7328_C000.

fm Page i Monday, October 16, 2006 11:03 AM

Biofluid
Mechanics
The Human Circulation
7328_C000.fm Page ii Monday, October 16, 2006 11:03 AM
7328_C000.fm Page iii Monday, October 16, 2006 11:03 AM

Biofluid
Mechanics
The Human Circulation

Krishan B. Chandran
Ajit P. Yoganathan
Stanley E. Rittgers

Boca Raton London New York

CRC is an imprint of the Taylor & Francis Group,

an informa business
7328_C000.fm Page iv Monday, October 16, 2006 11:03 AM

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2007 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper
10 9 8 7 6 5 4 3 2 1

International Standard Book Number-10: 0-8493-7328-X (Hardcover)

International Standard Book Number-13: 978-0-8493-7328-2 (Hardcover)

This book contains information obtained from authentic and highly regarded sources. Reprinted material
is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable
efforts have been made to publish reliable data and information, but the author and the publisher cannot
assume responsibility for the validity of all materials or for the consequences of their use.

No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photocopying, microfilming,
and recording, or in any information storage or retrieval system, without written permission from the
publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the
CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Chandran, K. B.
Biofluid mechanics : the human circulation / Krishnan B. Chandran, Ajit P.
Yoganathan, and Stanley E. Rittgers.
p. ; cm.
Includes bibliographical references and index.
ISBN-13: 978-0-8493-7328-2 (hardcover : alk. paper)
ISBN-10: 0-8493-7328-X (hardcover : alk. paper)
1. Blood flow. 2. Fluid mechanics. I. Yoganathan, A. P. (Ajit Prithiviraj), 1951- II.
Rittgers, Stanley E. III. Title.
[DNLM: 1. Blood Circulation--physiology. 2. Cardiovascular Physiology. 3. Heart
Valves--physiology. WG 103 C456b 2007]

QP105.C43 2007
612.1’181--dc22 2006031653

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the CRC Press Web site at

http://www.crcpress.com
7328_C000.fm Page v Monday, October 16, 2006 11:03 AM

Dedication

To my wife, Vanaja, and daughters, Aruna and Anjana, for their loving support.

KBC

To my wife, Tripti, and daughter, Anila, for their love, encouragement and support.

APY

To my wife, Eva, and sons, David and Andrew, who have taught me so much.

SER
7328_C000.fm Page vi Monday, October 16, 2006 11:03 AM
7328_C000.fm Page vii Monday, October 16, 2006 11:03 AM

Preface

The field of biomedical engineering has seen a rapid growth in the past
decade and numerous new undergraduate and graduate programs have
been established in many universities. With the increasing number of stu-
dents enrolled in these programs, there is also an increase in demand for
suitable text books. This work is an attempt to provide such a text book for
a course in the application of fluid mechanics to the study of the human
circulatory system. This book is intended as a first course on fluid mechanics
in the human circulation, which is suitable for senior undergraduate or for
first-year graduate students in biomedical engineering. The topics contained
in the various chapters were organized based on the experience gained by
the three authors in teaching courses on cardiovascular fluid mechanics in
their respective programs.
The book is organized into three parts. Part I consists of introductory
review material on fluid and solid mechanics and also a review of cardio-
vascular physiology pertinent to the topics covered in the subsequent chap-
ters. The first two chapters are an introduction to foundational material in
fluid and solid mechanics for those students not having had prior exposure
to these topics. In curricula where the students would have previously
enrolled in courses in fluid mechanics, mechanics of deformable bodies, and
in human physiology, this section may serve as a brief review.
Part II of the book discusses the fluid mechanics in the human circulation,
primarily applied to blood flow at the arterial level. The first chapter in this
section (Chapter 4) examines viscometry and the rheological behavior of
human blood. The solid mechanics of the arterial wall subject to transmural
pressure is also briefly discussed in this chapter, since the interaction
between blood and the arterial wall needs to be considered in unsteady
flow simulation. The application of steady flow models to derive some
useful diagnostic parameters, such as vascular resistance and Gorlin’s equa-
tions to describe time-averaged flow behavior past heart valves, is treated
next (Chapter 5). The third chapter in this section (Chapter 6) briefly
describes the Windkessel model for unsteady flow followed by a detailed
treatment of the Moens–Korteweg and Womersley models of pulsatile flow
in the human circulation. The relationship between flow-induced stresses
and the initiation and growth of atherosclerosis is also discussed with
qualitative treatment of flow in curved vessels, branches, and bifurcations,
as well as unsteady flow past stenoses and aneurysms. The final chapter in
this section (Chapter 7) examines the flow through native heart valves and
cardiac chambers.
7328_C000.fm Page viii Monday, October 16, 2006 11:03 AM

In Part III, vascular implants and measurements in the cardiovascular

system are discussed. The first chapter in this section (Chapter 8) examines
in detail the design and fluid mechanical evaluation of artificial heart
valves. The following chapter (Chapter 9) deals with the fluid mechanical
alterations related to vascular graft and stent implants. The final chapter
(Chapter 10) details measurement of blood flow, pressure, velocity, and
vascular impedance. The chapter concludes with discussions on sophisti-
cated fluid mechanical measurements employed in vitro and in vivo to
assess the complex time-dependent, three-dimensional flow in human
circulation.
Because this book is intended as a first course for exposing students to
fluid mechanics in the human circulation, only the heart and major arteries
are considered, while flow in other organ systems, such as the lungs, kidneys,
and the brain or flow in the microcirculation is not included.
The material contained in this book was developed over several years of
teaching these particular courses to undergraduate and graduate-level stu-
dents. The figures have been selectively included from publications of
numerous eminent researchers in the field and the captions include the
sources from which the figures were obtained (with permission). Citations
of references in the text, however, have not been included so as not to
distract the students from the subject matter. A reference list for each chapter
is provided at the end of the book. Students also are referred to leading
journals dealing with relevant subject matter (e.g., the Annals of Biomedical
Engineering; Journal of Biomechanical Engineering; Journal of Biomechanics;
Medical Engineering and Physics; Atherosclerosis; Circulation; Circulation Research;
Journal of Thoracic and Cardiovascular Surgery; and Annals of Thoracic Surgery)
as well as numerous other sources for current advances in the field.

Krishnan B. Chandran
Ajit P. Yoganathan
Stanley E. Rittgers
7328_C000.fm Page ix Monday, October 16, 2006 11:03 AM

Contents

Part I Fluid and Solid Mechanics

and Cardiovascular Physiology

Chapter 1 Fundamentals of Fluid Mechanics ..................................... 3

Chapter 2 Introduction to Solid Mechanics...................................... 47

Chapter 3 Cardiovascular Physiology................................................ 69

Part II Biomechanics of the Human Circulation

Chapter 4 Rheology of Blood and Blood Vessel Mechanics......... 115

Chapter 5 Static and Steady Flow Models ...................................... 167

Chapter 6 Unsteady Flow and Nonuniform

Geometric Models ............................................................ 191

Chapter 7 Native Heart Valves ......................................................... 257

Part III Cardiovascular Implants and Biomechanical

Measurements

Chapter 8 Prosthetic Heart Valve Fluid Dynamics ........................ 277

Chapter 9 Vascular Therapeutic Techniques................................... 315

Chapter 10 Fluid Dynamic Measurement Techiques ..................... 343

Bibliography ........................................................................................ 405

Index ..................................................................................................... 411

7328_C000.fm Page x Monday, October 16, 2006 11:03 AM
7328_S001.fm Page 1 Friday, July 28, 2006 4:35 PM

Part I

Fluid and Solid Mechanics

and Cardiovascular
Physiology
7328_S001.fm Page 2 Friday, July 28, 2006 4:35 PM
7328_C001.fm Page 3 Friday, October 6, 2006 3:48 PM

1
Fundamentals of Fluid Mechanics

CONTENTS
1.1 Introduction ....................................................................................................3
1.2 Intrinsic Fluid Properties..............................................................................4
1.2.1 Density ................................................................................................4
1.2.2 Viscosity ..............................................................................................4
1.2.3 Compressibility..................................................................................7
1.2.4 Surface Tension..................................................................................8
1.3 Hydrostatics....................................................................................................9
1.4 Macroscopic Balances of Mass and Momentum ....................................10
1.4.1 Conservation of Mass ..................................................................... 11
1.4.2 Conservation of Momentum .........................................................13
1.5 Microscopic Balance of Mass and Momentum.......................................15
1.5.1 Conservation of Mass .....................................................................15
1.5.2 Conservation of Momentum .........................................................18
1.5.3 Mathematical Solutions..................................................................22
1.6 The Bernoulli Equation...............................................................................26
1.7 Dimensional Analysis .................................................................................30
1.8 Fluid Mechanics in a Straight Tube..........................................................32
1.8.1 Flow Stability and Related Characteristics .................................33
1.8.1.1 Steady Laminar Flow .......................................................33
1.8.1.2 Turbulent Flow ..................................................................36
1.8.1.3 Flow Development ...........................................................38
1.8.1.4 Viscous and Turbulent Shear Stress...............................40
1.8.2 Effect of Flow Pulsatility................................................................41
1.9 Boundary Layer Separation .......................................................................44
1.10 Problems........................................................................................................45

1.1 Introduction
Before considering the mechanics of biological fluids in the circulation, it is
necessary to first consider some key definitions and specific properties. Once
established, these “pieces” will be used to construct important laws and

3
7328_C001.fm Page 4 Friday, October 6, 2006 3:48 PM

4 Biofluid Mechanics: The Human Circulation

principles that are the foundation of fluid mechanics. To begin, we will define
what we mean by a fluid. In general, a material can be characterized as a
fluid if it deforms continuously under the action of a shear stress produced
by a force, which acts parallel to the line of motion. In other words, a fluid
is a material that cannot resist the action of a shear stress. Conversely, a fluid
at rest cannot sustain a shearing stress. For our applications, we will treat a
fluid as a continuum (i.e., it is a homogeneous material) even though both
liquids and gases are made up of individual molecules. On a macroscopic
scale, however, fluid properties, such as density, viscosity, etc., are reasonably
considered to be continuous.

1.2 Intrinsic Fluid Properties

The intrinsic properties of a fluid are considered to be its density, viscosity,
compressibility, and surface tension. Following is a discussion and definition
of each of these properties.

1.2.1 Density
Density, commonly denoted by the symbol , is defined as the mass of a fluid
per unit volume and has units of mass/length3 [M/L3]. In the system meter/
kilogram/second (MKS) this would be represented by (kg/m3) or by (g/cm3)
in the CGS system (centimeter/gram/second) where 1 g/cm3 = 103 kg/m3.
Values of density for several common biofluids are:

ρwater = 999 kg/m3 at 15°C

ρair = 1.22 kg/m3 at standard atmospheric temperature and pressure
ρwhole blood = 1060 kg/m3 at 20°C (6% higher than water)

A related property of a fluid is its specific gravity, denoted as SG, which

is defined as its density divided by the density of water at 4°C (a reference
value that is quite repeatable). Thus, for whole blood at 20°C, SG = 1.06. The
specific weight of a fluid, denoted as γ (not to be confused with γ
, the symbol
for rate of shear), is the weight of a fluid per unit volume, or g. For blood
at 20°C, γ = 1.04 × 104 N/m3.

1.2.2 Viscosity
As we said earlier, a fluid is defined as a material that deforms under the
action of a shear force. The viscosity of a fluid (or its “stickiness”), denoted
7328_C001.fm Page 5 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 5

A
U Ft

h
y

FIGURE 1.1
Fluid subjected to simple shearing stress.

by the symbol, µ, is related to the rate of deformation that a fluid experiences

when a shear stress is applied to it. Just as with a solid, fluid shear stress is
defined as shear force per unit area applied tangentially to a surface and is
denoted by τ. To illustrate this, consider two parallel plates each of cross-
sectional area A [cm2] with fluid of viscosity µ between them, as shown in
Figure 1.1.
If a tangential force Ft is applied to the top plate as shown, it will result
in the plate moving with a velocity U [cm/s] relative to the lower plate.
The fluid adjacent to the top plate will move with the same velocity as that
of the plate, since the fluid is assumed to stick to the plate (known as the
“no-slip” condition). Similarly, the fluid adjacent to the bottom plate will
be at rest since it sticks to a stationary surface. Thus, a velocity gradient,
or change in velocity per unit change in height, is produced within the
fluid as shown. The shearing force Ft divided by the area A, over which it
acts, is defined as the shearing stress, τ, having the units [ML−1T−2]. The
velocity gradient, also referred to as the rate of shear, γ
, is the ratio U/h
where h is the distance between the two parallel plates. Thus, the rate of
shear has the dimension of sec−1. In general, the rate of shear is defined as
∂u/∂y where y is the distance perpendicular to the direction of shear as
shown in the figure. The viscous properties of all fluids are defined by the
relationship between the shear stress and the rate of shear over a range of
shear rates.
The relationship between viscous shear stress, τ, viscosity, µ, and shear
rate, ∂u/∂y, for flow in the x direction is given in Equation 1.1, with the
derivative taken in the y direction perpendicular to the direction of flow.

∂u
τ = −µ (1.1)
∂y

The coefficient in this relationship is known as the dynamic viscosity

[ML−1T−1] and is usually expressed as either Pa•sec (kg/m•s) in the meter/
kilogram/second [MKS] system or as Poise (gm/cm•s) in the centimeter/
gram/second [CGS] system. In many biological applications, it is conve-
nient to define a centiPoise (cP) where 1 cP = 0.01 Poise due to the relatively
7328_C001.fm Page 6 Friday, October 6, 2006 3:48 PM

6 Biofluid Mechanics: The Human Circulation

Casson’s Bingham plastic

fluid
Shear stress-τ
(dynes/cm2) Newtonian fluid

Power law
fluid
Yield
stress

⋅
Rate of shear-° (sec–1)

FIGURE 1.2
Shear stress vs. rate of shear plots for Newtonian and non-Newtonian fluids.

low values of this property. The constitutive relationship expressed in

Equation 1.1 can be plotted in a shear stress vs. rate of shear plot, as shown
in Figure 1.2.
A fluid in which the viscosity is constant is known as a Newtonian fluid
and the relationship between viscous shear stress and shear rate is rep-
resented in the figure by a straight line passing through the origin with
a slope equal to µ. In reality, many fluids do not follow this ideal linear
relationship. Those fluids in which the shear stress is not directly propor-
tional to the rate of shear are generally classified as non-Newtonian
fluids. In this case, the ratio of shear stress to the rate of shear at any
point of measurement is referred to as the apparent viscosity, µ app . The
apparent viscosity is not a constant, but depends on the rate of shear at
which it is measured. There are several classes of non-Newtonian fluids
whose constitutive relationships are shown in Figure 1.2. For example,
many fluids that exhibit a nonlinear relationship between shear stress
and rate of shear and pass through the origin are expressed by the
relationship

τ = K pl γ
n (1.2)

where n is less than unity. Such fluids are classified as power law fluids.
Another class of fluids is known as Bingham plastics because they will
initially resist deformation to an applied shear stress until the shear stress
exceeds a yield stress, τy . Beyond that point, there will be a linear relationship
between shear stress and rate of shear. The constitutive relationship for a
7328_C001.fm Page 7 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 7

Bingham plastic is given by

τ = τy + µbγ
(1.3)

where τy is the yield stress and µb is the plastic viscosity. Fluids that exhibit
a yield stress and also a nonlinear relationship between shear stress and rate
of shear may be classified as Casson fluids. The specific empirical relation-
ship for such fluids, which deviate from the ideal Bingham plastic behavior,
is known as Casson’s equation or

τ = τ y + k c γ
(1.4)

As pointed out earlier, it is important in many biomedical applications to

know the rheological characteristics of blood. In order to understand the
relationship between the shear stress and the rate of shear for blood, exper-
imental measurements are necessary. From these experiments, it has been
determined that blood behaves as a Newtonian fluid only in regions of
relatively high shear rate (>100 sec–1). Thus, for flow in large arteries where
the shear rate is well above 100 sec–1, a value of 3.5 cP is often used as an
estimate for the (assumed) constant viscosity of blood. In the microcircula-
tion (i.e., small arteries and capillaries) and in veins where the shear rate is
very low, blood must be treated as a non-Newtonian fluid (see Chapter 4,
section 4.3: Viscous Behavior of Blood).
The viscosity of a fluid is also strongly dependent upon its temperature.
Generally, the viscosity of liquids decreases with increasing temperature,
while the viscosity of gases increases with increasing temperature.

1.2.3 Compressibility
The compressibility of a fluid is quantified by the pressure change required
to produce a certain increment in either the fluid’s volume or density. This
property, known as the Bulk Modulus (k), is defined as

∆p ∆p
k= d∨ = dρ (1.5)
∨ ρ

Thus, for an incompressible fluid, k = ∞. For water, k = 2.15 × 109 N/m2,

indicating that it is practically incompressible. Since the transmission of
sound through a fluid such as air or water is simply the movement of
pressure waves through the fluid, the speed of sound, c, depends on the
fluid’s bulk modulus and density as

c = k/ρ (1.6)
7328_C001.fm Page 8 Friday, October 6, 2006 3:48 PM

8 Biofluid Mechanics: The Human Circulation

πR2∆p

2πRσ

FIGURE 1.3
Pressure-surface, tension force balance for a hemispherical drop.

Therefore, for water at 15°C,

c = (2.15 × 109/999)0.5 = 1470 m/s

The speed of sound in biological tissues and blood is used in various

ultrasound modalities to obtain anatomic images and also blood flow veloc-
ities (see Chapter 10, section 10.6: Ultrasound Doppler Velocimetry).

1.2.4 Surface Tension

The surface layer between two different liquids behaves like a stretched
membrane due to intermolecular forces. An expression can be found for the
surface tension in a fluid, σ, by considering a hemispherical fluid drop of
radius, R, as shown below (Figure 1.3).
Assuming static equilibrium between the pressure and surface forces and
applying a force balance in the vertical direction,

2 πRσ = ∆pπR 2

and, hence,

R∆p
σ= (1.7)
2

The effect of surface tension is particularly important in the pulmonary

airways as it is what maintains the openings in the alveoli (the smallest
regions of the lung where gas exchange occurs) and, thus, enables us to
breath sufficient air in and out.
7328_C001.fm Page 9 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 9

1.3 Hydrostatics
While most fluids in the body exist in a state of continuous motion, there
are important effects on the fluid, which are due to static forces. A fluid at
rest in a gravitational field, for example, is in hydrostatic equilibrium as
shown in Figure 1.4. Under these conditions, the weight of the fluid is exactly
offset by the net pressure force supporting the fluid. Here, the pressure at
the base of the fluid element is p while that at a distance dz above the base
is p plus the gradient of pressure in the z direction, dp/dz, times the incre-
mental elevation, dz, where p is the gauge pressure referenced to the atmo-
spheric pressure, pa (= 1.01 × 105 N/m2). Thus, if we sum the pressure and
gravitational forces acting on the element in the vertical direction, we get

 dp 
pdA −  p + dz dA − ρgdAdz = 0 (1.8)
 dz 

which requires that the pressure gradient be:

dp
= −ρg (1.9)
dz

where g is the gravitational acceleration. The negative sign for the pressure
gradient indicates that p decreases as z increases. Integrating the above

dp
( p+ dz ) dA
dz
z
dA

rg dA dz
dz

pdA

FIGURE 1.4
Hydrostatic equilibrium of a fluid element.
7328_C001.fm Page 10 Friday, October 6, 2006 3:48 PM

10 Biofluid Mechanics: The Human Circulation

equation between the base and the top of the element, z1 and z2, respectively,
with corresponding pressures p1 and p2 , yields

∫ ∫
p2 z2
dp = − ρgdz (1.10)
p1 z1

or

p2 − p1 = −ρg( z2 − z1 ) (1.11)

Thus,

∆p = ρgh (1.12)

for any element of height h. An equivalent expression would be ∆p = γ h.

Example 1.1:
For the case of a column of mercury where ρHg = 1.35 × 104 kg/m3, the
pressure increase over a height of 1 mm would be:

∆p = 1.35 × 104 (kg/m3) × 9.81 (m/s2) × 10−3 (m)

= 133 N/m2
= 133 Pa

1.4 Macroscopic Balances of Mass and Momentum

The ultimate goal of fluid mechanics is to identify relationships between
variables so that we can determine the value of one or more of these variables
in terms of given conditions. In order to do this, we will begin with basic
balances of properties or “Conservation Laws” that, in turn, involve multiple
variables of interest. Initially, we will do so on a Macroscopic basis where
we consider the dynamics associated with a relatively large volume of fluid.
Later, in section 1.5, Microscopic Balances of Mass and Momentum, we will
reevaluate these laws relative to an infinitesimal or Microscopic volume. By
doing so, we will be able to determine relatively simple, gross parameters
(i.e., flow rate, reaction forces, etc.) with the Macroscopic approach, but more
complex, detailed parameters (i.e., local velocities, pressures, etc.) with the
Microscopic approach. Classically, the three properties of a fluid that are
considered in this analysis are Mass, Momentum, and Energy. We will
establish balances for each based upon the Reynolds Transport Theorem.
7328_C001.fm Page 11 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 11

dA2
V2n

dA1
dV
A2

V1n

A1

FIGURE 1.5
Mass flux balance for a stream tube.

This theorem relates the time rate of change of each of these properties in a
System relative to corresponding changes that occur within and across a
Control Volume and is given by

dBsys ∂
dt
=
∂t ∫ bρd ∨ + ∫ bρVdA
CV CS
(1.13)

where B denotes the extensive (i.e., absolute) property, b denotes the amount
of that property per unit mass (i.e., b = B/m), CV denotes the Control Volume
and CS denotes the Control Surfaces of that volume.

1.4.1 Conservation of Mass

The requirement (or “law”) of conservation of mass applies to all materials
and is the logical starting point for an introduction to fluid motion. To begin,
we consider a conceptual 3-D space or Control Volume that is enclosed by
a surface across which fluid can move (Figure 1.5). This control volume is
constant in size and location over time and does not necessarily coincide with
any physical boundaries; that is, it is purely a tool for the analysis of physical
systems. If we apply Equation 1.13 for the property Mass, then the time rate
of change of Bsys would be zero since mass of a system is constant by definition.
Thus, the law of Conservation of Mass (also known as the Continuity Equa-
tion) states that any change in mass within this control volume must be equal
to the mass of fluid that enters the volume (Mass In) minus the mass that
exits (Mass Out). Therefore, for a given period of time, ∆t ,

(Rate of Mass In) – (Rate of Mass Out)

= Rate of Change of Mass within the Tube
7328_C001.fm Page 12 Friday, October 6, 2006 3:48 PM

12 Biofluid Mechanics: The Human Circulation

The rate of mass carried across a surface is equal to the density times the
flow rate, ρQ, where Q is the volume flow rate, or ρVn A, and Vn is the com-
ponent of velocity normal to the cross section. Furthermore, the rate of
change of mass within the tube is given by

∂
∂t ∫ ρd ∨
CV

Thus, for the case shown in Figure 1.5,

Rate of Mass In = ∫ ρV dA
A1
1n

and

Rate of Mass Out = ∫ ρV

A2
2n dA

where V1n and V2 n are the velocities normal to the differential areas at cross
sections 1 and 2, respectively.
Inserting these into the law of Conservation of Mass, gives us

∂
∫
− ρV2 ndA +
A2
∫ ρV dA = ∂t ∫ ρd ∨
A1
1n
∨
(1.14)

For liquids in general and blood in particular, a very good assumption is

that the fluid is incompressible; that is, its density, ρ , is constant. This,
together with the fixed size of the Control Volume, causes the time rate of
change of mass within the control volume to be zero. Furthermore, if the
flow is steady and there are only two surfaces across which fluid flows
(Surfaces 1 and 2), then the mean velocities (V) and cross-sectional areas
( A) can be related as shown in Equation 1.15)

A1V1 = A2V2 = Q (constant) (1.15)

If the velocity varies over the cross section as a function of radius, r, then
the mean velocities must first be obtained by

∫ V (r)dA
1
V1 = 1
A1
A1
7328_C001.fm Page 13 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 13

and

∫ V (r)dA
1
V2 = 2
A2
A2

1.4.2 Conservation of Momentum

The principle of Conservation of Momentum was initially formulated from
Newton’s
 Second Law of Motion, which states that “the sum of the forces

( ΣF) acting on an object is equal to its mass (m) times its acceleration ( a ) ” or

 
∑ F = ma (1.16)

 
Rewriting a as dV
dt
and bringing m inside the differential (since it is con-
stant for a System), results in

 
  dV  d(mV )
∑ F = m  = dt
 dt 
(1.17)

where the derivative is now the time rate of change of momentum.

Therefore, an alternative way of stating Newton’s Second Law of Motion
is that for a system

Sum of the External Forces acting on the System

= Time Rate of Change of Linear Momentum of the System

 
∑F ∫ Vρ d ∨
D
sys = (1.18)
Dt
sys

When the System and the Control Volume are coincident at an instant of
time, the forces acting on the System and the forces acting on the Control
Volume are identical or

 
∑F = ∑F sys CV (1.19)

Considering the change in linear momentum for such a System and

coincident Control Volume (Equation 1.18 R.H.S.), the Reynolds Transport
Theorem allows us to write
7328_C001.fm Page 14 Friday, October 6, 2006 3:48 PM

14 Biofluid Mechanics: The Human Circulation

Time rate of change of linear momentum in a System

= Time rate of change of linear momentum in Control Volume
+ Net rate of change of linear momentum through Control Volume Surfaces
or

 ∂    
∫ Vρ d ∨ = ∂t ∫ Vρ d ∨ + ∫ VρV ⋅ n dA
D
(1.20)
Dt
sys CV CS

Therefore, for a Control Volume that is fixed (i.e., with respect to an inertial
reference) and nondeforming

 ∂    
∑F CV =
∂t ∫ Vρ d ∨ + ∫ VρV ⋅ n dA
CV CS
(1.21)

The above equation is called the Linear Momentum Equation because we

only consider motion acting in an axial direction. The forces in this equation,
acting on the Control Volume, are both body forces and surface forces and
can be expressed as

  
∑F CV = ∫ ρ g d ∨ + ∫ t dA
CV CS
(1.22a)


where g is the body force per unit mass acting on the Control Volume con-
tents and t is the stress vector acting on the Control Volume surfaces.
When viscous effects are important, the surface area integral of the stress
vector is nonzero and must be determined empirically (i.e., from experi-
mental data) and given as friction factors or drag coefficients (i.e., constants
relating drag forces to other variables). However, if viscous effects are
negligible, then the stress vector is given by


t = − np
ˆ (1.22b)

where n̂ is the outward directed unit vector normal to the surface and p is
the pressure.
The previous two balances of mass and momentum are called macroscopic
or integral balances because they consider the Control Volume as a large,
discrete space and are written in terms of bulk-flow variables. In order to
derive more general forms of these equations, which provide spatial detail
throughout the flow field, we need to take a microscopic or differential
approach. Such an approach leads us to what are commonly referred to in
fluid mechanics literature as the Continuity and the Navier–Stokes
Equations, for Conservation of Mass and Momentum, respectively.
7328_C001.fm Page 15 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 15

1.5 Microscopic Balance of Mass and Momentum

1.5.1 Conservation of Mass
We begin by considering an infinitesimal control volume, CV, of dimen-
sion ∆x∆y∆z (Figure 1.6a).
A fluid flowing through the control volume will have a velocity denoted
by the vector


V = uiˆ + vjˆ + wkˆ (1.23)

where each of these variables may be a function of time.

The conservation of mass principle states that for a system,

The net rate of mass flux across the control

surface + The rate of change of mass inside control volume = zero

Treating each of the above terms individually, we get:

1. The net rate of mass flux across the control surfaces (Figure 1.6b)
in the x direction:

(ρu|x −ρu|x+∆x )∆y∆z

y
y
ρv|y+∆y
Control
v volume ρw|z+∆z
dy
u
w O
ρu|x ρu|x+∆x
dz
dx ρw|z
x x
ρv|y

z z
(a) (b)

FIGURE 1.6
(A) Differential control volume in rectangular coordinates. (B) Mass flux across surfaces of
control volume.
7328_C001.fm Page 16 Friday, October 6, 2006 3:48 PM

16 Biofluid Mechanics: The Human Circulation

y direction:

(ρv|y −ρv|y+∆y )∆x∆z

z direction:

(ρw|z −ρw|z+∆z )∆x∆y

2. The rate of change of mass inside the control volume:

∂
(ρ∆x∆y∆z)
∂t

Combining terms and rearranging yields

(ρu|x+ ∆x −ρu|x )∆y∆z + (ρv|y+ ∆y −ρv|y )∆x∆z + (ρw|z+ ∆z −ρw|z )∆x∆y

∂ (1.24)
=− (ρ∆x∆y∆z)
∂t

Furthermore, each of the mass flux terms can be equivalently written as

the gradient of mass flux in that coordinate direction times the distance
moved or

∂(ρu)
(ρu|x −ρu|x+∆x ) = ⋅ ∆x
∂x

etc.
Since the volume within the control element is time invariant, we can divide
each term by ∆x∆y∆z . Then, in the limit as ∆x∆y∆z approaches zero, we
obtain

∂ ∂ ∂ ∂ρ
(ρu) + (ρv) + (ρw) + =0 (1.25)
∂x ∂y ∂z ∂t

which is equivalent to

 ∂ρ
∇ ⋅ ρV + =0 (1.26)
∂t

The above equation is known as the Continuity Equation where the “del”
vector operator, ∇, is defined as

∂  ∂  ∂
∇=i +j +k (1.27)
∂x ∂y ∂z
7328_C001.fm Page 17 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 17

A fluid such as blood is incompressible and, thus, ρ = constant, so that the

Continuity Equation becomes


ρ∇ ⋅ V = 0 (1.28)

or


∇ ⋅V = 0 (1.29)

Note that this equation is valid for both steady and unsteady (including
pulsatile, three dimensional) flows of an incompressible fluid.
If we return to Equation 1.24 and differentiate each numerator, we obtain

∂ρ ∂ρ ∂ρ ∂ρ  ∂u ∂v ∂w  Dρ  ∂u ∂v ∂w 
+ u + v + w + ρ + +  = + ρ + + =0
∂t ∂x ∂y ∂z  ∂x ∂y ∂z  Dt  ∂x ∂y ∂z 
(1.30)

or

Dρ 
+ ρ∇ ⋅ V = 0 (1.31)
Dt

where

D ∂ ∂ ∂ ∂
= +u +v +w
Dt ∂t ∂x ∂y ∂z

is called substantive derivative.

The substantive derivative represents the time derivative of a scalar or
vector quantity, which follows the motion of the fluid. For example, for a
scalar quantity such as temperature, T, the substantive derivative would be
expressed as

DT ∂T ∂T ∂T ∂T
= +u +v +w (1.32)
Dt ∂t ∂x ∂y ∂z

∂T
where ∂t
is the local time rate of change of temperature and the terms

∂T ∂T ∂T
u +v +w
∂x ∂y ∂z
7328_C001.fm Page 18 Friday, October 6, 2006 3:48 PM

18 Biofluid Mechanics: The Human Circulation

represent the rate of change of temperature due to fluid motion (also known
as Convection).
Finally, for many fluid dynamic situations, such as the consideration of
liquids or gases at low speeds, it is quite common to assume an incompress-
ible fluid where ρ = Constant. In this case, Equation 1.31 can be further
reduced to


∇ ⋅V = 0

1.5.2 Conservation of Momentum

In deriving the differential form of this law, we once again consider the
dynamics associated with a fluid control volume ( ∆∨ = ∆x∆y∆z) shown in
Figure 1.6b. We now apply Newton’s Second Law of Motion, as written in
Equation. 1.16, in terms of the time rate of change of momentum:

Sum of the external forces acting on the Control Volume =

Net rate of efflux of linear momentum across the
Control Volume + Time rate of change of
linear momentum within the Control Volume

In general, linear momentum per unit volume of fluid can be expressed

as ρV so that by multiplying this term by the rate of volume change, we can
obtain the time rate of change of linear momentum. To determine the flux
of a property across the Control Volume
 surface, the appropriate expression
for the rate of volume change is (V ⋅ nˆ )dA, where n̂ is the outward directed
normal to a particular surface. For change of a property within the Control
Volume, the rate of volume change is simply given by ddt∨ . As with the deri-
vation of the Continuity Equation, the Control Volume considered is constant
and we can express the above equality as

    
∑F (
ρV V ⋅ n dA ) ∂ ρVd ∨
lim
∆ ∨→0 ∆x ∆y ∆z
= lim
∆ ∨→0 ∫∫ ∆x∆y∆z
+ lim
∆ ∨→0 ∂t ∫∫∫ ∆x∆y∆z
(1.33)

if we take the limit of ∆∨ = ∆x∆y∆z as it approaches zero.

Each of these terms can be evaluated separately as follows:

1. Sum of the external forces


lim
∑F 
= dF
∆ ∨→0 ∆x∆y∆z
7328_C001.fm Page 19 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 19

2. Net rate of momentum efflux across the control volume

  
ρV (V ⋅ n)dA ∂  ∂  ∂ 
lim
∆∨→0 ∫∫ ∆x∆y∆z
= (ρVu) + (ρVv) + (ρVw)
∂x ∂y ∂z
  
 ∂ ∂ ∂   ∂V ∂V ∂V 
= V  (ρu) + (ρv) + (ρw)  + ρ  u +v +w 
 ∂x ∂y ∂z   ∂x ∂z ∂z 

At this point, we can also use the Continuity Equation to substitute

terms

∂ ∂ ∂ ∂ρ
(ρu) + (ρv) + (ρw) = − (1.34)
∂x ∂y ∂z ∂t

and reduce this limit to

     
ρV (V ⋅ n)dA  ∂ρ  ∂V ∂V ∂V 
lim
∆∨→0 ∫∫ ∆x∆y∆z
= −V
∂t
+ ρ u
 ∂x
+v
∂y
+w 
∂z 

3. Time rate of change of momentum within the control volume

 
Vρd ∨ ∂  ∂V  ∂ρ
lim
∆∨→0 ∫∫∫ = (ρV ) = ρ
∆x∆y∆z ∂t ∂t
+V
∂t

Substituting for the limits and combining terms gives

   
  ∂V ∂V ∂V  dV
dF = ρ  u +v +w +ρ (1.35)
 ∂x ∂y ∂z  dt

We have now expressed the change of momentum in terms of its compo-

nent velocities. Let’s now look at the external forces in more detail. They
consist of the sum of the body forces, FB , and the surface forces, F S. The body
forces are typically due to the presence of gravitational, electromagnetic, and
electrostatic fields. If the only body force is gravity, then

FBx = ρgx ∆x∆y∆z (1.36)

The surface forces acting on the control volume are those due to the normal,
σ, and the shear, τ, stresses. These stresses can be assumed to vary contin-
uously from their nominal value at the center of the Control Volume in each
7328_C001.fm Page 20 Friday, October 6, 2006 3:48 PM

20 Biofluid Mechanics: The Human Circulation

y ∂␶yx dy
␶yx +
∂y 2

␶zx − ∂␶zx dz
∂z 2
σxx _ ∂σxx dx
∂x 2
σxx + ∂σxx dx
∂x 2
∂␶yx dy
␶yx _
∂y 2
␶zx + ∂␶zx dz
∂z 2
x

FIGURE 1.7
Normal and shear stresses along the x coordinate in the Control Volume.

of the coordinate directions. Figure 1.7 depicts the normal and shear stresses
acting on the Control Volume in the x direction alone. Similar figures can be
constructed for the normal and shear stresses acting in the y and z directions.
Thus, the net surface force acting in the x direction is given by

 ∂σ   ∂τ yx   ∂τ 
FSx =  xx  ∆x∆y∆z +   ∆x∆y∆z +  zx  ∆x∆y∆z (1.37)
 ∂x   ∂y   ∂z 

The total force acting on the control volume in the x direction then becomes

Fx = FBx + FSx (1.38)

which, in the limit, can be expressed as

∂σ xx ∂τ yx ∂τ zx
dFx = ρgx + + + (1.39a)
∂x ∂y ∂z

Similarly, the differential force components in the y and z directions are:

∂σ yy ∂τ xy ∂τ zy
dFy = ρg y + + + (1.39b)
∂y ∂x ∂z

∂σ zz ∂τ xz ∂τ yz
dFz = ρgz + + + (1.39c)
∂z ∂x ∂y
7328_C001.fm Page 21 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 21

Substituting the results of the above expressions 1.39 a–c back into the
expression for Newton’s Second Law (Equation 1.17) yields

∂σ xx ∂τ yx ∂τ zx  ∂u ∂u ∂u ∂u 
ρgx + + + = ρ + u + v + w  (1.40a)
∂x ∂y ∂z  ∂t ∂x ∂y ∂z 

∂τ xy ∂σ yy ∂τ zy  ∂v ∂v ∂v ∂v 
ρg y + + + = ρ + u + v + w  (1.40b)
∂x ∂y ∂z  ∂t ∂x ∂y ∂z 

∂τ xz ∂τ yz ∂σ zz  ∂w ∂w ∂w ∂w 
ρgz + + + = ρ +u +v +w  (1.40c)
∂x ∂y ∂z  ∂t ∂x ∂y ∂z 

In this form, we can see that the right-hand side of the above equa-
tions actually represents density (mass/volume) × acceleration, or force/
volume, where the acceleration terms can be separated into local accel-
eration (∂u/∂t, etc.) and convective acceleration (u ∂u/∂x, etc.) compo-
nents. The total acceleration can be expressed in terms of the substantive
derivative as

Du ∂u ∂u ∂u ∂u
= +u +v +w (1.41)
Dt ∂t ∂x ∂y ∂z

Equation 1.40a to Equation 1.40c represent the complete form of the differ-
ential Conservation of Momentum balances. These equations cannot be
solved, however, because there are more unknowns (i.e., dependent variables)
than equations. Thus, it is necessary to derive additional information in order
to provide those equations. In practice, we will consider the special, although
not uncommon, case of incompressible Newtonian fluids. Here, the normal
and shear stresses can be expressed as

∂u  ∂u ∂v 
σ xx = − p + 2µ τ xy = τ yx = µ  +  (1.42a)
∂x  ∂y ∂x 

∂v  ∂w ∂v 
σ yy = − p + 2µ τ yz = τ zy = µ  + (1.42b)
∂y  ∂y ∂z 

∂w  ∂u ∂w 
σ zz = − p + 2µ τ zx = τ xz = µ  + (1.42c)
∂z  ∂z ∂x 
7328_C001.fm Page 22 Friday, October 6, 2006 3:48 PM

22 Biofluid Mechanics: The Human Circulation

By substituting these relationships into Equation 1.40a to Equation1.40c,

we obtain the Equations of Motion in scalar form along the three coordinate
axes as

∂p  ∂2 u ∂2 u ∂2 u   ∂u ∂u ∂u ∂u 
ρgx − + µ  2 + 2 + 2  = ρ + u + v + w  (1.43a)
∂x  ∂x ∂y ∂z   ∂t ∂x ∂y ∂z 

∂p  ∂2 v ∂2 v ∂2 v   ∂v ∂v ∂v ∂v 
ρg y − + µ  2 + 2 + 2  = ρ + u + v + w  (1.43b)
∂y  ∂x ∂y ∂z   ∂t ∂x ∂y ∂z 

∂p  ∂2 w ∂2 w ∂2 w   ∂w ∂w ∂w ∂w 
ρgz − + µ  2 + 2 + 2  = ρ +u +v + w  (1.43c)
∂z  ∂x ∂y ∂z   ∂t ∂x ∂y ∂z 

The equivalent vector form of these equations is:


  DV
ρ g − ∇p + µ∇ V = ρ
2 (1.44)
dt

which can be alternatively written as


∂V   1  
+ (V ⋅∇)V = − ∇p + g + ν(∇ 2V ) (1.45)
∂t ρ

by expanding the material derivative for acceleration and dividing by ρ.

The above equations (in either scalar or vector form) are commonly called
the Navier–Stokes equations for Newtonian incompressible fluids in honor
of the mathematicians who originally derived these relationships.The Con-
tinuity and Navier–Stokes equations can also be derived in other coordinate
systems and, thus, are given in Cartesian, cylindrical, and spherical coordi-
nates in Table 1.1.

1.5.3 Mathematical Solutions

As mentioned earlier, in order to solve a set of equations, we must have at
least as many constraints as we have dependent variables. Examination of
Equation 1.43a to Equation 1.43c shows that there are four dependent
variables — pressure (p) and the three velocity components (u, v, and w) —
defined in terms of four independent variables — time (t) and the three
position coordinates (x, y, and z) — but only three equations. (Note: Similar
results would be found with the cylindrical and spherical coordinate systems
as well.) However, by including the Continuity equation, we obtain a fourth
constraint, which will allow us to uniquely define each dependent variable.
Mathematically, Equation 1.31 (Continuity) and Equation 1.44 (Navier–Stokes)
7328_C001.fm Page 23 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 23

TABLE 1.1
Continuity and Momentum Equations in Cartesian, Cylindrical (Polar),
and Spherical Coordinate Systems
(a) Cartesian Coordinate System:

∂u ∂v ∂w
+ + =0
∂x ∂y ∂z

 ∂u ∂u ∂u ∂u  ∂p  ∂2 u ∂2 u ∂2 u 
ρ  + u + v + w  = ρBx − +µ 2 + 2 + 2 
 ∂t ∂x ∂y ∂z  ∂x  ∂x ∂y ∂z 

 ∂v ∂v ∂v ∂v  ∂p  ∂2 v ∂2 v ∂2 v 
ρ  + u + v + w  = ρBy − +µ 2 + 2 + 2 
 ∂t ∂x ∂y ∂z  ∂y  ∂x ∂y ∂z 

and

 ∂w ∂w ∂w ∂w  ∂p  ∂2 w ∂2 w ∂2 w 
ρ +u +v +w  = ρBz − +µ 2 + 2 + 2 
 ∂t ∂x ∂y ∂z  ∂z  ∂x ∂y ∂z 

(b) Cylindrical Coordinate System:

 1 ∂ ( rVr ) 1 ∂V ∂V
θ
∇ ⋅V = + + z =0
r ∂r r ∂θ ∂z

 ∂V ∂V V ∂Vr Vθ2 ∂V 
ρ  r + Vr r + θ − + Vz r 
 ∂t ∂r r ∂θ r ∂z 

∂p  ∂2V 1 ∂Vr Vr 1 ∂2Vr 2 ∂Vθ ∂2Vr 

= Fr − + µ  2r + − 2 + 2 − 2 + 2 
∂r  ∂r r ∂r r r ∂θ2 r ∂θ ∂z 

 ∂V ∂V V ∂V V V ∂V 
ρ  θ + Vr θ + θ θ + r θ + Vz θ 
 ∂t ∂r r ∂θ r ∂z 

1 ∂p  ∂2V 1 ∂Vθ Vθ 1 ∂2Vθ 2 ∂Vr ∂2Vθ 

= Fθ − + µ  2θ + − 2 + 2 + 2 +
r ∂θ  ∂r r ∂r r r ∂θ2 r ∂θ ∂z 2 

and

 ∂V ∂V V ∂Vz ∂V 
ρ  z + Vr z + θ + Vz z 
 ∂t ∂r r ∂θ ∂z 

∂p  ∂2V 1 ∂Vz 1 ∂2Vz ∂2Vz 

= Fz − + µ  2z + + + 2 
∂z  ∂r r ∂r r 2 ∂θ2 ∂z 

(continued)
7328_C001.fm Page 24 Friday, October 6, 2006 3:48 PM

24 Biofluid Mechanics: The Human Circulation

TABLE 1.1 (CONTINUED)

Continuity and Momentum Equations in Cartesian, Cylindrical (Polar),
and Spherical Coordinate Systems

(c) Spherical Coordinate System:

∂Vr 2Vr 1 ∂Vθ Vθ cot θ 1 ∂Vφ

+ + + + =0
∂r r r ∂θ r r sin θ ∂φ

 ∂V ∂V V ∂Vr Vφ ∂Vr Vθ2 Vφ2 

ρ  r + Vr r + θ + − −
 ∂t ∂r r ∂θ r sin θ ∂φ r r 

 ∂2Vr 2 ∂Vr 2Vr 1 ∂2Vr cot θ ∂Vr 1 ∂2Vr 2 ∂Vθ 

 ∂r 2 + r ∂r − r 2 + r 2 ∂θ2 + r 2 ∂θ + r 2 sin 2 θ ∂φ 2 − r 2 ∂θ 
∂p  
= Fr − + µ 
∂r  2Vθ cot θ 2 ∂Vφ 
 − r 2
− 2
r sin θ ∂φ 

 ∂V ∂V V V V ∂V Vφ ∂Vθ Vφ2 cot θ 

ρ  θ + Vr θ + r θ + θ θ + − 
 ∂t ∂r r r ∂θ r sin θ ∂φ r 

 ∂2Vθ 2 ∂Vθ Vθ 1 ∂2Vθ cot θ ∂Vθ 

 ∂r 2 + r ∂r − r 2 sin 2 θ + r 2 ∂θ2 + r 2 ∂θ 
1 ∂p
= Fθ − + µ  

r ∂θ 1 ∂ 2V 2 ∂V 2 cot θ ∂V
+ θ φ 
 r 2 sin 2 θ ∂φ 2 + r 2 ∂θ − r 2 sin θ ∂φ
r


and

 ∂Vφ ∂Vφ Vr Vφ Vθ ∂Vφ VθVφ cot θ Vφ ∂Vφ 

ρ + Vr + + + +
 ∂t ∂r r r ∂θ r r sin θ ∂φ 

 ∂2Vφ 2 ∂Vφ Vφ 1 ∂2Vφ cot θ ∂Vφ 

 ∂r 2 + r ∂r − r 2 sin 2 θ + r 2 ∂θ2 + r 2 ∂θ 
1 ∂p  
= Fφ − + µ 
r sin φ ∂φ ∂2Vφ ∂Vr 2 cot θ ∂Vθ
+ 1 2 
 r 2 siin 2 θ ∂φ 2 + r 2 sin θ ∂φ + r 2 sin θ ∂φ 

are first- and second-order partial differential equations, respectively. Fur-

thermore, Equation 1.43a to Equation 1.43c are nonlinear because of the
presence of product terms such as u·∂u/∂x, v·∂u/∂y, etc. Unfortunately, no
exact analytical solution has been defined for equations of this form. There-
fore, in practice, two approaches have been taken. One is to first simplify
the equations until they have a mathematical form for which there is a
solution. For example, we could assume steady, two-dimensional flow (say,
in the x-y plane) along one axis (say, the x axis), which would eliminate all
terms involving ∂/∂t and two of the velocity components. If the flow were
in the x direction only, then Equation 1.31 would reduce to

∂u
=0
∂x
7328_C001.fm Page 25 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 25

or

u = f ( y) + C
where C is the integration constant, and Equation 1.44 would reduce to

∂p  ∂2 u 
= µ 2 
∂x  ∂y 
Here, it is now possible to solve for u( y) as an explicit function of p.
The other approach taken is to solve these equations numerically. This
approach is more complex, but provides the ability to solve problems with-
out making unrealistic simplifying assumptions. The basic technique is to
first sub-divide the flow into many small regions, or elements, over which
the governing equations are applied. Rather than use the differential form
of the equations, however, they are rewritten in algebraic form in terms of
changes that occur in variables due to incremental changes in position and
time. Solutions are then obtained locally at specific locations, or nodes, on
the finite elements across a mesh of elements (Figure 1.8). This set of solutions

1 3

FIGURE 1.8
CFD Mesh for a reconstructed human coronary artery with a stenosis segment. It can be
observed that a finer mesh is employed in the area of stenosis in order to obtain more accurate
results in this region of interest.
7328_C001.fm Page 26 Friday, October 6, 2006 3:48 PM

26 Biofluid Mechanics: The Human Circulation

is then updated at subsequent time intervals over the entire mesh until some
acceptable level of accuracy, or tolerance, is achieved based upon conver-
gence between successive values of certain output variables. Obviously, this
can be a very detailed and time-consuming process depending upon the
complexity of the geometry being analyzed and the initial and boundary
conditions imposed. Furthermore, additional features are sometimes
included in the equations to allow for simulation of non-Newtonian fluids
and turbulent flow conditions, for example. While it is always important to
validate such results, these computational fluid dynamic (CFD) software
programs are increasingly being used to solve challenging biomedical flow
problems unapproachable by any other means.

1.6 The Bernoulli Equation

One particularly useful tool known as the Bernoulli equation can be
obtained by simplification of the Navier–Stokes equations (Equation 1.43a
to Equation 1.43c). Specifically, flow is assumed to be inviscid (i.e., the viscous
forces are negligible), which then allows the Navier–Stokes equations to be
simplified to what is known as Euler’s Equation of Motion (Equation 1.46).


∂V   1 
+ (V ⋅∇)V = − ∇p + g (1.46)
∂t ρ

As can be seen, the shear stress terms have been set equal to zero since
there is no friction in the system. To further simplify the equation of motion,
we will focus on flow along a streamline (Figure 1.9). A streamline is an

ds
∂p dz ) dA
( p + ∂z

V
θ dz

p dA
dA

pg dA ds

FIGURE 1.9
Force balance for a fluid particle along a stream line.
7328_C001.fm Page 27 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 27

imaginary line in the flow field drawn so that it is always tangential to the
velocity vectors. The sum of the forces acting on a fluid particle along the
streamline is given by

 ∂p 
∑ F = pdA −  p + ∂s ds dA − ρg(sin θ)dAds
S (1.47)

where

dz
sin θ =
ds

Substituting this and simplifying Equation 1.47 results in

 ∂p 
∑ F =  − ∂s ds − ρgdz dA
s (1.48)

Furthermore, fluid acceleration along the streamline curve consists of

temporal and spatial components given by

  
dV ∂V  ∂V
= +V (1.49)
dt ∂t ∂s
tan g

Taking the mass of the particle as ρdAdS and inputting Equation 1.48 and
Equation 1.49 into Newton’s Second Law of Motion gives us

 
 ∂V  ∂V   ∂p 
ρdAdS  +V  =  − ∂s ds − ρgdz dA (1.50)
 ∂t ∂s   

  
 2
Since an equivalent mathematical expression for the term V ∂∂Vs is ∂∂s  V2  , we
will now substitute this into Equation 1.49 and integrate between any two
arbitrary points 1 and 2 to obtain


(V 2 − V12 ) + (p − p ) + ρg(z − z ) = 0
2
∂V
ρ ∫
1
∂t
ds + ρ 2
2 2 1 2 1 (1.51)

Equation 1.51 is known as the unsteady form of the Bernoulli equation

where p1, V1, and z1 are the pressure, velocity, and height at the upstream
location, and p2, V2 , and z2 are the pressure, velocity, and height at the down-
stream location. The integral term accounts for the flow acceleration or
7328_C001.fm Page 28 Friday, October 6, 2006 3:48 PM

28 Biofluid Mechanics: The Human Circulation

deceleration between the two locations and for steady flow conditions
(where ∂∂t = 0 ), it then becomes zero. Furthermore, since the points 1 and 2
are arbitrary, we can write a general expression of the steady flow Bernoulli
equation as

V2
p+ρ + ρgz = H (1.52)
2

where p , V, and z are evaluated at any point along the streamline. Here, H
is a constant and is referred to as the “total head” or total energy per unit
volume of fluid. This terminology is used because, even though we derived
this equation based on Conservation of Momentum principles, the terms in
Equation 1.52 all have the units of F/L2, which equivalently can be written
as FL/L3 or Energy/Volume.
In the absence of frictional losses, then, the Bernoulli equation states that
the total mechanical energy per unit volume at any point remains constant.
The mechanical energy of a system consists of a pressure energy component
(due to static pressure), a kinetic energy component (due to motion), and a
potential energy component (due to height) assuming there are no thermal
energy changes. Thus, between any two points, the form of the mechanical
energy may change from kinetic to potential and vice versa, for example,
but the total mechanical energy does not change.
The Bernoulli equation also states that the pressure drop between two
points located along a streamline at similar heights is only a function of the
velocities at these points. Thus, for z1 = z2, Bernoulli’s equation simplifies to

1
p1 − p2 = ρ(V22 − V12 ) (1.53)
2

Finally, when the downstream velocity is much higher than the upstream
velocity (V2  V1 ) , ρV12 can be neglected and Equation 1.53 then can be fur-
ther simplified to

1 2
p1 − p2 = ρV (1.54)
2 2

In the specific case of blood flow where we can assume a density of 1.06 g/cm3,
Equation 1.55 becomes

p1 − p2 = 4V22 (1.55)

where pi is in mmHg, V2 in m/s, and the constant term, 4, has units of

[mmHg/(m/s)2]. Equation 1.55 is referred to as the simplified Bernoulli
equation, and has found wide application clinically in determining pressure
7328_C001.fm Page 29 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 29

drops across stenoses, or vascular obstructions, using velocity measure-

ments made from noninvasive instruments such as Doppler ultrasound and
magnetic resonance phase velocity mapping (see Chapter 10, sections 10.6
and 10.8).
It is important to keep in mind that the Bernoulli equation is not valid in
cases where viscous forces are significant, such as in long constrictions with
a small lumen diameter, or in flow separation regions. In addition, it does
not account for turbulent energy losses in cases of severe stenosis since they
are time-varying. Such losses should be taken into account in any calculation
because they may substantially reduce the energy content of the fluid.

Example 1.2:
A patient has a stenotic aortic valve producing a pressure drop between the
left ventricle and the aorta. The mean velocity in the left ventricle proximal
(upstream) to the valve is 1 m/s, while the mean velocity in the aorta distal
(downstream) of the valve is 4 m/s. Applying Equation 1.55, the pressure
drop across the valve is:

p1 − p2 = 4V22

= 4(4 ms )
2

= 64 mmHg
By comparison, if we include the proximal velocity in the calculation, then
the pressure drop reduces to

p1 − p2 = 4 (V22 − V12 )

= 4 ( 4 m s ) − (1 m s ) 
2 2
 
= 60 mmHg

This example demonstrates an important clinical use of the Bernoulli

equation, since the degree of pressure drop is a good indicator of the
severity of the stenosis and, thus, of the need for treatment. Unfortunately,
it is a somewhat risky and complicated procedure to obtain pressure data
directly since it requires inserting a catheter manometer (pressure meter)
into the aorta. A much simpler and noninvasive technique — Ultrasound
Doppler Velocimetry (see Chapter 10, section 10.6) — is capable of mea-
suring the flow velocity in the valve, thus enabling an estimate of the
pressure drop by means of the simplified Bernoulli equation. Several cau-
tions should be observed in using this approach, however, due to the basic
assumptions made in deriving the original Bernoulli equation and those
made in further simplifying it, which may either over- or underestimate
its true value.
7328_C001.fm Page 30 Friday, October 6, 2006 3:48 PM

30 Biofluid Mechanics: The Human Circulation

1.7 Dimensional Analysis

If we return to the general form of the Conservation of Momentum or the
Navier–Stokes equations, we can write


∂V    
ρ + ρV ⋅ ∇V = −∇p + ρ g + µ∇ 2 V (1.56)
∂t

  
where ρ ∂∂Vt is the local acceleration, ρV ⋅ ∇V is the convective accelera-
tion, −∇p is the pressure
 force per unit volume, ρg is the body force per unit
volume, and µ∇ 2 V is the viscous forces per unit volume.
If we represent


g = −∇φ (1.57)

where

φ = −( g x x + g y y + g z z ) (1.58)

then,


−∇p + ρ g = −∇p − ρ∇φ = −∇( p + ρφ) = −∇p (1.59)

or


∂V   
ρ + ρV ⋅ ∇V = −∇p + µ∇ 2 V (1.60)
∂t

Based on the principles of Dimensional Analysis, the dependent vari-

ables – V and p are functions of the independent variables x , y , z, and t as
well as the constants ρ, µ, and g , along with any other parameters that
appear in the boundary conditions. Furthermore, examination of the above
equation shows that each term has units of force per unit volume, or F L3 .
Multiplying numerator and denominator by L results in an equivalent
expression ( F⋅L) ( L3 ⋅L), or Energy/(Volume·L). Now, if we divide each term by
a constant having those same units, we would obtain a dimensionless
equation. For the moment, let’s designate a characteristic (i.e., reference or
representative) length and a characteristic velocity as Lo and υ o , respectively.
7328_C001.fm Page 31 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 31

ρV 2
Now, the ratio of 0 L0 also has the dimension of ( F⋅L) ( L3 ⋅L) , so multiplying
L
every term by 0 ρV 2 would result in
0


V
∂    
 Vo   V  V  p   µ  2 2 V
+ ⋅ ( L ∇) = ( Lo∇)  2  +  ( Lo∇ )  
 V   Vo  o  Vo 
(1.61)
 ρVo   ρVo Lo   Vo 
∂t o 
 Lo 

where individual terms have been grouped into dimensionless variables

defined as


V
 =   , dimensionless velocity
 Vo 

 = L ∇ , dimensionless vector gradient operators

∇ o

 P 
P =  2  , dimensionless pressure
 ρVo 

υ 
Τ = t  o  , dimensionless time
 Lo 

 ρV L 
Re =  o o  , dimensionless parameter, called the Reynolds number
 µ 

Writing the equation in terms of these dimensionless variables, we get the

dimensionless form of the Navier–Stokes equation

∂ p+ 1 ∇
  = −∇  2
+  ⋅∇ (1.62)
∂T Re

For incompressible flows, the dimensionless Continuity Equation becomes

 ⋅ = 0
∇ (1.63)

Examination of Equation 1.62 and Equation 1.63 shows that, for a given
value of Re, there is only one solution to these equations for each driving
 . One consequence of this is that by matching values of Re for
function, ∇p
various systems, which are similar but of different scale, it is possible to
achieve dynamic similarity and obtain predictive information about proto-
types from scaled model measurements.
7328_C001.fm Page 32 Friday, October 6, 2006 3:48 PM

32 Biofluid Mechanics: The Human Circulation

Example 1.3:
It is desired to study the fluid dynamics in a coronary artery (diameter = 3 mm)
using a laser Doppler velocimetry (see Chapter 10, section 10.7: Laser
Doppler Velocimetry). However, the resolution of the LDV system is 300 µm,
resulting in a maximum of 10 velocity points across the artery diameter and
low accuracy for obtaining wall shear stresses from the fitted velocity profile.
In order to improve the resolution by ×10, an in vitro model is constructed
with a diameter of 3 cm. What flow rate should be used to ensure dynamic
similarity between the model and the prototype? (Assume the working fluid
has the same kinematic viscosity as blood — ν = 0.035 cm2/s).
Since Dp = 3 mm, Dm = 30 mm, dynamic similarity can only be achieved if

Rem = Re p

or

VmDm VpDp
=
νm νp

For νm = ν p,

Dp
Vm = V = 0.1Vp
Dm p

Therefore, flow in the model must be:

 πD2   π10Dp2 
Qm = Vm  m  = 0.1Vp   = 10Qp
 4   4 

1.8 Fluid Mechanics in a Straight Tube

Much of the blood flow in the human circulation occurs within tubular
structures — arteries, capillaries, veins, etc. It is for that reason that the study
of fluid mechanics in a straight tube is of particular interest in biofluid
mechanics. While the human vasculature is not strictly a series of straight
tubes of constant diameter, results from this analysis do provide good esti-
mates or starting points for further evaluation. Before we look at this topic
in detail, however, we need to make several definitions of terms, which are
relevant to common flow conditions.
7328_C001.fm Page 33 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 33

(τ + dτ dr) 2π(r + dr) dx

dr

dr
r τ(2πr dx)

dp
(p + dx) 2πr dr
dx
p(2πr dr) dx

FIGURE 1.10
Force balances for steady flow through a straight, horizontal, circular tube.

Blood flow in the circulatory system is invariably unsteady and, in most

regions (e.g., the systemic arteries and the microcirculation), it is pulsatile.
The term “Unsteady” is very general and refers to any type of flow that is
simply not constant. If the flow has a periodic behavior and a net directional
motion over the cycle (i.e., the average flow is > 0), then it is called Pulsatile.
On the other hand, if the flow has a periodic behavior but oscillates back
and forth without a net forward or reverse output (i.e., the average flow = 0),
it is called Oscillatory Flow. Despite the fact that blood flow is unsteady, it
is helpful to first describe the principles of fluid flow under more simplified
conditions before moving to complex physiologic situations. The simplest
case to consider, therefore, is that of steady flow of a Newtonian fluid through
a straight, rigid, circular tube aligned in a horizontal position (Figure 1.10).

1.8.1 Flow Stability and Related Characteristics

The nature of flow of a Newtonian fluid in a straight, rigid, circular tube is
controlled by the inertial (accelerating) and the viscous (decelerating) forces
applied to the fluid elements. When viscous forces dominate, the flow is called
laminar and is characterized by a smooth motion of the fluid. Laminar flow
can be thought of as if the fluid is divided into a number of layers flowing
parallel to each other without any disturbances or mixing between the layers.
On the other hand, when inertial forces strongly dominate, the flow is called
turbulent. Here, the fluid exhibits a disturbed, random motion in all direc-
tions, which is superimposed on its repeatable, main motion.

1.8.1.1 Steady Laminar Flow

The key characteristic of laminar flow is that it is well organized and very
efficient, whereas turbulent flow is chaotic and accompanied by high energy
losses. Therefore, turbulent flow is undesirable in blood circulation because
of the excessive workload it would put on the heart and also because of
potential damage to blood cells. A helpful index used to determine whether
the flow in a tube is laminar or turbulent is the ratio of the inertial forces to
7328_C001.fm Page 34 Friday, October 6, 2006 3:48 PM

34 Biofluid Mechanics: The Human Circulation

the viscous forces. This ratio is classically known as the Reynolds Number
(Re), which is dimensionless since both terms have units of force [F]. It is
defined as

Inertial forces ρVd

Re = = (1.64)
Viscous forces µ

where ρ [kg/m3] is the density of the fluid, V [m/s] is the average velocity
of the fluid over the cross section of the tube, d [m] is the tube diameter,
and µ [kg/m·s] is the dynamic viscosity of the fluid. Although inertial forces
obviously begin to dominate for Re > 1, it has been determined experimen-
tally that in a smooth-surfaced tube, flow is laminar for all conditions where
Re < 2100. Furthermore, if the tube is long enough to have stabilized any
entrance effects (see section 1.8.1.3: Flow Development, below), then the
velocity profile takes on a parabolic shape and the flow is called fully devel-
oped laminar flow.
In section 1.4: Macroscopic Balances of Mass and Momentum, we dis-
cussed the principle of Conservation of Momentum and derived the
Navier–Stokes Equations. For incompressible Newtonian flow, the equation
of motion in vector notation is given by Equation 1.45


∂V   1  
+ (V ⋅∇)V = − ∇p + g + ν(∇ 2V ) (1.45)
∂t ρ

 
where V [m/s] is the velocity vector, p [Pa] is the pressure, g [m/s2] is the
gravitational acceleration, and ρ [N-s2/m4] and ν [m2/s] are the density and
the kinematic viscosity of the fluid, respectively.
If we now apply these equations (in cylindrical coordinates, [Table 1.1]) to
the case of steady flow in a straight, circular, horizontal tube (Figure 1.10),
then the momentum balance in the z (axial) direction reduces to

∂p  1 ∂  ∂Vz  
− +µ r  = 0 (1.65)
∂z  r ∂r  ∂r  

since the time rate of change (i.e., ∂ ∂t ), secondary velocity (i.e., Vr and Vθ ),
∂V
and circumferential velocity gradient (i.e., ∂θz ) terms are zero. As a conse-
∂V
quence, the conservation of mass balance results in z ∂z also being zero.
Rearranging terms then yields

∂p  1 ∂  ∂Vz  
= µ r  (1.66)
∂z  r ∂r  ∂r  
7328_C001.fm Page 35 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 35

Since pressure is only a function of length and axial velocity is only a

function of radius, however, this equation can be written equivalently in
terms of ordinary derivatives, or

dp  1 d  dVz  
= µ r  (1.67)
dz  r dr  dr  

We can further observe that, for the two terms of the equation (i.e., L.H.S.
and R.H.S.) to be equal for all values of the independent variables r and z
(each of which is only present in one of the terms), each term must be
constant. Equation 1.67 can then be integrated twice to yield

dVz 1  dp  c
= r+ 1
dr 2µ  dz  r

1  dp  2
Vz = r + c1 ln r + c2 (1.68)
4µ  dz 

The constant terms, c1 and c2, can be evaluated by applying known values
of axial velocity at specific boundary locations. For example, Vz = 0 at r = R
due to the “no-slip” condition at the tube wall. The value of Vz , however,
is not known at the tube center, r = 0, although we can assume that it is a
maximum at that point due to overall symmetry of the tube. Thus, the
dV
appropriate boundary condition here is that drz = 0, which requires that c1 = 0
and which also constrains all velocities to be finite.
Evaluating c2 and substituting it into Equation 1.68 results in

1  dp  2
Vz = [r − R 2 ] (1.69)
4µ  dz 

If we replace the differential pressure gradient term by the pressure gra-

dient along the entire tube, ∆p
L , then the velocity variation, or “profile” in a
tube is given by

 ∆pR 2    r  
2
Vz (r ) =   1 −   
 4µL    R  

or

  r 2 
Vz (r ) = Vmax 1 −    (1.70)
  R  
7328_C001.fm Page 36 Friday, October 6, 2006 3:48 PM

36 Biofluid Mechanics: The Human Circulation

where V [m/s] is the velocity of the fluid at a distance r [m] from the center
of the tube, Vmax [m/s] is the maximum (centerline) velocity, R [m] is the
radius of the tube, d [m] is the diameter of the tube, and ∆p [Pa] is the
pressure drop along a length L [m] of the tube. By integrating this velocity
profile over the tube’s cross section and dividing by the area, we can obtain
the average velocity,

Vmax
Vave = (1.71)
2

Since flow rate in the tube, Q [m3/s], is equal to the average velocity, Vave ,
times the cross-sectional area, we can write

Vmax
Q = Vave ( πR 2 ) = ( πR 2 )
2
(1.72)
 ∆pR 2  ∆pπR 4
=  ( πR 2 ) =
 8µL  8µL

In terms of tube diameter, this becomes

∆pπ d 4
Q= (1.73)
128µ L

Solving for the pressure difference, we obtain

µLQ
∆p = 128 (1.74)
πd 4

which is commonly known as the Hagen–Poiseuille Equation.

1.8.1.2 Turbulent Flow

Because of our assumption in the above derivation that the flow was steady,
the Hagen–Poiseuille equation will only be valid for laminar flow condi-
tions; i.e., Re < 2100. For Re > 2100, the flow starts to become turbulent in
parts of the fluid and particle motions begin to vary with time. Such flow
is called transitional flow and the range of Re for which this type of flow
exists varies considerably, depending upon the application (i.e., system
configuration and stability) and the roughness of the tube surface. Usually,
flows in a smooth-surfaced tube with 2100 < Re < 4000 are characterized as
transitional. Fully developed turbulent flow is observed starting at Re = 104
for a rough-surfaced tube and starting at Re = 108 for a smooth-surfaced
tube. The velocity profile of fully developed turbulent flow is flatter than
7328_C001.fm Page 37 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 37

1.0
n = 10

n=6

Laminar n=8

r
R 0.5

Turbulent

0
0 0.5 1.0
–
µ
Vc

FIGURE 1.11
Plots of laminar (Poiseuille) and turbulent velocity profiles for n = 6, 8, and 10.

that for laminar flow, although it is not considered uniform or “plug” flow
(Figure 1.11).
In order to quantitatively describe turbulent flow, we think of the instan-
taneous velocity, V (t), at a given radial location as the sum of a time-averaged
mean velocity, V [m/s], and a randomly fluctuating velocity component,
V ′ [m/s], or

V(t) = V + V ′ (1.75)

Expressing turbulent flow in this way allows us to separate out one com-
ponent of a given variable, which can be quantified and expressed as an
analytic function while also allowing us to modify each variable by super-
imposing a random, disorderly motion on top of it. In the case of velocity,
the mean velocity profile (i.e., the quantifiable component) of a fully devel-
oped turbulent flow at Re < 105 is described by

1
 r n
V = V max  1 −  (1.76)
 R
7328_C001.fm Page 38 Friday, October 6, 2006 3:48 PM

38 Biofluid Mechanics: The Human Circulation

where V [m/s] is the time-averaged velocity at a distance r from the center

of the tube and V max [m/s] is the maximum (centerline) time-averaged
velocity. The exponent, 1/n, (where 6 < n < 10) is not constant, but decreases
as Re increases. When n = 7, for example, the average cross-sectional velocity
is equal to 4/5 of the maximum centerline velocity. This illustrates the flatter
nature of the velocity profile compared to that for the laminar flow case.
The reason is that axial momentum under turbulent conditions is rapidly
transported across the tube diameter due to nonzero radial velocity com-
ponents, reaching a plateau level closer to the tube wall. Under laminar
flow, this can only occur as a result of relatively slow viscous interactions
between layers since radial velocity is zero. The amount of turbulence
present in a flow field is quantified by the turbulence intensity, I, and can
be determined by

Vrms
I= 100% (1.77)
V ave

where Vrms = V ′ 2 is the root mean square of the fluctuating velocity [m/s].
It should be emphasized that blood flow in the circulatory system is
normally laminar, although in the ascending aorta it can destabilize briefly
during the deceleration phase of late systole; however, this time period is
generally too short for flow to become fully turbulent. Certain disease con-
ditions, though, can alter this condition and produce turbulent blood flow,
particularly downstream of an arterial stenosis (a vessel narrowing, usually
due to atherosclerosis), distal to defective (i.e., stenotic or regurgitant) heart
valves, or distal to prosthetic heart valves.

1.8.1.3 Flow Development

We should further point out that the above velocity profiles, both laminar
and turbulent, are only valid in a fully developed flow region; that is, at a
distance in a straight, smooth-surfaced tube that is far enough from the
entrance for any upstream disturbances to have dissipated. These distur-
bances can be caused by a number of geometric changes or discontinuities,
such as vessel curvature or taper, presence of a bifurcation, an anastomosis,
an aneurysm, an occlusion, etc. The effects of these flow disturbances are
significant only in the immediate region downstream of the discontinuity. It
is in this region that “flow development’”or stabilization occurs and, down-
stream of this region, the flow profile is called “fully developed.” Even
though these changes are occurring in the flow, it is important to remember
that, in both the flow development and the fully developed flow regions,
the nonslip condition at the wall remains valid and that the average velocity
across the tube will remain the same for a constant cross-sectional area
because of the Principle of Mass Conservation.
7328_C001.fm Page 39 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 39

r
U0 u D
x

Entrance length Fully-developed

velocity profile

FIGURE 1.12
Flow in the entrance region of a pipe.

An illustration of a flow development region can be seen in Figure 1.12,

which shows the flow field in a small diameter tube downstream of a gradual
outlet from a large tank.
Immediately after entering the tube, the flow profile is essentially flat since
the flow is dominated by higher velocity particles near the tube center. Just
downstream of the entrance, the layer of fluid closest to the wall decelerates
as some particles adhere to the wall (dashed lines in Figure 1.12). It then
slows adjacent inner fluid layers through viscous (shear) forces. At the same
time, the fluid layers near the center of the tube accelerate since the axial
pressure drop is not opposed by large frictional forces in this region. This
acceleration is also necessary in order to satisfy the Mass Conservation
principle. The region of flow near the tube inner surface where viscosity
effects alter the velocity profile is called the boundary layer.
For laminar flow, the thickness of this boundary layer, δ, increases with
distance downstream because of the continued action of significant drag or
retarding forces acting on the fluid through its viscosity by the tube wall.
When the boundary layer finally grows enough to fill the entire tube cross
section (i.e., δ = D/2), the flow is said to have become fully developed, i.e.,
there is no room left for further adjustment to take place. From this position
on, all flow conditions (velocity profile, pressure gradient, etc.) remain con-
stant. For turbulent flow, the viscous forces are insignificant compared to iner-
tial forces (as seen by the extremely high Re values required for turbulence)
everywhere except in a small region very close to the wall. Here, there is
a laminar (since the local Re is relatively low) boundary layer, which is
very thin. For both laminar and turbulent cases, the axial length required
for this flow development to occur, known as the entrance length, Le, can
be estimated as

Le = 0.06 ⋅ d ⋅ Re for laminar flow (1.78)

1
Le = 0.693 ⋅ d ⋅ Re 4 for turbulent flow (1.79)
7328_C001.fm Page 40 Friday, October 6, 2006 3:48 PM

40 Biofluid Mechanics: The Human Circulation

Example 1.4:
For flow in the proximal aorta with a diameter of ~ 2.5 cm (0.025 m) and an
average velocity of ~ 0.25 m/s, the Reynolds number would be:

ρVD
Re = = (1.07 g/cm3 × 25 cm/s × 2.5 cm)/(0.035 Poise) [in cgs system]
µ
~ 1,900

This would result in an entrance length on the order of 2.5 to 3 m. Since

an adult human aorta is < 1 m in length, this means that flow would be
within an entrance length in the entire aorta, implying that it never becomes
fully developed.

Example 1.5:
Flow in the intercostal arteries derives from the thoracic aorta (diameter of
~ 2.0 cm) as they branch off to perfuse the musculature of the chest wall. These
vessels have a diameter of ~ 2 mm with an average velocity of ~ 10 cm/s.
Therefore, the Reynolds number would be:

ρVD
Re = = (1.07 g/cm3 × 10 cm/s × 0.2 cm)/(0.035 Poise)
µ
~ 60

In this case, the entrance length would be on the order of 0.7 cm. As a
result, flow development occurs in a relatively short distance and the major-
ity of the intercostal artery (several centimeters long) experiences Poiseuille-
type flow (in the mean).

1.8.1.4 Viscous and Turbulent Shear Stress

In the study of hemodynamics and, in particular, of vascular disease, one of
the most important variables is the shear stress, τ [N/m2], at the vessel wall
(τ w ) . Wall shear stress has considerable clinical relevance because it provides
information about both the magnitude of the force that the blood exerts on
the vessel wall as well as the force exerted by one fluid layer on another. In
healthy blood vessels, the shear stress is generally low (~ 15 to 20 dynes/cm2)
and is not harmful to either the blood cells or to the cells, which line the
inner surface of the vessel, called endothelial cells. Shear stress varies with
flow conditions (cardiac output, heart rate, etc.) as well as with the local
geometry of the vessel (curves, branches, etc.) as will be discussed in later
sections. Excessively high levels of shear stress caused, for example, by
atherosclerotic lesions or artificial heart valves, may damage the red blood
cells (a condition called “hemolysis”) or the endothelium of the vessel wall.
Other abnormal shear stresses, such as very low or strongly oscillatory shear
stresses, may also change the biological behavior of some cells, such as
7328_C001.fm Page 41 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 41

platelets in the blood stream in which they become activated, leading to

thrombus formation. These stresses may also act on endothelial cells lining
the vessel wall, which then secrete vasoactive compounds, leading to vessel
constriction or wall hypertrophy. [Note: Each of the above topics will be
discussed in greater detail in subsequent chapters.]
Based on our earlier discussions, we can readily determine the shear stress
for laminar flow of a Newtonian fluid as being linearly related to the shear
rate (dV/dr) according to Equation 1.1 expressed in terms of cylindrical
coordinates

dV
τ = −µ (1.1)
dr

where V is the velocity [m/s] at radial position r [m] and µ is the dynamic
viscosity [N-s/m2] of the fluid. For turbulent flow, we said that the boundary
layer is very thin and the velocity profile is flatter than for laminar flow. The
combination of a flatter profile with the nonslip condition at the wall results
in much higher shear rates and, thus, wall shear stresses for turbulent flow
are large as compared to laminar flow. For either case (laminar or turbulent),
the wall shear stress can be determined from a force balance within a control
volume if the pressure drop, ∆p, is known along a length L of the tube (refer
again to Figure 1.10)

d ∆p
τ wall = − (1.80)
4 L

As mentioned earlier, both viscous and turbulent shear stresses, if large

enough, can potentially activate or lyse (i.e., rupture) blood cells. However,
the origin and, consequently, the scale of viscous and turbulent shear stresses
are different. Viscous shear stresses act on a molecular scale, i.e., they arise
from the tendency of one molecule to remain in close proximity to its neigh-
bors. This is quantified in fluids through the measure of their “viscosity.”
Because viscous stresses act on a scale much smaller than the diameter of a
blood cell (on the order of a few µm), blood cells will always experience a
viscous shear stress if one is present. In contrast, turbulent shear stresses
arise from the inertia contained within the fluctuations of turbulent flow.
Turbulent shear stresses act on a scale comparable to the length of the
smallest turbulent fluctuations, which are often much larger than the diam-
eter of a blood cell. If such a turbulent shear stress is present, a blood cell
may not experience it, particularly if it resides within a turbulent eddy.

1.8.2 Effect of Flow Pulsatility

The previous section outlined the characteristics of steady flow in a tube
when the driving pressure gradient is constant. However, we said earlier
7328_C001.fm Page 42 Friday, October 6, 2006 3:48 PM

42 Biofluid Mechanics: The Human Circulation

that blood flow in the heart and arteries is quite pulsatile, meaning that
the pressure and velocity profiles vary periodically (here, the period
denotes the duration of a cardiac cycle [s]) with time. When the heart
contracts during systole, a pressure originates from the left ventricle, which
then travels out as a wave due to the elasticity of the arteries. The most
immediate consequence of this is that the pressure in the left ventricle
(upstream) exceeds that in the aorta (downstream) resulting in opening of
the aortic valve and the blood being ejected from the heart. Due to the
compliant nature of the arteries and their finite thickness, the pressure
travels like a sound wave at a speed, which is much faster than the flow
velocity (~ 500:1). The pulsatile nature of the flow also affects the pressure
distribution out into the vessels and the velocity profiles within them as
well as the point of transition from a laminar to a turbulent regime. This
later effect is due to the fact that flow accelerates rapidly in early systole,
when, based on the instantaneous Reynolds number in the ascending aorta,
flow would be expected to be turbulent during a major part of systole.
Despite this, however, it has been observed that aortic blood flow remains
laminar and well streamlined under these conditions. The reason for this
is partly due to the stabilizing effect that systolic acceleration has on the
flow. One might anticipate, then, that the onset of turbulence would occur
during the later part of systole when flow decelerates. This has, in fact,
been shown by the presence of erratic fluctuations in velocity recordings
(Figure 1.13).

100

50
Velocity (cm s–1)

–20

0.0 0.5 1.0 1.5

Time (s)

FIGURE 1.13
Disturbances during flow deceleration phase in a pulsatile (periodic unsteady) flow velocity
patterns in the aorta. (Redrawn from Seed, W.A. and Wood, N.B., Cardiovascular Research
5: 319–330, 1971. With permission.)
7328_C001.fm Page 43 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 43

The flow destabilization that occurs in late systole is mainly due to the
development of an adverse (i.e., against the flow) pressure gradient during
the flow deceleration, but that is not the entire reason. Additionally, flow
in the aorta doesn’t become turbulent as soon or to the degree expected
based on the instantaneous Re values because there is not sufficient time
available for flow to become turbulent. To understand this, think of tur-
bulence as a process that requires both a minimum amount of energy to
be present and a triggering or seeding mechanism to initiate it. In this
case, the triggering is provided by some “disturbance” to the flow (e.g.,
a surface roughness or an external vibration) at some site. For “fully
developed” turbulence to occur, however, this disturbance must propagate
throughout the entire flow field. With steady flow, these conditions are
always present, but with pulsatile flow, they only exist for a limited time
(up to the cycle period) and then they reverse. Therefore, in a healthy
artery, turbulence is generally absent because the flow destabilization time
interval (mid-to-end systole) is short (~150 ms) and is immediately followed
by diastole when the velocity and instantaneous Re are low. When the
subsequent systolic phase begins, a new acceleration phase restabilizes
the flow. It should be pointed out here that, although complex vortex-
containing flows are present in regions like bifurcations or branches, such
flow is not necessarily turbulent in the sense that it is either random or
unpredictable. Many of these flows contain very laminar flow behavior,
just not simply in linear directions.
Based on the above discussion, Re alone is clearly not enough to completely
characterize pulsatile flow. Another dimensionless parameter, the Womersley
number (denoted as α), is also used to characterize the periodic nature of
blood flow. The definition of α is:

d ρω
α= (1.81)
2 µ

where ω [radians/sec] is the heart rate. The Womersley number in unsteady

flow has a physical significance similar to that of Re in steady flow in that
it provides a comparison between unsteady inertial forces and viscous
forces. In the human circulatory system, α ranges from 10−3 in capillaries
to nearly 20 in the ascending aorta at rest. When < 1.0, viscous forces
dominate in every region in the tube and the conditions are known
as quasi-steady flow. As α increases, the inertial forces become more
important and start to dominate, initiating at the center of the tube. As a
result, a delay (with respect to the driving pressure gradient) can be
observed in the bulk flow, and the velocity profile becomes flat in the central
region of the tube.
7328_C001.fm Page 44 Friday, October 6, 2006 3:48 PM

44 Biofluid Mechanics: The Human Circulation

Example 1.6:
The ascending aorta has a diameter, d = 2.5 cm and the heart rate is 70 bpm
(beats per minute). The density, ρ, and the viscosity of blood, µ, are 1.06 g/cm3
and 0.035 Poise, respectively. Based on these data, the Womersley parameter
would be:

d ρω
α=
2 µ

= (2.5 cm/2) × [1.06 g/cm3 × 2 (70/60)/0.035 Poise]1/2

= 1.25 × 14.9
= 18.6

What this example illustrates is that flow in the major arteries is charac-
terized by both large mean flow and large oscillatory flow components due
to the intermittent nature of the heart pump. As we’ll see later in Chapter 6
(Unsteady Flow and Nonuniform Geometric Models), the Womersley param-
eter is an important index of the net amount of forward flow and the shape
of the local velocity profile.

1.9 Boundary Layer Separation

As we described earlier for flow development in a circular tube, when a
moving fluid contacts a solid boundary, the fluid immediately adjacent to
the boundary takes on the same velocity as the boundary. Since this velocity
is usually different from that of the free stream, a transition must occur
between the two locations and, thus, a boundary layer is formed. The bound-
ary layer is the region in which flow velocity changes from zero (relative to
the wall) to the value of the free stream. Although boundary layers may
actually be extremely thin, they are very important because it is only in this
region of the flow that viscous effects are significant. This may seem para-
doxical since the fluid has the same viscosity everywhere. However, for a
force to be exerted between fluid layers (i.e., presence of a shear stress), there
must be both a nonzero viscosity and a nonzero shear rate. These require-
ments are satisfied in the boundary layer where velocity varies with distance,
but they are not satisfied in the main flow where velocity is nearly uniform.
As the fluid proceeds downstream, viscous diffusion occurs farther into the
free stream as slower near-wall layers interact with faster free-stream layers
and viscous effects are felt at a greater depth from the wall. Thus, it is said
that the boundary layer “grows.”
When a fluid encounters a solid object and flows over it, it is forced to
deviate from an axial direction and, thus, changes direction in order to pass
7328_C001.fm Page 45 Friday, October 6, 2006 3:48 PM

Fundamentals of Fluid Mechanics 45

throughout the body. For this redirection to occur, a pressure gradient must
exist. The presence of an adverse pressure gradient (i.e., a pressure gradient
acting against the flow direction and tending to decrease the fluid velocity)
may have a magnitude that is large enough to cause the fluid within the
boundary layer to stop and deflect away from the surface. When this hap-
pens, boundary layer separation is said to have occurred. Boundary layer
separation begins at a separation point on the boundary, which is defined
as a location where, not only is the flow velocity, u, equal to zero, but the
rate of change of u along a direction normal to the surface, n, is also zero,
or ∂u/∂n = 0. Beyond the separation point, there may exist a region of
reversed, turbulent, or disturbed flow where recirculating or vortical flows
are commonly seen. These characteristics are important biologically since
they can lead to reverse and/or oscillatory shear stresses and to long particle
residence times, both of which can have a wide variety of clinical implica-
tions, including the initiation of atherosclerosis (see Chapter 6, section 6.4:.
Wall Shear Stress and Its Effect on Endothelial Cells).
In pulsatile flows, flow separation can be generated by a geometric
adverse pressure gradient or by time-varying changes in the driving pres-
sure. Geometric adverse pressure gradients, for example, are present
behind all prosthetic heart valves because of their small orifice areas. As
the area downstream of a prosthetic heart valve increases, the mean flow
velocity must decrease in accordance with the continuity equation. This
decrease in velocity is also accompanied by an adverse pressure gradient
due to the energy balance. Similar effects are seen as a result of the con-
tractile nature of the heart in which blood flow experiences both accelera-
tion and deceleration during a cardiac cycle. An adverse pressure gradient
exists during the deceleration phase of a particular flow and may lead to
boundary layer separation. Consequently, flow separation is even more
likely in regions where both geometric and temporal adverse pressure
gradients exist (e.g., arterial stenoses, stenotic heart valves, prosthetic heart
valves, and aneurysms).

1.10 Problems
1.1 For Hagen–Poiseuille flow, show that the shear stress at the tube wall
( τ w = τ r=R ) can be written as

4µQ
τw =
πR 3

1.2 For a laminar flow of a Newtonian fluid in a tube of circular cross

section, show that the ratio of mean shear to the wall shear rate is 2/3.
7328_C001.fm Page 46 Friday, October 6, 2006 3:48 PM

46 Biofluid Mechanics: The Human Circulation

1.3 For Hagen–Poiseuille flow, show that the mean velocity over the cross
section is 1/2 of the peak velocity in the tube.
1.4 What pressure will be required to force 1 cc/s of blood serum through
an intravenous tube of radius 0.50 mm and length 3 cm and into an artery with
a mean pressure of 100 mmHg? (Assume: Blood serum viscosity, µ = 7 × 10−3
Poise.)
1.5 A patient has atherosclerosis, which produces a stenosis of his aorta of
16% diameter reduction.

(a) What is the resultant reduction in flow rate?

(b) Assuming laminar flow, how much pressure increase is necessary
to compensate for this reduction?
7328_C002.fm Page 47 Friday, October 6, 2006 12:23 PM

2
Introduction to Solid Mechanics

CONTENTS
2.1 Introduction to Mechanics of Materials...................................................47
2.1.1 Elastic Behavior ...............................................................................47
2.1.2 Engineering and True Strain .........................................................50
2.1.3 Incremental Elastic Modulus.........................................................51
2.1.4 Poisson’s Ratio.................................................................................51
2.1.5 Shearing Stresses and Strains........................................................53
2.1.6 Generalized Hooke’s Law..............................................................53
2.1.7 Bulk Modulus ..................................................................................56
2.2 Analysis of Thin-Walled Cylindrical Tubes ............................................57
2.3 Analysis of Thick-Walled Cylindrical Tubes...........................................60
2.3.1 Equilibrium Equation .....................................................................60
2.3.2 Compatibility Condition ................................................................61
2.4 Viscoelasticity ...............................................................................................63
2.5 Problems........................................................................................................67

As blood flows in the arterial system in the human circulation, there is a constant
interaction between the flowing blood and the wall of the arteries. In order to
study the unsteady blood flow dynamics in the arteries, it becomes necessary
to consider the material behavior of the arterial wall. In this chapter, the basic
stress–strain relationship for an elastic material is reviewed and basic relation-
ships for stresses in thin-walled and thick-walled cylindrical tubes are derived
that will be useful in modeling unsteady blood flow in the arteries. Visco-elastic
behavior of materials is also included, as this has arterial wall significance.

2.1 Introduction to Mechanics of Materials

2.1.1 Elastic Behavior
A material is defined to be elastic if it deforms when a force is applied to
it, but returns to its original configuration without energy dissipation when
the applied force is removed. Consider the uniform thin cylindrical wire

47
7328_C002.fm Page 48 Friday, October 6, 2006 12:23 PM

48 Biofluid Mechanics: The Human Circulation

Stress (σ)
ᐍo
ᐍ

d
P
Strain (⑀)
(a) (b)

FIGURE 2.1
Stress–strain behavior of Hookean elastic material.

shown in Figure 2.1 with an undeformed length
0 subjected to a normal

tensile load P. The wire will elongate by a magnitude δ due to the axial load.
The average normal stress that the wire is subjected to can be computed as
σ = P A where A is the cross-sectional area of the wire in the undeformed
configuration. The normal stress has the units of force per unit area. In the
SI system, stress has the dimensions of N/m2 and is referred to as Pascal
(Pa). The magnitudes for stress occurring in engineering practice are rela-
tively large and, hence, prefixes, such as kPa (kilo Pascal = 103 Pa), MPa
(Mega Pascal = 106 Pa), or GPa (Giga Pascal = 109 Pa), are usually employed.
When the wire is stretched in the axial direction, there will be a reduction
in the cross-sectional area. This fact can be demonstrated easily by stretching
an elastic band and observing that the cross-sectional area decreases with
increase in axial stretch. The decrease in dimensions is not large enough to
be observed by visual inspection in most engineering materials. In usual
engineering practice, the stress is defined based on the cross-sectional area
of the undeformed configuration rather than the instantaneous value and
the error introduced will be negligible.
The ratio of the increment in length over the initial length is defined as
the normal strain. This relationship is given by

−
0 δ
ε= = (2.1)

0
0

In this equation,
is the instantaneous length. By varying the load, the

applied stress can be varied and by measuring the increase in length, the
corresponding strain can be obtained. The normal strain, ε, defined above
is the ratio of length units and, hence, is dimensionless. However, it is a
7328_C002.fm Page 49 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 49

common practice to specify normal strain as m/m or mm/mm. In engineer-

ing applications, the strain will be very small and will be in the range of
10−6 m/m or µm/m.
The stress–strain relationship for an ideal linear elastic material is shown
in Figure 2.1b. The slope of the stress–strain plot is a constant, E, referred to
as the elastic modulus or Young’s modulus. A material exhibiting a linear
stress–strain relationship is known as a Hookean elastic material, named
after Robert Hooke, who demonstrated such a behavior in 1678. The
Hookean relationship can be expressed as

σ = Eε (2.2)

It is apparent that the elastic modulus will have the same units as stress.
A material is said to be homogeneous if the elastic modulus, E, is the same
at every point on the bar. This will result in the elongation being uniformly
distributed along the length of the bar. A material is defined as isotropic if
the stress–strain behavior does not change when the direction of loading is
changed. The stress–strain diagram will be linear until the proportionality
limit is reached and then the relationship will become nonlinear. Engineering
materials can be divided into two broad categories based on the stress–strain
relationship. Figure 2.2 shows the typical stress–strain relationship for a
ductile and for a brittle material. For a ductile material, the stress initially
increases linearly with strain until a critical value of the yield stress, σy , is
reached. If the applied stress exceeds the yield stress for the material, the
material will be subject to a large deformation with a small increase in load
exhibiting plastic deformation.

σ
σU
Rupture σB ×
×
Stress (σ)

Stress (σ)

σY
σB

Yield Strain-hardening Necking

Strain (⑀) Strain (⑀)

(a) (b)

FIGURE 2.2
Stress–strain diagrams for typical ductile (a) and brittle (b) material.
7328_C002.fm Page 50 Friday, October 6, 2006 12:23 PM

50 Biofluid Mechanics: The Human Circulation

After the maximum value of stress, or the ultimate stress, σult, has been
reached, the diameter of a region of the specimen decreases significantly.
This phenomenon is referred to as necking. After necking occurs, a lower
load is sufficient to induce further deformation of the specimen until rup-
ture takes place. The load to induce rupture is known as the breaking
strength, σb, of the material. In the case of a brittle material, there is no
noticeable change in the rate of elongation before rupture takes place. In
such a material, the ultimate strength and the breaking strength are the
same. Any material for which the stress–strain curve does not follow the
linear relationship is classified as being nonlinear. We will observe later
that most biological materials generally do not exhibit a linear stress–strain
relationship.

2.1.2 Engineering and True Strain

In Equation 2.1, we defined strain as

−
0 δ
ε= =

0
0

This relationship is defined as the engineering strain where the ratio of the
elongation δ and the initial length of the specimen
0 are used. However,
strain can also be computed by using the successive values of the instanta-
neous length as they are recorded. Taking the value of the instantaneous
length and the corresponding increment in length, the true strain can be
defined as

∆

ε1 = ∑ ∆ε = ∑

Replacing the summation by integrals, the true strain can be expressed as

∫
= ln

d

ε1 = (2.3)
0

0

The relationship between the true strain and the engineering strain can be
obtained as

ε1 = ln(1 + ε) (2.4)

In most of the engineering materials undergoing small deformations and

strains, the magnitude of engineering strain will be very close to that of the
true strain.
7328_C002.fm Page 51 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 51

Stress (σ)

Einc = dσ/d⑀
dσ
d⑀

Strain (⑀)

FIGURE 2.3
Schematic of a nonlinear stress vs. strain relationship for biological soft tissue.

2.1.3 Incremental Elastic Modulus

As shown in Figure 2.1, any material that follows Hooke’s law will exhibit a
linear stress–strain relationship. However, when the material does not follow
Hooke’s law, it will have a nonlinear stress–strain relationship as shown in
Figure 2.3. In such cases, an elastic modulus can be defined as the slope of
the curve at any value of the stress and the corresponding strain. This value
is called the incremental modulus, and is given by the relationship

dσ
Einc = (2.5)
dε

The incremental elastic modulus (and the material stiffness) increases with
the increase in slope of the nonlinear stress–strain curve schematically shown
in Figure 2.3.

2.1.4 Poisson’s Ratio

As discussed earlier, when a slender bar is loaded axially (Figure 2.4), an
axial elongation takes place and a linear stress–strain relationship is observed
within the elastic limit. In this case, the normal strain can be given by the
relationship (from Equation 2.2)

σx
εx = (2.6)
E
7328_C002.fm Page 52 Friday, October 6, 2006 12:23 PM

52 Biofluid Mechanics: The Human Circulation

d
A

P
x

sx = P
A

FIGURE 2.4
An axially loaded slender bar.

In this equation, subscript x denotes the coordinate along the axial direc-
tion. It was also pointed out earlier that, when the bar elongates axially, there
would be a corresponding contraction in the transverse directions. In other
words, even if there is no applied stress in the transverse directions, there
will be strains present in these directions. The Poisson’s ratio is defined as

lateralstrain
ν=
axialstrain

or

−ε y −εz
ν= = (2.7)
εx εx

In this relationship, the material is assumed to be isotropic so that the

transverse strain in the y and z directions is the same. The negative sign in
Equation 2.7 is introduced to make the Poisson’s ratio a positive quantity
since εy will be negative for a positive εx and vice versa. From Equations 2.6
and Equation 2.7, we can derive

σx σx
εx = and ε y = ε z = − ν (2.8)
E E
7328_C002.fm Page 53 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 53

y y π +γ
␶yx xy
2

␶xy ␶xy

␶yx π _γ
xy
2
x x

FIGURE 2.5
An element subjected to shearing stresses.

2.1.5 Shearing Stresses and Strains

In Figure 2.5, a planar element is shown, which is subjected to shearing stresses
τxy and τyx. The first subscript of each variable denotes that the shearing stress
acts on a plane perpendicular to the axis denoted by that subscript and the
second subscript denotes the direction of the shear stress. The effect of such
stresses will be to deform the element such that two of the angles are reduced
from π2 to π2 − γ xy , while the other two angles are increased from π2 to π2 + γ xy . The
small angle γxy, expressed in radians, is defined as the shearing strain. When
the deformation results in a reduction of an angle formed by two faces oriented
towards the positive x- and y-axes, the shearing strain is defined as a positive
quantity; otherwise, it is negative. By measuring the values of the shearing
strains for various values of the shearing stresses, a shear stress–strain diagram
can be plotted for each material. The plots will be similar to those obtained for
the axial stress–strain relationship. For shearing stress values that do not exceed
the elastic limit, a Hooke’s law for shear can be written as

τ xy = Gγ xy (2.9)

In this relationship, G is the modulus of rigidity or the shear modulus.

In a similar manner, for a homogeneous isotropic material, two additional
relationships can be written as

τ yz = Gγ yz and τ zx = Gγ zx (2.10)

2.1.6 Generalized Hooke’s Law

Consider the general stress condition represented in Figure 2.6. As long as
none of the stresses represented here exceed the proportional or elastic limit,
7328_C002.fm Page 54 Friday, October 6, 2006 12:23 PM

54 Biofluid Mechanics: The Human Circulation

σy

␶yz ␶yx
␶zy ␶xy

␶xz
␶zx
σx
σz

x
z

FIGURE 2.6
General stress conditions on a cubic element.

the Principle of Superposition can be applied, and we can write the gener-
alized Hooke’s law as

σ x νσ y νσ y
εx = − −
E E E
σ y νσ x νσ z
εy = − −
E E E
(2.11)
σ νσ νσ y
εz = z − x −
E E E
τ xy τ yz τ
γ xy = ; γ yz = ; γ zx = zx
G G G

Here, σx, σy , and σz are the normal stresses in the respective coordinate
directions and εx, εy , and εz, the corresponding normal strains. The normal
strain, ε, is a measure of the elongation or contraction of a line segment in
the body. Hence, normal strains cause a change in the volume of a rectangular
element shown in Figure 2.6. τxy , τyz, and τzx are the shear stresses and γxy , γyz,
and γzx are the corresponding shear strains. In these terms, the first subscript
indicates the plane normal to the coordinate direction on which the stress
acts and the second subscript indicates the direction of the stress. The shear
strain is a measure of the change in angle between two small line segments
that are originally perpendicular to each other and, thus, represents a change
in the shape of the rectangular element.
7328_C002.fm Page 55 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 55

The above equations can be rewritten to express stress in terms of strain as

E
σx = (1 − ν)ε x + ν(ε y + ε z )
(1 + ν)(1 − 2 ν)
E
σy = (1 − ν)ε y + ν(ε z + ε x )
(1 + ν)(1 − 2 ν)
E (2.12)
σz = (1 − ν)ε z + ν(ε x + ε y )
(1 + ν)(1 − 2 ν)
τ xy = Gγ xy
τ yz = Gγ yz
τ zx = Gγ zx

In Equation 2.11, stresses exist only in the x – y plane if σz = 0. Such problems

are considered as plane stress analysis. In these cases, the stresses can be
expressed in terms of strain as

E
σx = (ε + νε y )
(1 − ν2 ) x
(2.13)
E
σy = (ε + νε x )
(1 − ν2 ) y

In the case of blood vessels, a subject of interest in this textbook, the vessel
geometry can be approximated as a tube and, hence, the use of polar (r, θ, z)
coordinates will be appropriate. The Hooke’s law in polar coordinates can
be written as

σ r νσ 0 νσ z
εr = − −
E E E
σ 0 νσ z νσ r
ε0 = − − (2.14a)
E E E
σ z νσ r νσ θ
εz = − −
E E E

The above equations can be rewritten to express stress in terms of strain as

E
σr = [(1 − ν)ε r + ν(ε θ + ε z )]
(1 + ν)(1 − 2 ν)
E
σθ = [(1 − ν)εθ + ν(ε z + ε r )] (2.14b)
(1 + ν)(1 − 2 ν)
E
σz = [(1 − ν)ε z + ν(ε x + ε θ )]
(1 + ν)(1 − 2 ν)
7328_C002.fm Page 56 Friday, October 6, 2006 12:23 PM

56 Biofluid Mechanics: The Human Circulation

and the corresponding expressions for shear stresses are:

τ rθ = G γ rθ

τθz = Gγ θz (2.14c)

τ zr = Gγ zr

In the case of plane stress analysis (where σ z = 0 ), the stresses can be

written in terms of strains as

E
σr = (ε + νεθ )
(1 − ν2 ) r
(2.15)
E
σθ = (ε + νε r )
(1 − ν2 ) θ

2.1.7 Bulk Modulus

Consider a cubic element (Figure 2.6) where each side of the cube has a unit
length. If only normal stresses σx, σy , and σz are present, the cube will deform
to the shape of a rectangular parallelepiped. The volume of the deformed
element is:

V = (1 + ε x )(1 + ε y )(1 + ε z ) (2.16)

Because the strains are much smaller than unity, their products will be
even smaller and can be neglected and, thus, V can be written as

V = 1 + εx + εy + εz (2.17)

Since the original volume V0 = 1, the change in volume is given by

V − V0 = ∆V = ε x + ε y + ε z

The change in volume per unit volume, also referred to as the volumetric
strain or cubical dilatation, is given by

∆V
= εx + εy + εz (2.18)
Vθ
7328_C002.fm Page 57 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 57

By rewriting the expressions for the strains in terms of stresses from

Equation 2.12, we have

∆V (1 − 2 ν)
= (σ x + σ y + σ z ) (2.19)
V0 E

This expression shows that when the Poisson’s ratio is 0.5, the volumetric
strain will be equal to zero. A special case to be considered is when the body
is subjected to a uniform hydrostatic pressure p. Then, each of the stress
components is equal to – p and the above relationship can be rewritten as

∆V −3(1 − 2V )
= p (2.20)
V0 E

Introducing the constant k = E

3(1− 2 ν ) , we have

∆V p
=− (2.21)
V0 k

The constant k is defined as the bulk modulus or the modulus of com-

pression. A body subjected to a uniform compressive force will decrease in
volume and, hence, the volumetric strain will be negative, yielding a positive
value for the bulk modulus. A material with an infinite bulk modulus is
known as an incompressible material. Water has a relatively high bulk
modulus (300,000 psi) and, thus, can be considered as incompressible for
most applications. In general, blood is also considered as an incompressible
fluid in the analysis of flow in the human circulation.
For a homogeneous isotropic material, an expression relating E, G, and ν
can be derived as

E
G= (2.22)
2(1 + ν)

Therefore, the complete material description of an isotropic, linear elastic

material can be specified by two independent constants where the third is
derived using the above relationship.

2.2 Analysis of Thin-Walled Cylindrical Tubes

As a first approximation, the geometry of major blood vessels can be consid-
ered as thin-walled elastic tubes of circular cross section. Therefore, in order
to model the dynamics of vascular tissues, we will develop a relationship for
7328_C002.fm Page 58 Friday, October 6, 2006 12:23 PM

58 Biofluid Mechanics: The Human Circulation

σθ
t
σz

σz

σθ

FIGURE 2.7
Stresses on a thin-walled cylindrical vessel.

the stresses present in the walls of tubes of cylindrical geometry. A cylindrical

thin-walled vessel of internal radius R and a thickness t is shown in
Figure 2.7. Because the vessel walls offer little resistance to bending, it can
be assumed that the forces developed in the wall are tangential to the surface
of the vessel as shown in the figure. Because of the axi-symmetric geometry,
no shear forces are generated and only the normal stresses along the axial
and circumferential directions exist. The normal stress in the circumferential
direction is the hoop stress and, in the axial direction, it is the longitudinal
stress. In order to derive an expression for the hoop stress, σθ , we will
consider a segment of the tube as shown in Figure 2.8a. The forces acting
on the segment are also shown in the figure. The pressure p in the figure
denotes the transmural pressure, which is the difference between the inside
and outside pressures. In employing this analysis for blood vessels, the
pressure acting on the outer surface of the vessel is usually neglected and

∆z
σdA t
t
σzdA

r
x r
pdA z
r
z
σdA

t
pdA
(a) (b)

FIGURE 2.8
Forces on segments of the cylindrical vessel.
7328_C002.fm Page 59 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 59

the blood pressure is the transmural pressure load resulting in the deforma-
tion of the vessel.
Static force equilibrium in the x direction can be written as

2 σ θt∆z − p(2 R∆z) = 0 (2.23)

and solving this equation, we obtain

pR
σθ = (2.24)
t

Similarly, assuming a closed-ended vessel, as in Figure 2.8b, force equilib-

rium along the z direction will yield the relationship

σ z (2 πRt) − p( πR 2 ) = 0 (2.25)

An expression for the longitudinal stress is derived from this as

pR
σz = (2.26)
2t

From Equation 2.24 and Equation 2.26, we note that

σ θ = 2σ z (2.27)

In open-ended tubes, σ z = 0 and only the hoop stress will exist. Note that
the radius R used in this derivation is the internal radius of the tube. The
initial circumferential length is 2πR and, if the radius increases by a small
amount, ∆R, due to the applied internal pressure, then the circumferential
strain can be computed as

∆R
εθ = [2 π(R + ∆R) − 2 πR] 2 πR = (2.28)
R

and the Young’s modulus expressed as

σ θ pR 2
E= = (2.29)
εθ t∆R
7328_C002.fm Page 60 Friday, October 6, 2006 12:23 PM

60 Biofluid Mechanics: The Human Circulation

In this derivation, we assumed the cylindrical tube to be thin-walled and,

hence, neglected the variation of stresses in the radial direction. This assump-
tion is valid only when t/R << 1. Usually, the assumption of a thin-walled
vessel is made only if the ratio is less than 0.1. When this condition is not
satisfied, the stresses developed in a thick-walled vessel need to be considered.

2.3 Analysis of Thick-Walled Cylindrical Tubes

If t/R ≥ 0.1 , then the thick-walled cylinder formulation needs to be employed
in the analysis of deformation under internal pressure loading. For many
cardiovascular applications of interest, the wall thickness-to-radius ratio of
arteries generally requires us to use the thick-walled cylindrical formulation.
The arterial wall is also tethered and, thus, a plane strain formulation
(displacement in the axial direction is neglected) is commonly used.

2.3.1 Equilibrium Equation

Consider an open-ended thick cylinder with an internal radius, R1, and exter-
nal radius, R2, with an internal pressure, p1, and external pressure, p2, as shown
in Figure 2.9. A differential element of thickness, dz, together with the radial
and circumferential stresses acting on the element are also shown in the figure.
Due to the axi-symmetric geometry and loading, there are no shear stresses
generated at either the inner or outer surfaces of the cylinder. An equilibrium
equation obtained by summation of forces in the radial direction yields

dθ
(σ r + dσ r )(r + dr )dθdz − σ r rdθdz − 2 σ θdrdz sin =0 (2.30)
2

P2
dz σr + dσr

dr σ
R2
R1 σr
dθ σ
o z Pl
Pl

P2

(a) (b)

FIGURE 2.9
Stresses on a thick-walled, open-ended vessel.
7328_C002.fm Page 61 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 61

Since Sin d2θ ≅ d2θ for small angles, after gathering terms, discarding higher
order terms, and dividing by rdrdθ dz, we obtain

dσ r ( σ r − σ θ )
+ =0 (2.31)
dr r

The above is the equilibrium equation for axi-symmetric cylinders.

2.3.2 Compatibility Condition

The strain displacement and stress–strain relations are developed as follows.
Displacements are functions of r only. All points deform radially, although
points at different radii deform by different amounts. From the definition of
strain, the radial and circumferential strains are written as

du
εr = (2.32)
dr

and

2 π(r + u) − 2 πr u
εθ = = (2.33)
2 πr r

where u is the radial displacement. Substituting the above compatibility

conditions into the stress–strain relationship, we have

E E  du u
σr = (ε + νεθ ) = +ν  (2.34)
1 − ν2 r 1 − ν2  dr r

and

E E u du 
σθ = (ε + νε r ) = +ν  (2.35)
1 − ν2 θ 1 − ν2  r dr 

Substituting the above into the equilibrium equation (Equation 2.31), we

obtain the second order differential equation

d 2 u 1 du u
+ − =0 (2.36)
dr 2 r dr r 2

The above equation incorporates equilibrium, stress–strain and compati-

bility conditions. The equation can be written in the form

d 1 d 
 (ur )  = 0 (2.37)
dr  r dr 
7328_C002.fm Page 62 Friday, October 6, 2006 12:23 PM

62 Biofluid Mechanics: The Human Circulation

Following the two integrations, a function u(r) satisfying the above equation
is given as

c2
u = c1r + (2.38)
r

where c1 and c2 are constants of integration. Substitution of Equation 2.38

into Equation 2.34 and Equation 2.35 yields

E  1− v 
σr =  (1 + v)c1 − 2 c2  (2.39)
1− v 
2 r 

and

E  1− v 
σθ = (1 + v)c1 + r 2 c2  (2.40)
1 − v2  

The constants c1 and c2 in Equation 2.39 and Equation 2.40 are evaluated
by applying the boundary conditions. From Figure 2.9, the boundary con-
ditions can be specified as σr = −p1 at r = R1 (inner surface) σr = −p2 and at
r = R2 (outer surface). The negative signs in the above equations indicate that
the pressures act into the surface as shown in Figure 2.9. Substituting for c1
and c2 in the above equations, the radial and circumferential stresses are
derived as

 R2   R2   R2   R2 
σ r = p1  2 1 2  1 − 22  − p2  2 2 2  1 − 21  (2.41)
 R2 − R1   r   R2 − R1   r 

and

 R2   R2   R2   R2 
σ θ = p1  2 1 2  1 − 22  − p2  2 2 2  1 + 21  (2.42)
 R2 − R1   r   R2 − R1   r 

The stress equations (Equation 2.41 and Equation 2.42) were first published
in 1833 and are known as the Lame relationships.
In the stress analysis of blood vessels, the external pressure p2 is generally
assumed to be zero and the force acting on the vessel wall is the transmural
blood pressure. The expression for the radial displacement in an artery
considered as a thick-walled tube is given by the expression

 p R 2 (1 + v)(1 − 2 v) r   p1R22 R12 (1 + v) r 

u=  1 1 2 +   (2.43)
 (R2 − R12 ) E   (R22 − R12 )r 2 E 
7328_C002.fm Page 63 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 63

While the above equation can be used to determine the deformation of

thick-walled cylinder with known internal pressure and material properties,
in practice it is used to establish the material property employing the defor-
mation measured experimentally. Rearranging the above equation in terms
of the elastic modulus, we obtain

 p R 2 (1 + ν)(1 − 2 ν) r   p1R22 R12 (1 + ν) r 

E= 1 1 2 +   (2.44)
 (R2 − R12 ) E   (R22 − R12 )r 2 u 

Bergel (1961a and 1961b) obtained an expression for the incremental mod-
ulus corresponding to an increase in outer radial dimension ∆R2 due to an
increase in internal pressure ∆p. By replacing p1 with ∆p, r with R2 and u
with ∆R2 in the above equation, the incremental modulus can be shown to be:

2(1 − ν2 )R12 R2 ∆p
Einc = (2.45)
(R22 − R12 )∆R2

In this expression, (R22 − R12 ) is a constant since the wall is assumed to be

incompressible and the vessel wall is tethered. This expression requires
measurement of the internal pressure, the radius, and the wall thickness of
the blood vessels in situ. The magnitude of the incremental modulus will
depend upon the computed mean pressure.
In several cases (including blood vessels in vivo), the internal pressure and
the external radius can be measured. In such cases, a pressure-strain mod-
ulus, Ep is given by

R2
Ep = ∆ p (2.46)
∆R2

This relationship ignores the tube wall thickness and, thus, Ep represents a
structural modulus rather than the elastic property of the tube material.

2.4 Viscoelasticity
The relationships for material properties derived above are applicable for
purely elastic material in which a load applied to the material produces an
instantaneous response in the form of deformation. Upon removal of the
external load, the specimen will instantaneously return to the undeformed
state (assuming inertial effects are small). In the presence of viscous behavior,
however, recovery of deformation will not occur instantly. This is because
the applied shear stress in viscous material is a function of rate of shear,
7328_C002.fm Page 64 Friday, October 6, 2006 12:23 PM

64 Biofluid Mechanics: The Human Circulation

σ σ

G
µ
1 1
⑀ η ⑀
G

⑀ ⑀

σ σ σ σ
(a) (b)

G µ
σ σ σ
σ
µ

(c) (d)

FIGURE 2.10
A typical viscoelastic material schematically represented by (a) a linear spring representing the
elastic component with the corresponding stress–strain relationship and (b) the viscous com-
ponent represented by a dashpot with the corresponding stress–strain rate relationship; The
combined viscoelastic behavior of a solid represented by (c) a Maxwell model with the spring
and dashpot in series, and (d) a Kelvin model with spring and dashpot in parallel.

as discussed in Chapter 1, section 1.2: Intrinsic Fluid Properties. Material

behaviors that incorporate both elastic and viscous characteristics are
referred to as being viscoelastic. The elastic component for such materials,
in a one-dimensional form, can be represented by a linear elastic spring with
a spring constant Ks (Figure 2.10a) and is given by the relationship

σ = Ksε (2.47)

The viscous behavior can be represented by a dashpot as shown in

Figure 2.10b with the relationship between the stress, σ, and the rate of
strain, ε = dε , being given by
dt

σ = µε (2.48)

where µ is the viscosity coefficient. The combined viscoelastic behavior of a

solid is represented by the spring and dashpot in series (the Maxwell model)
7328_C002.fm Page 65 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 65

or in parallel (the Kelvin or Voigt model). In the case of the Maxwell model
(Figure 2.10c), the constitutive relationship is given by

σ σ
+ = ε (2.49)
Ks µ

In the case of the Kelvin model (Figure 2.10d), the constitutive relationship
can be expressed as

σ = K s ε + µε (2.50)

These two-element models are not generally adequate to represent the vis-
coelastic behavior of real materials and, thus, models with three or more
elements are employed that can incorporate the flexibility needed to describe
the response of actual materials. Creep and stress relaxation experiments are
generally performed to describe the viscoelastic properties of materials. In
the case of creep tests (Figure 2.11), the viscoelastic specimen is subjected to an
instantaneous constant stress σ0 and the strain (creep response) is measured

⑀∝
Strain (ε)

⑀0

t=0 t=∝
Time

σ0
Stress (σ)

Time

FIGURE 2.11
Creep test for a viscoelastic material.
7328_C002.fm Page 66 Friday, October 6, 2006 12:23 PM

66 Biofluid Mechanics: The Human Circulation

σ0

Stress (σ)
σ∝

t=0 t=∝
Time

⑀0
Strain (ε)

Time

FIGURE 2.12
Stress relaxation experiment for a viscoelastic material.

as a function of time. For the stress relaxation experiment (Figure 2.12), the
specimen is subjected to an instantaneous constant strain and the stress (relax-
ation) is measured as a function of time. Thus, in viscoelastic material (includ-
ing blood vessels), a finite time is required after the application of the load
before the complete deformation is attained with a static load. With such a
material, the radius of the tube for a step increase in pressure will take a finite
time to increase to its steady state value (Figure 2.13). The same phenomenon
Pressure

Static loading Dynamic loading

Φ
Radius

Time

FIGURE 2.13
Response to time-dependent pressure loading of a viscoelastic structure.
7328_C002.fm Page 67 Friday, October 6, 2006 12:23 PM

Introduction to Solid Mechanics 67

will also occur when the pressure is removed instantly. To determine the elastic
modulus under static conditions of applied internal pressure, adequate time
should be allowed for the vessel to expand to its final radius before the
measurement is made. With such experiments, only the elastic response of
the material is determined and no information is obtained about the viscous
properties.
To determine the viscoelastic behavior of the material, dynamic experiments
are performed with the application of sinusoidally varying pressure. A typ-
ical response to such loading is shown in Figure 2.13. Even though the
oscillations of the radius have the same frequency as that of the applied
pressure, the effect of viscosity is to delay the response by a phase angle, φ.
Moreover, it is also observed that the value of peak distension under sinu-
soidal loading is smaller than that for static deflection. This is due to the fact
that before the delayed response can attain its peak, the load has already
reached its down slope and, thus, reversing the process so that the amplitude
of the distension is smaller.
Due to the phase difference between the applied pressure and the radial
changes, the ratio of stress to resultant strain will be a complex quantity,
which can be expressed by the relationship

Ec = Ec e iφ = Ec cos φ + i Ec sin φ (2.51)

The modulus, Ec, is a function of the applied oscillatory frequency and its
values can be determined by applying sinusoidal loading at varying fre-
quencies. For the case of dynamic loading in viscoelastic material, the incre-
mental elastic modulus expression given in Equation 2.45 can be modified as

2(1 − v 2 )R12 R2 ∆ p iφ
Einc = e (2.52)
(R22 − R12 )∆R2

As pointed out earlier, the measured strains obtained in dynamic testing

of viscoelastic materials will be smaller than those measured under static
conditions and, hence, the dynamic elastic modulus will be higher than the
corresponding static value.

2.5 Problems
2.1 A circular aluminum tube 40 cm in length is subject to a tensile load
of 2 kN. The outside and inside diameters of the tube are 4.2 and 4.0 cm,
respectively. What is the amount of tensile stress on the bar? Aluminum has
an elastic modulus of 73.1 GPa. Determine the axial strain and the increase
7328_C002.fm Page 68 Friday, October 6, 2006 12:23 PM

68 Biofluid Mechanics: The Human Circulation

in length of the bar for the given load. Assuming a typical elastic modulus
of about 0.1 MPa for a typical artery, what would be the corresponding tensile
load on an arterial specimen of the same dimension that would result in the
same axial strain?
2.2 A high-strength steel rod (E = 200 GPa and ν = 0.32) with a diameter
of 5 cm is being subjected to a compressive load of 10 kN. Determine the
increase in diameter of the tube after the load is applied.
2.3 A brass specimen 10 mm in diameter and a length of 50 mm is loaded
with a 20 kN force in tension. If the length increases by 0.12 mm, determine
the elastic modulus of the brass. If the diameter of the bar decreases by 0.0083
mm, calculate the Poisson’s ratio of the material.
2.4 An aluminum soda can (E = 73.1 GPa and ν = 0.35.) has a radius-to-
thickness ratio of 200:1 and holds soda under pressure. When the lid is
opened to release the pressure, the strain in the longitudinal direction is
measured as 170 µm/m. What was the internal pressure in the can? Express
your answer in the units of mm Hg. Compare this pressure magnitude with
the typical mean blood pressure in an artery.
7328_C003.fm Page 69 Monday, September 18, 2006 10:57 AM

3
Cardiovascular Physiology

CONTENTS
3.1 Introduction ..................................................................................................69
3.2 The Heart ......................................................................................................70
3.2.1 Overview ..........................................................................................70
3.2.2 Cardiac Structure.............................................................................73
3.2.3 Cardiac Conduction ........................................................................74
3.2.4 Cardiac Function .............................................................................77
3.3 Cardiac Valves..............................................................................................82
3.4 Systemic Circulation....................................................................................84
3.5 Coronary Circulation ..................................................................................91
3.6 Pulmonary Circulation and Gas Exchange in the Lungs .....................94
3.7 Cerebral and Renal Circulations ...............................................................98
3.7.1 Cerebral Circulation........................................................................98
3.7.2 Renal Circulation.............................................................................99
3.8 Microcirculation ...........................................................................................99
3.9 Regulation of the Circulation ..................................................................103
3.10 Atherosclerosis ...........................................................................................105
3.10.1 Morphology of Atherosclerosis...................................................106
3.10.2 Sequence of Growth of the Lesion .............................................107
3.10.3 Physiological Implications ...........................................................107
3.11 Problems...................................................................................................... 110

3.1 Introduction
This chapter contains a brief review of the cardiovascular physiology
relevant to fluid mechanics in the human circulation. For a detailed study,
the reader is referred to any textbook on Physiology (e.g., Guyton and Hall,
2000 or Silverthorn, 2001). The cardiovascular system includes the heart and
blood vessels of the systemic and pulmonary circulation. Fundamental
requirements of the circulatory system are to provide adequate blood flow
without interruption and to regulate blood flow according to the various

69
7328_C003.fm Page 70 Monday, September 18, 2006 10:57 AM

70 Biofluid Mechanics: The Human Circulation

demands of the body (Rushmer, 1976). The contracting heart supplies the
energy required to maintain the blood flow through the vessels. The pressure
gradient developed between the arterial and the venous end of the circula-
tion is the driving force causing blood flow through the blood vessels. The
energy is dissipated in the form of heat due to the frictional resistance. Blood
picks up oxygen in the lungs and nutrients in the intestine and delivers them
to the cells in all parts of the body. The circulating blood also removes cellular
wastes and carbon dioxide from the cells for excretion through the kidneys
and the lung, respectively. The circulating blood also maintains the visceral
organs, such as the heart, the kidney, the liver, and the brain, at a constant
temperature by convecting the heat generated and dissipating the same
through transfer across the skin. In the following brief review of the cardio-
vascular system, we will concentrate on the mechanical aspects of circulation.

3.2 The Heart

3.2.1 Overview
The human heart consists of four chambers whose function is to pump
blood throughout the body. The four chambers of the heart are schematically
illustrated in Figure 3.1. The arrows indicate the direction of blood flow
through the heart, which is ensured by the presence of heart valves. The
heart consists of two pumps in series circulating blood through the pulmo-
nary and systemic circulations, respectively. The right ventricle, a low-
pressure pump, supplies the pulmonary circulation, whereas the left ventricle,
a high-pressure pump, supplies the systemic circulation. The fundamental
skeleton of the heart is formed by the four rings of interconnected dense
tissue, which secure the valves (Figure 3.2). The atrial chambers are attached
to the superior surface of the annulus of the atrio-ventricular valves, while
the trunks of the aorta and the pulmonary artery are attached to the semi-
lunar valves. The right and left ventricles are attached below and around
the circumference of the fibrous skeleton and the interventricular septum is
attached to the line of fusion between the mitral and tricuspid valve rings
(Rushmer, 1976).
The right atrium receives venous blood from the superior and inferior
vena cavae (SVC and IVC, respectively). During ventricular relaxation, blood
flows through the tricuspid valve into the right ventricle. During ventricular
contraction, blood is pumped through the pulmonary arteries into the cap-
illaries of the lung where carbon dioxide is excreted and oxygen absorbed
from the air in the alveolar sac. The oxygenated blood returns to the heart
via the pulmonary veins and is collected by the left atrium. Blood then fills
the left ventricle through the open mitral (bicuspid) valve during ventricular
relaxation. Upon subsequent ventricular contraction, blood is pumped
through the aortic valve into the systemic circulation via the aorta.
7328_C003.fm Page 71 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 71

Superior Aorta
vena cava
Pulmonary
semilunar valve
Right pulmonary Left pulmonary
arteries arteries

Left pulmonary
veins
Right
atrium

Cusp of left AV
(bicuspid) valve

Chordae
Cusp of right AV tendineae
(bicuspid) valve
Papillary
muscles
Inter ventricular
Inferior vena septum
cava Left ventricle

Descending
Right ventricle aorta

One-way flow through the heart is ensured

by two sets of valves.

FIGURE 3.1
A schematic diagram of the four chambers of the heart and the heart valves. The arrows indicate
the direction of blood flow. (Redrawn from Silverthorn, D.U. (2001) Human Physiology: An
Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)

The heart muscle, or myocardium, consists of three major types — atrial,

ventricular, and specialized excitatory or conducting muscle. The atrial and
ventricular muscles contract due to electrical excitation causing an increase
in the blood pressure that induces pumping of the blood. The excitatory
and conducting fibers conduct the electrical activity to all parts of the heart.
These specialized conducting fibers contain few contractile fibrils and, thus,
contract very feebly. The cardiac (atrial and ventricular) muscle consists of
myofibrils containing actin and myosin filaments. These actin and myosin
filaments slide across each other during the excitation–contraction function
of the cardiac muscles producing contraction. Individual cardiac muscle
cells are separated by intercalated discs giving the cardiac muscle a striated
appearance (Figure 3.3). Even though these intercalated discs separate
7328_C003.fm Page 72 Monday, September 18, 2006 10:57 AM

72 Biofluid Mechanics: The Human Circulation

Pulmonary
Left artery
atrium Aorta
Right
atrium

P
A
M
T
Fibrous
skeleton

Right Left
ventricle ventricle

FIGURE 3.2
The fibrous skeleton and the four chambers of the heart. (Redrawn from Rushmer, R.F. (1976)
Cardiovascular Dynamics, W.B. Saunders Company, Philadelphia. With permission.)

Intercalated disks
Myocardial muscle cell

Myocardial muscle cells are branched, have a

single nucleus, and are attached to each other by
specialized junctions known as intercalated disks.

FIGURE 3.3
The striated appearance of the cardiac muscle. (Redrawn from Silverthorn, D.U. (2001) Human
Physiology: An Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)
7328_C003.fm Page 73 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 73

Right Left
ventricle ventricle

FIGURE 3.4
A schematic drawing depicting the left and right ventricles. The left ventricle has a more
rounded shape, whereas the right ventricle has a semilunar shape and wraps around the left
ventricle.

myocardial cells anatomically, the electrical resistance through the discs is

low. Thus, the cardiac muscle cells can functionally be considered a con-
tinuum and referred to as a functional syncytium.

3.2.2 Cardiac Structure

Figure 3.4 schematically illustrates the geometry of the left and right ventri-
cles. The left ventricle has a more circular cross section, whereas the right
ventricle assumes a semilunar shape that wraps around part of the left
ventricle. The anatomical features of the ventricles are adapted to the type
of work done by the respective chambers. For example, the left ventricle is
the high-pressure pump and has a cavity surface area, which is small relative
to the blood volume due to its cylindrical shape. The contraction of the left
ventricle involves both a reduction in the diameter of the cylindrical portion
and a shortening along the longitudinal direction. The contraction of the
circumferential fibers in the relatively thick-walled left ventricle induces a
very high internal pressure. This high pressure supplies energy for flow of
blood through the high-resistance systemic circulation. On the other hand,
the right ventricle has a roughly triangular shape with a convex septal wall
and a concave free wall, yielding a crescent-shaped ventricular cavity. The
surface area of the cavity wall is relatively high in comparison to the volume
of blood within the chamber. Blood pumped from the right ventricle is
affected by (1) shortening of the longitudinal axis of the chamber, (2) the free
wall moving towards the interventricular septum, and (3) contraction of the
circumferential fibers of the right ventricle, resulting in a greater curvature
of the septal wall. Thus, pumping of blood from the right ventricle resembles
the action of a bellows, where a large volume of fluid is displaced at a
relatively low pressure. This pumping action is suitable for the pulmonary
7328_C003.fm Page 74 Monday, September 18, 2006 10:57 AM

74 Biofluid Mechanics: The Human Circulation

circulation where the resistance to flow is relatively small and high pressures
are not needed. However, if the pulmonary resistance increases due to dis-
ease, the right ventricular myocardium cannot produce the sustained high
pressures needed and, hence, failure of the right ventricle (clinically known
as congestive heart failure or CHF) will result. Similarly, if a large pumping
volume is demanded of the left ventricle, the left ventricular chamber dilates
to accommodate the increased volume.

3.2.3 Cardiac Conduction

Figure 3.5 illustrates the anatomical details of the cardiac conduction system.
The sino-atrial (S-A) node is a specialized group of autorythmic cells located
in the posterior wall of the right atrium in the vicinity of the superior vena
cava. The S-A node has intrinsic rhythmicity that produces approximately
70 to 80 action potentials per minute in the normal adult human heart. These
action potentials spread through the atrial muscle at a speed of about 0.3 m/s.
As the cardiac impulse spreads through the atria, the muscles contract and
pump the blood through the atrio-ventricular valves into the ventricles. A few
specialized conducting fibers in the atrial muscle conduct the impulse rapidly
to the atrio-ventricular (A-V) node and the impulse reaches the A-V node
approximately 40 ms after the start of the impulse in the S-A node. The A-V
node complex consists of junctional fibers, the node, transitional fibers, and the
A-V bundle. The atrial muscles are electrically separated from the ventricular
muscles with the exception of the A-V bundle, which is the only pathway
through which the electrical activity is conducted from the atrial to the ventric-
ular chambers. A delay of approximately 110 ms occurs as the impulse is
conducted through this complex. This electromechanical delay allows sufficient
time for blood to be pumped into the ventricles prior to ventricular contraction.
The Purkinje fibers are specialized conducting fibers in the ventricles. The
Bundle of His starts from the A-V complex and is large in diameter. It conducts
the electrical impulses into the ventricular muscles at a velocity of 1.5 to 2.5
m/s. The A-V bundle divides into left and right bundle branches below the
valves at the interventricular septum and spreads toward the apex of the respec-
tive ventricle before turning toward the base of the heart. Thus, the ventricular
muscle near the apex is activated first, followed by the basal muscle. The
membrane action potentials for the autorhythmic and contractile cells are
shown in Figure 3.6a. The details of the action potentials of the Purkinje fibers
and ventricular muscle are shown in Figure 3.6b. The resting membrane poten-
tial of the S-A node is about – 55 to – 60 mv, while the potential for the Purkinje
fiber is about – 90 to – 100 mv, and for the ventricular muscles it is about – 80
to – 85 mv. Upon electrical excitation, however, the cells are depolarized to
about + 20 mv inside with respect to the outside environment.
The phases of a ventricular muscle action potential include a sharp upstroke
from the resting membrane potential (0), a brief overshoot (1), a plateau region
(2), rapid recovery or repolarization (3), and electrical diastole (4), as depicted
7328_C003.fm Page 75 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 75

SA node depolarizes.
SA
node

AV
node
Electrical activity
goes rapidly to
AV node via
internodal
(b) pathways.

(c)
SA node
Internodal Depolarization
pathways spreads more
slowly across atria.
Conduction
slows through
AV node.
AV node

Bundle of (d)
his Bundle Depolarization moves
branches Purkinje rapidly through
fibers ventricular conducting
(a) system to the apex
of the heart.

(e)
Depolarization wave
spreads upward from
the apex.
(f )

FIGURE 3.5
The electrical conduction of the heart. (Redrawn from Silverthorn, D.U. (2001) Human Physiology:
An Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)

in Figure 3.6b. In the action potential of the S-A node, each time the resting
membrane potential is reestablished, the potential gradually decays until it
reaches the threshold for self-excitation. At this point, depolarization takes
place and the action potential is established. The action potential of the S-A
node repolarizes within about 15 ms after depolarization, whereas 15 to 30 ms
is required for the other cardiac muscles. During this plateau region, or refrac-
tory period, only extremely high electrical stimulation can initiate a new spike
and the normal cardiac impulses cannot initiate a new spike.
7328_C003.fm Page 76 Monday, September 18, 2006 10:57 AM

76 Biofluid Mechanics: The Human Circulation

Membrane potential
of autorhythmic cell

Electrical Membrane
current potential of
contractile cell

Contractile cells
Cells of
SA node

Intercalated disk
with gap junctions
(a)

Plateau
+20
0 (b)
–20
–40
–60
–80
Millivolts

–100 (1) Purkinje fiber

+20 (2) Plateau
0
(c)
–20 (3)
–40 (O)
–60
(4)
–80
–100 Ventricular muscle

0 1 2 3 4
Seconds
(b)

FIGURE 3.6
(A) Electrical conduction in myocardial cells resulting in depolarization. (Redrawn from
Silverthorn, D.U. (2001) Human Physiology: An Integrated Approach, 2nd ed., Prentice Hall, Upper
Saddle River, NJ. With permission.) (B) The rhythmic action potentials of the Purkinje fiber and
the ventricular muscle. (Redrawn from Guyton, A.C. and Hall, J.E. (2000) A Textbook of Medical
Physiology, 10th ed., W.B. Saunders Company, Philadelphia. With permission.)

The intrinsic discharge or depolarization rate of the S-A node is 70 to 80

times a minute. For the Purkinje fibers, it is 40 to 60 times a minute, and for
the ventricular muscles, it is 15 to 40 times a minute. In the absence of excitation
from the S-A node, the A-V node or the Purkinje fibers can provide rhythmic
excitation of the heart. Since the S-A node excitation is relatively fast, its action
potential normally controls the rhythmicity of the heart. The period between
the beginning of the rhythmic excitation of the S-A node to the next excitation
is defined as a cardiac cycle. The contractile motion of the ventricular muscle
7328_C003.fm Page 77 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 77

1 sec.
+2

+1 R

Millivolts
P T P
0
Q
S
–1

FIGURE 3.7
A typical normal electrocardiogram.

also follows the order in which the action potential reaches the particular
region. Hence, the interventricular septum starts contracting first followed by
the endocardial surface of the apical region, and the contraction spreads
toward the base (Figure 3.5). From a mechanical point of view, this sequence
of contraction enables an efficient ejection of blood from the ventricles to the
respective blood vessels through an effective “squeezing”of the heart.
Electrical potentials generated by the heart also spread to the surrounding
tissues and this electrical activity can be recorded by placing electrodes on
the surface of the skin. Such a recording is known as the electrocardiogram
(ECG) and a typical ECG signal is shown in Figure 3.7. The P wave and the
QRS complex shown in the figure are depolarization waves. The electrical
currents cause the P wave as the atrial muscles depolarize followed by the
contraction of the atrial muscles. The QRS complex represents the depolar-
ization of the ventricular muscles, which is followed by the ventricular
muscular contraction. The T wave is generated as the ventricular muscles
repolarize and, hence, it is known as the repolarization wave. In Figure 3.7,
the time between consecutive R waves is 0.8 sec or, in other words, the heart
rate is 75 cycles or beats per minute (bpm). For a normal adult, the heart
rate is between 70 to 75 bpm under resting conditions.

3.2.4 Cardiac Function

The cardiac cycle has two phases — diastole and systole. Diastole is that
portion of the cardiac cycle in which the cardiac muscles are relaxing, while
systole is the time during which the cardiac muscles are contracting
(Figure 3.8). At the beginning of diastole, both the atrial and ventricular
muscles are relaxing and blood from the vena cavae and the pulmonary veins
is filling into the right and left atrium, respectively. The A-V (i.e., the mitral
and tricuspid) valves between the atria and the ventricles open and blood
flows into the ventricles, increasing the ventricular volume. Atrial systole
 Start
78

(a) Late diastole—both sets of

chambers relaxed. Passive
ventricular filling.

(e) Isovolumic ventricular (b) Atrial systole—atrial

relaxation—as ventricles relax contraction forces
pressure in ventricles drops, a small amount of
blood flows back into cups additional blood into
of semilunar valves and ventricles.
snaps them closed.
iastole
ar d
c ul At
tr i

n
r ia

Ve
l sy
stol
e
EDV = End-diastolic volume.
7328_C003.fm Page 78 Monday, September 18, 2006 10:57 AM

The maximum amount of blood

ESV = End-systolic volume, Atrial diastole le in ventricles occurs at the end of
or minimum amount of blood Ventricular systo ventricular relaxation. EDV ≅ 135 ml.
in ventricles. ESV ≅ 65 ml.

(d) Ventricular ejection—

as ventricular pressure (c) Isovolumic ventricular
rises and exceeds pressure contraction—first phase
in the arteries, the semilunar of ventricular contraction pushes
valves open and blood is ejected. AV valves closed but does not
create enough pressure
to open semilunar valves.

FIGURE 3.8
Schematic of the cardiac cycle with the atrial and ventricular systolic and diastolic phases. (Redrawn from Silverthorn, D.U. (2001) Human Physiology: An
Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)
Biofluid Mechanics: The Human Circulation
7328_C003.fm Page 79 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 79

follows with the depolarization of the S -A node, resulting in contraction of

the atrial muscles, which pumps blood into the ventricles. After the depolar-
ization wave moves through the A-V node and the Purkinje fibers in the
ventricles, the ventricular muscles start contracting from the apex towards
the base of the ventricle. Ventricular pressure increases with the ventricular
contraction and the A-V valves close to prevent blood from flowing back to
the atria. Ventricular muscles continue to contract while both the A-V and
the semilunar (i.e., the aortic and pulmonary) valves remain closed. Ventric-
ular chamber pressures rise rapidly during the isovolumic contraction and,
once they exceed those of the pulmonary or aortic arteries, the semilunar
valves open and blood is forced out into the pulmonary and systemic circu-
lations. At the end of systole, the ventricular muscles begin to relax and the
chamber pressures fall rapidly. As soon as the arterial pressures exceed that
of the ventricular chambers, the semilunar valves close and the ventricles
continue to relax isovolumically. When the ventricular pressures fall below
those of the atria, the A-V valves open and the next cardiac cycle begins.
The relationship between the ventricular chamber pressure and volume
can be represented by the pressure–volume curve shown in Figure 3.9 for

EDV = End-diastolic volume

ESV = End-systolic volume

Stroke volume
120
D
Left ventricular pressure (mm Hg)

Aortic valve
Aortic valve Ventricular ejection opens
closes;
ESV
80 C
One Isovolumic
Isovolumic
cardiac contraction
relaxation
cycle

40
Mitral valve
Mitral valve closes; EDV
opens Ventricular diastole + filling
A B
0 65 100 135
Left ventricular volume (ml)
A B: Passive filling and atrial contraction
B C: Isovolumic contraction
C D: Ejection of blood into aorta
D A: Isovolumic relaxation

FIGURE 3.9
A ventricular pressure volume curve. (Redrawn from Silverthorn, D.U. (2001) Human Physiology:
An Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)
7328_C003.fm Page 80 Monday, September 18, 2006 10:57 AM

80 Biofluid Mechanics: The Human Circulation

the left ventricle. Point A in the figure depicts the point when the ventricular
chamber pressure and volume are the minimum. The mitral valve will open
once the atrial pressure exceeds that of the ventricle and the AB portion of
the curve represents diastolic ventricular filling as the ventricular muscles
continue to relax. Point B represents the end diastolic volume (EDV) and is
the maximum ventricular volume during a cardiac cycle. With the beginning
of ventricular contraction, the chamber pressure rises resulting in the closure
of the mitral valve and the isovolumic contraction phase shown as the BC
portion of the curve. At point C, the ventricular pressure exceeds that of the
aorta and the aortic valve opens. The ventricular muscles continue to contract
increasing the chamber pressure while ejecting the blood into the aorta;
hence, the ventricular volume decreases. The amount of blood left in the
ventricle at the end of contraction (point D) is the end systolic volume (ESV)
and represents the minimum ventricular volume in a cardiac cycle. The
difference between the EDV and the ESV is called the stroke volume (SV)

SV (ml/beat ) = EDV − ESV (3.1)

and represents the volume of blood pumped out by the left ventricle into
the systemic circulation in a cardiac cycle (ml/beat). At the end of contrac-
tion, the ventricular muscles start to relax, the ventricular pressure falls
below that of the aorta and the aortic valve closes. The ventricular muscles
continue through the isovolumic relaxation (segment “DA” of the curve)
and the cardiac cycle starts again. The amount of blood that is pumped out
of the ventricles in 1 minute is referred to as the cardiac output (CO) and is
the product of the stroke volume and the heart rate (HR):

CO(ml/min) = SV (ml/beat )xHR( beats/min ) (3.2)

A similar pressure–volume curve can also be constructed for the right

ventricle with the pressures generated in the right ventricles being smaller.
From elementary thermodynamics, we know that the total work done in a
heat cycle is the area under the pressure–volume curve. This follows from
the fact that mechanical work is the product of force and distance and, since
force is the product of pressure and area, the work done can be written as
(pressure × area) × distance, or pressure × volume. Thus, the mechanical
work done by the ventricles can be computed from the pressure–volume
curve for a cardiac cycle. The AB portion of the curve in Figure 3.9 represents
the filling phase of the cardiac cycle where work is being done by the blood
in the left ventricle to increase the ventricular volume. Segment BC repre-
sents the isovolumic contraction phase when no work is done and the
energy is stored as elastic energy in the muscles. CD denotes the ejection
phase and work is done by the cardiac muscles on blood, and DA represents
the isovolumic relaxation phase where no work is done. The area within the
pressure–volume curve represents the net work done by the ventricle on blood.
7328_C003.fm Page 81 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 81

The heart muscles receive the energy to perform this work from the oxygen
in the blood. Computing the input and output energies, efficiency of the
heart can be computed. Typically, the work done by the heart amounts to
only 10 to 15% of the total input energy and the remainder of the energy is
dissipated as heat.
A combined representation of the electrical activity of the heart (ECG), the
pressure pulses in the cardiac chambers and the aorta, and the ventricular
blood volume is illustrated in the Wiggers diagram (Figure 3.10). Between the

Time (m sec)
0 100 200 300 400 500 600 700 800

QRS QRS
Electro- complex Cardiac cycle complex
cardiogram
(ECG) P T P

120 Aortic valve

Aortic valve closes
opens
Pressure (mm Hg)

90 Aorta
Dicrotic notch
Left
ventricular
pressure
60

Mitral valve
Left atrial closes Mitral valve
30 pressure
open

Heart S1 S2
sounds
135 End diastolic
Left volume
ventricular
volume
(ml)
65 End systolic volume
Atrial Ventricular systole Ventricular diastole Atrial
systole systole

Atrial Isovolumic Ventricular Early Late Atrial

systole systole ventricular ventricular systole

FIGURE 3.10
The Wiggers diagram depicting the relationship with the electrocardiogram, the ventricular
and the atrial pressures, the heart sounds, and the volume changes in the ventricular chamber.
(Redrawn from Silverthorn, D.U. (2001) Human Physiology: An Integrated Approach, 2nd ed.,
Prentice Hall, Upper Saddle River, NJ. With permission.)
7328_C003.fm Page 82 Monday, September 18, 2006 10:57 AM

82 Biofluid Mechanics: The Human Circulation

pressure and volume curves in the Wiggers diagram, the heart sounds gener-
ated during the cardiac cycle are also illustrated. Ventricular systole begins at
the apex of the heart and the blood is squeezed toward the base of the heart
during the ventricular contraction. The increase in pressure in the ventricular
chambers and the blood pushing upward on the ventricular side of the atrio-
ventricular (mitral and tricuspid) valves keeps the leaflets closed so that blood
does not flow back into the atria. Vibration induced following the closure of
the valves generates the first heart sound (S1 in Figure 3.10) that can be heard
through a stethoscope. Similarly, at the end of contraction, the ventricles relax
with a rapid pressure reduction in the chambers. Once the ventricular pressure
falls below that of the arteries, blood starts to flow back toward the ventricles
and the semilunar (i.e., aortic and pulmonic) valves close. The second heart
sound (S2) is generated due to the induced vibrations with the closure of the
semilunar valves. Third and fourth heart sounds (not illustrated in the figure)
can also be recorded with sensitive electronic stethoscopes. The third heart
sound is generated by the turbulent or agitated flow of blood into the ventricles
during the filling phase. The fourth heart sound is associated with turbulence
created by the atrial contraction. Diseased states, such as stenosed (i.e., stiffer)
or incompetent (i.e., leaking) valve leaflets, will result in changes in character-
istics of the heart sounds, such as frequency and amplitude. Hence, heart
sounds can also be used in diagnosis of valvular diseases.

3.3 Cardiac Valves

As is evident from the above description, the function of the heart valves is
to prevent back flow of blood from the ventricles into the atria or from the
great arteries (i.e., the aorta and pulmonary) into the ventricles. The valves
open and close passively. They open and close relatively rapidly and, when
closed, completely seal their respective orifices. The thin leaflets of the valves
withstand very high repetitive loads for billions of cycles during the human
lifetime. Cross-sectional and frontal views of the heart with the four valves
are shown in Figure 3.11. The atrio-ventricular (A-V) valves consist of
the tricuspid valve between the right atrium and the right ventricle and the
bicuspid (mitral valve) between the left atrium and the left ventricle. The
pulmonary valve between the right ventricle and the pulmonary artery, and
the aortic valve between the left ventricle and the aorta have three symmet-
rical valve cusps in the shape of a half moon and, hence, are called the
semilunar valves. The systolic phase of the cardiac cycle with the A-V valves
closed and the semilunar valves open is shown in the top panels of
Figure 3.11 and the diastolic phase is illustrated in the bottom panel.
Behind the aortic valve cusps, there are three aortic sinuses called the
sinuses of Valsalva. Two of these sinuses have openings for the origin of
the coronary arteries where the presence of sinuses prevent the obstruction
 (a) Transverse section (b) Frontal section
Left atrium
Tricuspid Mitral (left AV),
Fibrous Aorta
(right AV) valve skeleton or bicuspid,
(closed) valve (closed) Mitral valve
Aortic (closed)
sinus

Aortic semilunar
Chordae tendinea
valve (open)
(tense)

Papillary
Cardiovascular Physiology

Pulmonary semilunar
muscles (tense)
valve (open)
Left ventricle
(contracted)
7328_C003.fm Page 83 Monday, September 18, 2006 10:57 AM

(d) Frontal section

Pulmonary veins
(c) Transverse section Mitral (left AV), or bicuspid,
valve (open)
Mitral valve
(open)
Chordae tendineae
Aortic semilunar (relaxed)
valve (closed)
Papillary muscles
(relaxed)
Tricuspid valve Pulmonary semilunar
(open) valve (closed) Left ventricle
(dilated)

FIGURE 3.11
An illustration of the four heart valves during the systolic and diastolic phases of a cardiac cycle. The top panel depicts the ventricular contractile phase
with the atrio-ventricular valves in the closed position and the semilunar valves in the open position. The bottom panel represents the ventricular filling
phase with the atrio-ventricular valves in the open position and the semilunar valves in the closed position. (Redrawn from Silverthorn, D.U. (2001) Human
83

Physiology: An Integrated Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)
7328_C003.fm Page 84 Monday, September 18, 2006 10:57 AM

84 Biofluid Mechanics: The Human Circulation

of the coronary arteries by the valve cusps in the fully open position. During
systole, the three cusps open to the full dimensions of the valve ring resulting
in axial flow of blood into the aorta. The A-V valves have two large opposing
cusps and small intermediary cusps at each end. For the mitral valve,
chordea tendinae originating from the edge of the two larger leaflets are
connected to two papillary muscle groups. The chordea tendinae from the
tricuspid valves are attached to three groups of papillary muscles. The mitral
valve consists of a large anteromedial cusp that hangs like a curtain at the
fully open position and a shorter posterolateral cusp. The combined surface
area of the two leaflets is almost twice the valve ring area. In the open
position, the upper portion of the mitral valve resembles that of a funnel.
During contraction of the ventricle, the papillary muscles also contract and
pull the valve leaflets toward the ventricle for efficient closure of the valves
at the beginning of systole. The anatomy and function of the tricuspid valve
is similar to that of the mitral valve. The detailed analysis of fluid mechanics
past the heart valves is included in Chapter 7.

3.4 Systemic Circulation

A schematic of the human circulatory system is shown in Figure 3.12. The
flow of blood from the left ventricle into the aorta to the peripheral regions
of the body and back to the right atrium is defined as the systemic circulation.
The arteries and the arterioles carry the blood to the capillaries in the tissues
and the blood returns to the right atrium through the venules and the veins.
Blood flow from the right ventricle into the lungs and back to the left atrium
is defined as the pulmonary circulation. The pulmonary artery distributes
blood from the right ventricle into the lungs and the blood returns to the
left atrium via the pulmonary vein. Blood is pumped through the pulmonary
and systemic circulations at a rate of about 5.2 liters per minute (l/m) under
normal conditions. The driving force for propelling blood through any seg-
ment of the circulatory system is the pressure gradient across the segment.
Even though the blood flow through the arterial segments is highly pulsatile
and, thus, not amenable to the steady flow Hagen–Poiseuille model
(Chapter 1, section 1.8: Fluid Mechanics in a Straight Tube), a basic relation-
ship between the pressure difference and the flow rate can be determined if
we consider the time averaged values. If ∆p is the time averaged pressure
difference across an arterial segment (i.e., difference between mean arterial
pressure at the aortic root and the venous pressure in the right atrium, in
the case of systemic circulation) and Q is the time averaged flow rate (i.e.,
cardiac output), then from basic fluid mechanics, we have the relationship

∆p
R= (3.3)
Q
7328_C003.fm Page 85 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 85

Veins Capillaries Arteries

Head and brain

Arms

Pulmonary veins
Superior Ascending
vena cava arteries
Pulmonary
arteries Lungs
Right atrium
Left Aorta
atrium

Inferior Left ventricle Abdominal

Venae cavae
vena cava aorta
Right ventricle
Coronary arteries
Heart
Trunk

Hepatic
vein Hepatic artery

Liver Hepatic portal vein Digestive tract

Renal veins Renal arteries

Descending
arteries
Valve Kidneys

Pelvis and legs

FIGURE 3.12
Schematic of blood flow to the visceral organs of systemic and pulmonary circulation. (Redrawn
from Silverthorn, D.U. (2001) Human Physiology: An Integrated Approach, 2nd ed., Prentice Hall,
Upper Saddle River, NJ. With permission.)

where R is the resistance to the flow of blood in that segment. Under these
conditions, the resistance to blood flow is predominantly due to the viscous
shear stresses. The resistance computed using units of pressure in mm Hg
(millimeters of mercury) and flow rate in cm3/s in Equation 3.3 is referred to
as Peripheral Resistance Units (PRUs). For example, in the normal systemic
circulation, if the mean pressure difference between the systemic arteries and
the veins is about 100 mmHg and the flow rate through the systemic circulation
is about 100 cm3/s, the resistance is 1 PRU. In the pulmonary circulation, the
time-averaged pressure difference between the pulmonary artery and the left
atrium is normally about 10 mmHg. Since the pulmonary circulation will also
7328_C003.fm Page 86 Monday, September 18, 2006 10:57 AM

86 Biofluid Mechanics: The Human Circulation

maintain the same flow rate as the systemic circulation over a period of time,
the resistance in the pulmonary circulation is about 0.1 PRU. Clinically, the
resistance is computed in ‘‘Woods Units’’ where mmHg is used for pressure
and l/min is used for the flow rate.
Again, in Chapter 1, the relationship between the flow rate and the pressure
gradient was derived for steady flow of a viscous fluid through a rigid cylin-
drical pipe. The Hagen–Poiseuille flow relationship was given by the equation

πR 4 ∆p
Q= (1.72)
8µL

From Equation 3.3 and Equation 1.72, we get

∆p 8µL
R= = (3.4)
Q πR 4

From Equation 3.4, it can be observed that the resistance to blood flow is
inversely proportional to the fourth power of tube radius. Thus, small changes
in the radius of arteries and arterioles will significantly alter the vascular
resistance. This is important since one of the characteristics of blood vessel
walls is that they are very responsive to stimulation. With sympathetic stim-
ulation, the tone of the vessel walls increases, causing a decrease in diameter
and, thereby, a significant increase in resistance to the flow of blood. On the
other hand, with sympathetic inhibition, the tone of the vessels decreases
resulting in an increase in the vessel diameter and a corresponding decrease
in flow resistance. Thus, sympathetic stimulation or inhibition is effectively
used in controlling the flow rate of blood through any region of the body and
this control is affected at the arteriolar level. For example, an increase in vessel
diameter of just 5% would produce a reduction in resistance of over 21%.
The ratio of change in volume to a change in pressure gives a measure of
the distensibility of the blood vessel. In practice, the increase in volume
normalized to the initial volume is used to define the volumetric strain of
the vessel. The definition for vascular compliance, C, is given by

∆V ∆V
C= V
= (3.5)
∆p V ∆p

where ∆V is the change in volume and V is the initial volume. A pressure–

volume curve can be used to describe the relationship between the pressure and
volume for a vessel. Figure 3.13 shows such a relationship for the arterial and
venous system in man. The effect of sympathetic stimulation is to reduce the
distensibility of the vessel, whereas sympathetic inhibition has the opposite
effect. In the venous system, a large increase in blood volume results in a rela-
tively small increase in pressure when compared to the arterial system. In other
words, the venous system has a greater distensibility or capacitance. As a result,
the veins act as the primary volume reservoir for blood in the circulatory system.
7328_C003.fm Page 87 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 87

140

imulation
Pressure (mm Hg) 120

100 Normal volume

stem
Sympathetic st

ic Inhibition
80

Arterial sy
60

40 Sympathet n
atio
s t i mul
etic tem
20 path ous sys
Sym Ven hibition
Normal volume In
0
0 500 1000 1500 2000 2500 3000 3500
Volume (ml)

FIGURE 3.13
Pressure–volume curves for the arterial and venous system. (Redrawn from Guyton, A.C. and
Hall, J.E. (2000) A Textbook of Medical Physiology, W.B. Saunders Company, Philadelphia. With
permission.)

The aorta and other major arteries are high-pressure conduits transporting
blood to the various regions and visceral organs in the body. The normal
arterial wall is made up of three layers — the intima, media, and adventitia.
The intimal layer of the artery consists of a layer of endothelial cells attached
to a basement membrane. The medial layer consists of vascular smooth mus-
cle cells, collagen, and elastin fibers in an extracellular matrix. The adventitial
layer consists of connective tissue that is attached, or tethered, to the surround-
ing organs in the body. Since blood flows under high pressure through the
arteries, these vessels have strong walls. The arterial walls are also less com-
pliant compared to the venous vessels as shown in Figure 3.13. In the systemic
arteries, relatively small increases in arterial volume will result in large
increases in pressure. Therefore, the systemic arteries serve as a pressure res-
ervoir (Rushmer, 1976). The small arteries further branch into arterioles that
also have strong muscular walls where the vascular tone is controlled by
regulatory mechanisms. Thus, the blood flow distribution to various regions
of the body is controlled by changes in resistance offered by various arterioles.
The blood from the arterioles then enters into capillaries where the exchange
of nutrients between the blood and the interstitial space takes place. The
capillary walls are very thin and permeable to small molecular substances to
allow for an efficient exchange of nutrients. Blood from the capillaries is
collected into venules that coalesce into larger veins, which eventually return
it to the right atrium. Table 3.1 shows the total cross-sectional area, the time-
averaged velocity of blood, and the percentage of blood at any given time in
the various vessels in the systemic circulation (Milnor, 1989).
7328_C003.fm Page 88 Monday, September 18, 2006 10:57 AM

88 Biofluid Mechanics: The Human Circulation

TABLE 3.1
Cross-Sectional Area of the Various Blood Vessels in the Systemic and Pulmonary
Circulation along with the Percentage Volume of Blood for a 20 kg Dog
Total Percentage
Mean Mean Cross Total of Total
Diameter Number Length Section Volume Blood
Vessels (mm) of Vessels (mm) (cm2) (ml) Volume
Systemic Circulation
Aorta (19–45) 1 (2.0–16.0) 60
Arteries 4.000 40 150.0 5.0 75
Arteries 1.300 500 45.0 6.6 30
Arteries 0.450 6000 13.5 9.5 13 11%
Arteries 0.150 110,000 4.0 19.4 8
Arterioles 0.050 2.8 × 106 1.2 55.0 7
Capillaries 0.008 2.7 × 109 0.65 1357.0 88 5%
Venules 0.100 1.0 × 107 1.6 785.4 126
Veins 0.280 660,000 4.8 406.4 196
Veins 0.700 40,000 13.5 154.0 208
Veins 1.800 2.100 45.0 53.4 240 67%
Veins 4.500 110 150.0 17.5 263
Venae cavae (5–14) 2 (0.2–1.5) 92
Subtotal 1406

Pulmonary Circulation
Main Artery 1.600 1 28.0 2.0 6
Arteries 4.000 20 10.0 2.5 25 3%
Arteries 1.000 1550 14.0 12.2 17
Arterioles 0.100 1.5 × 108 0.7 120.0 8
Capillaries 0.008 2.7 × 109 0.5 1357.0 68 4%
Venules 0.110 2.0 × 106 0.7 190.0 13
Veins 1.100 1650 14.0 15.7 22
Veins 4.200 25 100.0 35 5%
Main veins 8.000 4 30.0 6
Subtotal 200
Heart
Atria 2 30
Ventricles 2 54 5%
Subtotal 84
Total 1690 100%
Source: Milnor, W.R. (1989) Hemodynamics, 2nd ed., Williams and Wilkins, Baltimore. With
permission.

At the root of the aorta, the systolic pressure of a normal adult is about
120 mmHg and the diastolic pressure is about 80 mmHg. The difference
between the systolic and diastolic pressures is referred to as the pulse pres-
sure. A typical pressure pulse in the aorta as a function of time is shown in
Figure 3.14. The pressure pulse in systole starts with a steep rise, which
gradually decreases to a plateau at peak pressure. This is followed by a sharp
incisura, or dichrotic notch, due to the closing of the aortic valve. During
diastole, there is essentially an exponential decay of the pressure pulse.
7328_C003.fm Page 89 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 89

Sharp
Slow rise incisura
to peak Exponential diastolic
decline
120 Sharp 16.0
upstroke

Pressure (k Pa)
Pressure (mm Hg)
13.3
80 10.6

6.65
40

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Seconds

FIGURE 3.14
Schematic of a normal pressure pulse in the aorta depicting the sharp pressure rise in the
beginning of systole and the exponential pressure decay during diastole. The incisura during
the pressure decay represents the closing of the aortic valve.

Since systole lasts for about 1/3 of the cardiac cycle followed by diastole for
about 2/3 of the cardiac cycle, the mean arterial pressure (MAP) over a
cardiac cycle can be computed as

( ps + 2 pd )
MAP = (3.6)
3

where ps and pd represent the systolic and diastolic pressure, respectively.

The mean aortic pressure for the waveform shown in Figure 3.14 is about
93 mmHg.
The magnitudes of blood pressures in the systemic and pulmonic circula-
tions are shown in Figure 3.15. At the root of the aorta, a pulse pressure of
40 mmHg can be observed. The pressure pulses travel along the arterial wall
at a speed much greater than the blood velocity in the vessels and, as the
pressure pulse travels downstream through the larger arteries, the pulse
pressure increases. The velocity of the pressure pulse in the aorta is ~ 3 to
∼ 5 m/s and increases to ~ 7 to ∼ 10 m/s in the large arteries and ~ 35 m/s
in small arteries. It will be shown later that the velocity of transmission of
the pressure pulses is directly proportional to the stiffness of the arterial wall.
The augmentation of the pressure pulse downstream is due to several factors:

1. The blood vessels in the distal portion are stiffer and, hence, the
velocity of transmission increases.
2. Curvature and branching sites in the blood vessels reflect the pulse
wave and the reflected wave interacts with the waves in the forward
direction.
7328_C003.fm Page 90 Monday, September 18, 2006 10:57 AM

90 Biofluid Mechanics: The Human Circulation

120

100

Pulmonary arteries
Pressure (mm Hg)

Pulmonary veins
Venae cavae
Large veins
Small veins
Capillaries

Capillaries
80

Arterioles
Venules

Venules
60
Large arteries

Small arteries
40

Arterioles
Aorta

20

0
Systemic Pulmonary

FIGURE 3.15
Pressure pulses in the systemic and pulmonary circulation. (Redrawn from Guyton, A.C. and
Hall, J.E. (2000) A Textbook of Medical Physiology, W.B. Saunders Company, Philadelphia. With
permission.)

3. The high pressure portion of the pressure pulse travels faster than
the other parts of the pulse and results in nonlinear interactions.

Blood traveling downstream has to overcome the resistance to flow in

the smaller vessels and, hence, the blood pressure decreases in the distal
arteries. As pointed out earlier, the major resistance to flow is provided by
the arteriolar segments and, thus, a sharp decrease in blood pressure is
observed at the origin of the capillaries where the pressure pulse is mini-
mal. This continues through the veins so that at the end of the vena cavae,
as the blood enters the right atrium, the blood pressure is close to 0 mmHg
(less than 5 mmHg). Since all the blood from the venous system empties
into the right atrium, measurement of the right atrial pressure is also
referred to as the central venous pressure (CVP). The CVP is modulated
by the ability of the heart to pump blood forward and by the amount of
blood that flows back from the peripheral circulation (called the Venous
Return). If the heart is not pumping efficiently, the CVP will tend to rise
and vice versa.
Venous valves (shown in Figure 3.12) and the ‘‘venous pump’’ also play
a major role in the transport of blood back to the heart. If a person is standing,
the pressure in the veins in the legs could be as high as 90 mmHg relative
to that of the right atrium due to hydrostatic forces. With muscle contractions
in the legs, however, the veins are compressed and the blood is ‘‘pumped’’
toward the heart. Venous valves prevent flow back to the legs, which would
normally occur due to gravity. A defect in the venous valves that allows
blood to leak back toward the legs will result in venous distention and stasis
that may progress to the development of varicose veins and deep vein
thrombosis (DVT).
7328_C003.fm Page 91 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 91

3.5 Coronary Circulation

Nutrients are supplied to the heart muscles by the coronary arteries. The
distribution of the main coronary arteries around the heart is illustrated in
Figure 3.16. The top panel shows the origin of the left and right coronary

Aortic valve
Right ventricle
Right coronary a.
Pulmonary valve
Left coronary a.
Orifices of
coronary a.s
Left ventricle
Tricuspid
Mitral valve valve
Coronary sinus
Atrioventricular (A-V) node
(a)

Aorta

Superior vena cava Left coronary a.

Left atrial appendage

Right coronary a.
Circumflex a.
Pulmonary a.
Right atrial
appendage Anterior descending
branch
Posterior descending (interventricular)
branch
(interventricular)

(b)

FIGURE 3.16
The anatomical details of the left and right coronary arteries: (a) top view showing the left and
right coronary ostia and the origin of the coronary arteries, and (b) main branches of the
coronary vessels and its perfusion field. (Redrawn from Jacob, S.W. and Francone, C.A. (1974)
Structure and Function in Man, 3rd ed., W.B. Saunders Company, Philadelphia. With permission.)
7328_C003.fm Page 92 Monday, September 18, 2006 10:57 AM

92 Biofluid Mechanics: The Human Circulation

arteries from the left and right coronary sinuses. The bottom panel illustrates
the principal branches of the left and right coronary arteries. The openings
for the coronary arteries, the coronary ostia, are situated in the right and left
sinuses of Valsalva. The left coronary artery originates from the left aortic
sinus and divides immediately into the left anterior descending and the left
circumflex branches. The left coronary artery supplies most of the left ven-
tricle, the anterior two-thirds of the interventricular septum, the anterior left
margin of the free wall of the right ventricle, the apex, the left atrium, and
the lower half of the interatrial septum. The right coronary artery, originating
from the right coronary sinus, supplies the anterior and posterior walls of
the right ventricle, the right atrium and the sinus node, the posterior one-
third of the interventricular septum, the atrio-ventricular node, the upper
half of the interatrial septum, and the posterior base of the left ventricle. In
approximately half of human beings, more blood flows through the right
coronary artery, while the left coronary flow is predominant in about 20%
of humans. The coronary flow from the large conduit arteries, which are on
the epicardium (i.e., surface of the heart), branch into arterioles and then
capillaries that supply the heart muscle. The blood from the capillaries
circulates through the venules and the large conduit veins back to the heart.
Of the returning venous blood, 85% drains through the coronary sinus. The
remaining 15% flows through the anterior superficial veins and the thesbian
veins and directly empties into the cardiac chambers (Marcus, 1983).
The coronary artery circulation has been extensively studied because of
the importance of the coronary blood supply in maintaining the proper
function of the heart. The resting coronary blood flow is about 225 ml/min,
or about 0.8 ml/min/g of the heart muscle. When the heart muscles work
harder (e.g., during exercise conditions), the coronary blood supply can
increase by 4 to 10 times the resting blood flow rate. The rate of blood flow
through the coronary arteries during a cardiac cycle is illustrated in
Figure 3.17. During systole, when the heart muscles are contracted, very little
flow occurs in the coronary arteries due to the constriction of the intramus-
cular branch vessels of the coronary arteries. During diastole, the heart
muscles are relaxed, the lumen of the intramuscular arteries open fully, and
the major portion of the coronary flow occurs. Measurements of coronary
blood flow with electromagnetic flow meters have indicated that the flow
through the arteries in systole constitutes from 7 to 45% of the total coronary
flow during a cardiac cycle (Rushmer, 1976).
During resting conditions, 65% of the oxygen is extracted from the coro-
nary arterial blood as it passes through the heart. The rate of blood flow
through the coronary arteries is regulated by the myocardial oxygen
demand. Under severe exercise conditions, the cardiac output can increase
three to six times the resting value. To meet the increased oxygen demand
as the cardiac muscles are working harder to increase the cardiac output,
the coronary blood flow also increases by several fold. The fact that coronary
blood flow is regulated in response to the myocardial oxygen demand can
be illustrated by occluding the coronary arteries for a few seconds and then
7328_C003.fm Page 93 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 93

300 Left ventricle

capillaries

Coronary blood flow (ml/min)

200

100

0
Systole Diastole

FIGURE 3.17
The blood flow rate in the human left coronary vessels in systole and diastole extrapolated
from dog data. (Redrawn from Guyton, A.C. and Hall, J.E. (2000) A Textbook of Medical Physiology,
W.B. Saunders Company, Philadelphia. With permission.)

releasing the occlusion. The myocardium distal to the occlusion has a reduc-
tion in blood supply and oxygen during the occlusion. Hence, immediately
after the occlusion is released, the blood flow through the vessel increases
significantly and stays high for a length of time depending upon the length
of occlusion. This phenomenon, referred to as reactive hyperemia, is illus-
trated in Figure 3.18.
Chest pain, or angina pectoris, results if the myocardium receives inade-
quate oxygen supply. Angina pectoris can be initiated by physical exertion
or emotional stress and is usually relieved by rest or administration of nitrites

Peak flow or velocity

Coronary blood flow

Control flow Oxygen deficit

or velocity

or velocity

Debt 15 sec
area

Duration of Duration of deficit

occusion
Time (sec)

FIGURE 3.18
Reactive hyperemia caused by occlusion of the coronary vessels.
7328_C003.fm Page 94 Monday, September 18, 2006 10:57 AM

94 Biofluid Mechanics: The Human Circulation

(drugs that dilate the arteries). Myocardial infarction (MI) is the result of
acute occlusion of coronary vessels. Coronary artery occlusion is due to the
presence of vascular lesions called atherosclerosis and will result in lack of
blood supply to the portion of the myocardium perfused by the artery
together with injury or death to the affected segment. A description of this
disease, its predilection, and attempts to relate abnormal flow-induced
stresses to the etiology of these lesions are discussed later in this chapter
and in subsequent chapters in this book.

3.6 Pulmonary Circulation and Gas Exchange in the Lungs

The exchange of oxygen and carbon dioxide in the lungs takes place between
the air in the alveolar sac and the blood in the lung capillaries. The lungs
contract or expand due to the upward or downward movement of the
diaphragm and by the depression or elevation of the ribs to increase the
anterior–posterior diameter of the chest cavity, although normal, quiet
breathing is principally accomplished by movement of the diaphragm alone.
The diaphragm motion is downward, pulling the lower surfaces of the lung
along with it during inspiration. The diaphragm then relaxes and the elastic
lung tissue recoils during expiration. In heavy breathing, such as during
exercise, additional forces due to the contraction of abdominal muscles assist
in the motion of the diaphragm. The lung is surrounded by a thin layer of
pleural fluid that lubricates the movement of the lung within the thoracic
cavity. The pressure within the pleural space between the lungs and the
thoracic cavity is slightly negative (approximately – 5 cm of water) at the
beginning of inspiration and as the expansion of the thoracic cavity pulls
the lungs outward and the pleural pressure becomes more negative (approx-
imately – 7.5 cm water). Obviously, when no air is moving in and out of the
lungs, the pressure in all parts of the respiratory tract, including the alveoli,
is equal to atmospheric pressure. During inspiration, the pressure in the
alveoli falls to about – 1 cm water with respect to atmospheric pressure and
about 0.5 liter of air fills the lung. During expiration, the alveolar pressure
increases to + 1 cm water to expire an equal amount of air. The breathed air
passes through the trachea, several generations of bronchi, and the bronchi-
oles before reaching the alveolar sacs.
Under normal respiratory conditions, a pressure gradient between the
atmosphere and alveoli of 1 cm of water is sufficient to maintain airflow into
and out of the lungs. Multiple cartilage rings in the trachea and bronchi
provide these tubes with a reasonable amount of rigidity to prevent collapse
and to allow passage of air. These cartilage rings are not present in the later
generations of the bronchi or in the bronchioles where the walls in these
segments are composed mainly of smooth muscles. Obstructive diseases in
the lung (e.g., COPD — chronic obstructive pulmonary disease) occur from
7328_C003.fm Page 95 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 95

the narrowing of the smaller bronchi and bronchioles due to the constriction
of smooth muscle cells in these smaller-sized passages. Histamine and slow
reactive substances of anaphylaxis, which are released in the lungs due to
allergic reactions (caused, for example, by pollen in the air), can induce
bronchiolar constriction. The respiratory passages are kept moist by a layer
of mucus that coats the entire luminal surface. The mucus also traps small
particles from the inspired air and keeps most of the particles from reaching
the alveoli. The respiratory passages are also lined with epithelial cells con-
taining about 200 celia, or flagella, in each cell. These celia beat continually
at a rate of 10 to 20 times per second, inducing motion of the mucus towards
the pharynx which, together with the entrapped particles, is either swal-
lowed or coughed out to the exterior.
Blood flow to the lungs is provided by the contraction of the right ventricle
into the pulmonary artery that extends about 4 cm beyond the pulmonary
valve before branching into the right and left pulmonary arteries that feed
the two lungs. The pressure pulse curve for the right ventricle is compared
to that of the aorta in Figure 3.19a and the pressures in the blood vessels of
the lung are shown in Figure 3.19b. The systolic and diastolic right ventric-
ular pressures are about 22 and 0 mmHg, respectively. In the pulmonary
artery, the corresponding pressures are 22 and 8 mmHg with a pulse pressure
of 14 mmHg. By conservation of mass, the rate of blood flow from the right
ventricle into the lungs averaged over time must be the same as the rate at
which blood is pumped by the left ventricle into the systemic circulation.
Thus, the flow rate through the lungs is very high compared to that of other
organs because the entire cardiac output must flow through the lungs to
become oxygenated. This is possible fluid dynamically because the distance
traversed by the blood in the pulmonary circulation is relatively short com-
pared to that in the systemic circulation. Additionally, the pulmonary arte-
rioles are larger in size and more distensible than their systemic counterparts,
causing the resistance in the pulmonary circulation to be low. Therefore, the
lower pressures generated by the right ventricle are still sufficient to maintain
the same cardiac output through the pulmonary circulation and back into
the left atrium (see Equation 1.74). The lungs themselves are supplied with
oxygenated blood for nutrition through the bronchial arteries, which origi-
nate from the systemic circulation. The flow in this circuit amounts to only
1 to 2% of the cardiac output. The bronchial circulation supplies oxygen to
supporting tissues of the lungs, including connective tissue as well as the
bronchi. The blood from the bronchial circulation empties into the pulmo-
nary veins and returns to the left atrium, thus bypassing the lung capillaries
(i.e., alveoli).
Because of the low pulmonic pressures, distribution of blood flow in the
lung capillaries is noticeably affected by gravity. More specifically, when a
person is in the standing position, pressure in the base (inferior) region is
higher than in the apical (superior) region of the lung. Another important
phenomenon is that the capillaries in the alveolar walls are distended by
blood pressure, but are also compressed by air pressure in the alveoli. Hence,
7328_C003.fm Page 96 Monday, September 18, 2006 10:57 AM

96 Biofluid Mechanics: The Human Circulation

Aortic pressure curve

120

Pressure (mm Hg)

75

Right ventricular curve

Pulmonary artery curve
25

8
0
0 1 2
Seconds
(a)

25 S
mm Hg

15 M

8 D
7

2
0
Pulmonary Pulmonary Left
artery capillaries atrium
(b)

FIGURE 3.19
(a) Typical pressure pulses in the right ventricle, pulmonary artery, and the aorta; (b) distribution
of pressures in the blood vessels in the lung. (Redrawn from Guyton, A. C. and Hall, J. E. (2000)
A Textbook of Medical Physiology, W.B. Saunders Company, Philadelphia. With permission.)

in the regions where the alveolar air pressure becomes larger than that of
the capillary blood pressure, the capillaries collapse and prevent blood flow.
Therefore, in the basal regions of the lungs, where the blood pressure is
always higher than that of the alveolar pressure, there is continuous blood
flow. In the apical zones, however, the blood pressure becomes higher than
the alveolar pressure only during the systolic phase and, thus, results in an
intermittent flow through the capillaries. During exercise when there is an
increase in cardiac output, the flow through the lungs may increase by
fourfold up to sevenfold. This additional flow is accomplished by increasing
7328_C003.fm Page 97 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 97

the number of open capillaries, distension of the capillaries, and by an in-

crease in the pulmonary arterial pressure.
The exchange of oxygen from the fresh inspired air in the alveolar sac to
blood in the pulmonary capillaries and carbon dioxide in the opposite direc-
tion takes place through the process of diffusion from a region of high to
low species concentration. One-dimensional transport of nonelectrolytes
across a membrane is governed by Fick’s law, which is expressed as

dCi ∆Ci
Ni = − Di = − Di (3.7)
dz δ

In this relationship, Ni is the molar flux of the species i, Di is the diffusion

coefficient, and Ci is the molar concentration; δ is the thickness of the mem-
brane across which the mass transport occurs. The diffusion flux can also be
expressed in terms of the partial pressure of the species dissolved in air or
in blood. The partial pressure of the dissolved gas is given by Henry’s law,
or ∆Ci = K D ∆pi where K D is the solubility coefficient and ∆p is the partial
pressure difference. The transport of gases in the lungs is induced by the
difference in partial pressures between the alveolar air and the pulmonary
capillary blood across the respiratory membrane. The respiratory mem-
brane consists of a layer of fluid lining the alveolus containing a surfactant
(to reduce the surface tension of the alveolar air), the alveolar epithelium,
an epithelial basement membrane, a thin interstitial space, a capillary base-
ment membrane, and the capillary endothelial cell layer. The thickness of
the respiratory membrane averages about 6 µm and the total surface area of
the respiratory membrane in normal adults is about 70 m2 for efficient
exchange of gases. The total amount of gas transported across the respiratory
membrane can be obtained as the product of the mass flux and the area of
the respiratory membrane available for the transport and is given by

Di K Di A
w=− ∆pi = − DLi ∆pi (3.8)
δ

In this equation, w is the total amount of gas transported across the respi-
ratory membrane cross-sectional area, A. Since A and δ are very difficult to
determine accurately, all the parameters in the equation are lumped together
as the diffusing capacity, DL. Under resting conditions, the diffusing capacity
of O2 is 21 ml min− mmHg and that of CO2 is 400 to 450 ml min− mmHg. The time-
averaged, partial pressure difference across the respiratory membrane under
resting conditions is about 11 mmHg and the total amount of oxygen trans-
ferred across the respiratory membrane is about 252 ml/min. The total amount
of CO2 transferred is about 200 ml/min. Since the solubility of oxygen in aque-
ous solution is rather poor, oxygen diffuses across the red blood cell membrane
in the capillary and then reacts with hemoglobin in the intracellular fluid
volume so that it is transported principally as oxy-hemoglobin.
7328_C003.fm Page 98 Monday, September 18, 2006 10:57 AM

98 Biofluid Mechanics: The Human Circulation

3.7 Cerebral and Renal Circulations

3.7.1 Cerebral Circulation
The carotid arteries and the vertebral arteries supply blood circulation to the
brain. The brachiocephalic, left common carotid, and left subclavian arterial
branches originate from the aortic arch. The brachiocephalic artery further
divides into the right subclavian and the right common carotid arteries. In
the neck region, the carotid arteries bifurcate into the internal and external
carotid arteries. The external carotid arteries supply blood to the face and
the skull, while the internal carotids feed the cranial cavity. In addition,
vertebral arteries branch out from the two subclavian arteries and ascend to
the cranial cavity protected by the bony arches of the cervical vertebrae. The
vertebral arteries join to form the basilar artery in the hind (back) brain,
which is connected to the internal carotid arteries through the Circle of
Willis as shown in Figure 3.20.

Anterior Anterior
communicating cerebral
artery artery

Internal
carotid Anterior
artery choroidal
artery

Middle Posterior
cerebral communicating
artery artery

Posterior
cerebral
artery
Superior
Pontine cerebellar
arteries artery
Basilar artery

Anterior spinal Anterior inferior

artery
Vertebral artery Posterior inferior cerebellar artery
cerebellar artery

FIGURE 3.20
Schematic of blood flow in the human brain showing the Circle of Willis connecting flow from
the vertebral and carotid arteries. (Redrawn from DeArmond, S.J., Fusco, M.M., and Dewey,
M.M. (1974) Structure of the Human Brain, Oxford University Press, London. With permission.)
7328_C003.fm Page 99 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 99

The Circle of Willis is a protective mechanism in that blood flow can be

shunted to different parts of the brain around the circle if one of the feeding
arteries is occluded. The normal blood flow to the brain is about 750 ml/min
or 50 to 55 ml/min per 100 g of brain matter. The control of blood supply
to the brain is such that the blood supply is maintained at this level even
under severe conditions.

3.7.2 Renal Circulation

The right and left renal arteries originate from the descending aorta and
supply blood to the two kidneys. In addition to removing the body waste
material that is either ingested or produced by metabolism, the kidneys also
regulate water and electrolyte balances, body fluid osmolality and electrolyte
concentration, acid-base balance, and blood pressure and volume. The two
kidneys lie on the posterior wall of the abdomen, outside the peritoneal
cavity. The renal circulation is unique in that it has two capillary beds, the
glomerular and peritubular capillaries. The blood flow rate to the kidneys
is about 1100 to 1200 ml/min or about 25% of the cardiac output. The renal
artery branches progressively to form the interlobar arteries, arcuate arteries,
and interlobular arteries (also called radial arteries) that lead into the afferent
arterioles. The glomerular capillaries that lie at the end of the afferent arte-
rioles filter large amounts of fluid and solutes, except for the plasma proteins,
to begin the process for urine formation. High hydrostatic pressures in the
glomerular capillaries cause rapid filtration across these capillaries into the
Bowman’s capsule (Figure 3.21). Distally, these capillaries coalesce to form
the efferent arterioles leading to the peritubular capillaries. Lower hydro-
static pressures in these capillaries enable fluid to be rapidly absorbed from
the filtrate transported from the Bowman’s capsule via the tubules traversing
adjacent to the capillary bed. By adjusting the resistances of the afferent and
efferent arterioles, the kidneys can alter the amount of glomerular filtrate as
well as the amount of fluid reabsorbed based on the body’s demand. The
peritubular capillaries empty into the interlobular veins that progressively
coalesce into the arcuate veins, interlobar veins and into the renal veins.
These vessels join the venous circulation (inferior vena cava) that returns the
blood to the right atrium.

3.8 Microcirculation
In the previous sections, we reviewed blood flow in both the systemic and
pulmonary circulations. We started with the contractile function of the heart
in generating a sufficient pressure gradient in order to induce flow along the
circulatory system. We also discussed the arterial system, which acts as the
7328_C003.fm Page 100 Monday, September 18, 2006 10:57 AM

100 Biofluid Mechanics: The Human Circulation

Renal artery

Renal vein

(a)

Glomerulus
Efferent (capillaries)
arteriole Peritubular
capillaries
Juxtaglomerular
apparatus

Afferent
arteriole

(b)

FIGURE 3.21
Schematic of the blood supply to the kidneys: (a) renal artery and vein, and (b) details of the
afferent and efferent arterioles along with the Bowman’s capsule used in the filtration of blood
of waste products. (Redrawn from Silverthorn, D.U. (2001) Human Physiology: An Integrated
Approach, 2nd ed., Prentice Hall, Upper Saddle River, NJ. With permission.)
7328_C003.fm Page 101 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 101

Arteries Vein

Venule
Precapillary
sphincters
Arteriole
Metarteriole

Capillaries

Small venule

Arteriovenous
bypass
Precapillary
sphincters

FIGURE 3.22
Schematic of the blood vessels in a typical microcirculation. (Redrawn from Silverthorn, D.U.
(2001) Human Physiology: An Integreted Approach, 2nd ed., Prentice Hall, Upper Saddle River,
NJ. With permission.

main conduits transporting blood to the various visceral organs and the
peripheries of the human body. In large blood vessels, the typical size (i.e.,
diameter) of the blood vessel is several magnitudes larger than the formed
elements of blood, such as the red blood cells and the white blood cells, so
that blood may be treated as a homogeneous fluid. These large vessels mainly
function as conduits for supplying blood to the capillary network.
The exchange of nutrients and cellular excreta between the tissues and the
flowing blood occurs at the level of capillaries. The fluid mechanics within
the arterioles and capillaries are characteristically different from those in
large arteries. Hence, flow of blood in the system of blood vessels consisting
of arterioles, capillaries, and venules is termed the microcirculation. A sche-
matic of a typical micro-vascular bed is shown in Figure 3.22. The arteries
divide into arterioles and the diameters of arterioles can range up to 100µm.
The walls of the arterioles are highly muscular and have the ability to alter the
vessel diameter several fold and, hence, vary the resistance to flow across the
micro-vascular bed. The arterioles divide into metarterioles with diameters
of about 20 µm. The musculature in the metarterioles is not continuous, as
7328_C003.fm Page 102 Monday, September 18, 2006 10:57 AM

102 Biofluid Mechanics: The Human Circulation

smooth muscle fibers are found intermittently surrounding the wall. The
branching pattern of the capillary network varies from tissue to tissue as
well as within any one bed. Some of the capillary vessels (referred to as “true
capillaries”) that branch from the metarterioles have precapillary sphincters
at their origin. The sphincters consist of one or two smooth muscle cells
wrapped around the arteriolar–capillary junction. The diameter of the true
capillary ranges from 5 to 8µm, barely large enough for red blood cells and
other elements to squeeze through. Some of the capillaries are large and do
not have the precapillary sphincters and are referred to as the preferential
channels. If the precapillary sphincters constrict, the blood is shunted
through the low-resistance preferential channels and, hence, the micro-
vascular bed unit is bypassed.
The concentration of oxygen in the tissue supplied by the unit regulates
the degree of opening and closing of the precapillary sphincters and amount
of blood flowing through the true capillaries. The wall of the true capillaries
consists of a unicellular layer of endothelial cells surrounded by a thin
basement membrane of about 0.5 µm in thickness. Pores or passageways,
far apart from one another with a fraction of the lumen surface area, are
present in the capillary membrane through which water and water-soluble
substances pass between the lumen of the capillary and the interstitial fluid.
Several of these water-soluble substances, such as sodium and chloride ions
and glucose, are not lipid-soluble and, therefore, can only pass through the
pores. Transport of fluid between the capillary lumen and the interstitial
fluid depends on the hydrostatic and colloid osmotic pressure difference
between the capillary blood and the interstitial fluid. The rate at which water-
soluble substances transport across the pores depends upon the permeability


of the capillary pores. The diameter of the pores is about 80 to 90 Α . Water
molecules are much smaller than the pore diameter and, hence, have a larger
permeability compared to that of the plasma protein molecules. Lipid soluble
substances, such as oxygen and carbon dioxide, are transported by means
of diffusion across the cell membrane into the capillary. The amount of
diffusion is directly proportional to the concentration difference across the
two sides of the membrane. At the distal end, the capillaries merge into post
capillaries and then into venules. In the systemic circulation, venules merge
into larger veins terminating in the superior and inferior vena cavae, which
drain the blood into the right atrium. The lymphatic circulation drains the
excess fluid from the interstitial locations. Lymphatic capillaries are blind-
ended sacs present in the interstitial space around the collecting venules and
which merge into collecting lymphatics. These become lymphatic vessels,
which run parallel to the veins and which drain via the thoracic duct into
the subclavian vein in the neck.
Even the largest arterioles in the micro-vascular bed have a diameter that
is only about 15 times larger than that of the red blood cells. The true capillary
diameter is about the same size or smaller than that of the red blood cell
and, thus, the cells must squeeze through the capillaries. In such flow situ-
ations, the blood cannot be considered as a continuum and the interaction
7328_C003.fm Page 103 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 103

between particulates (i.e., cells) and the plasma must be taken into account.
Due to the small diameters and low velocities in the microcirculation, viscous
effects dominate over inertial effects and the magnitude of the Reynolds
number ranges between 0.005 and 0.5. However, the velocity gradient at the
wall or the wall shear rates in the microcirculation is of the order of 1000 s-1
and are relatively large compared to that of those in the arteries as we will
observe in subsequent chapters. The study of microcirculatory dynamics is
beyond the scope of this textbook, but can be found in treatises such as Caro
et al. (1978) and Fung (1984).

3.9 Regulation of the Circulation

The human body has mechanisms for regulation of the heart function and
the circulation based on the needs of the various regions and the visceral
organs of the body. Regulation of the circulation consists of systemic control
as well as the ability for each tissue to control its own blood flow based on
its metabolic needs. The metabolic needs include the delivery of oxygen and
nutrients, such as glucose, amino acids, and fatty acids, removal of carbon
dioxide and hydrogen ions from the tissues, transport of hormones, and
maintenance of a concentration balance of other ions. Acute control of local
blood flow is achieved by altering the resistance to blood flow into the region
by rapid changes in the tone of the arterioles, metarterioles, and precapillary
sphincters within a few seconds to several minutes. The widely accepted
mechanism through which the local vascular tone is altered is that with
higher metabolism of a tissue, the availability of oxygen and other metabo-
lites decreases. In response, the tissue releases vasodilators, such as adenosine
and carbon dioxide, among others, in order to reduce the vascular resistance
and increase blood flow to the tissue. Another possible mechanism is that,
due to lack of sufficient oxygen, the blood vessels naturally relax and dilate
to allow more blood flow.
A typical example of the local regulation of blood flow is the reactive
hyperemia in the coronary circulation described earlier (Figure 3.18). The
local mechanism for increasing blood flow into the tissues is dilation of the
micro vessels in the tissue. Rapid flow of blood in the arterioles and small
arteries also causes an increase in the wall shear stress acting on the endo-thelial
cells. The endothelial cells respond to this stimulus by releasing endo-thelial-
derived relaxation factor (EDRF), or nitric oxide (NO). Nitric oxide acts to
relax the local arterial wall and, hence, the vessels dilate locally. Both the
kidneys and the brain have special mechanisms for control of blood flow. In
the case of kidneys, the composition of the fluid in the early distal tubule is
monitored by a tubular epithelial structure called the macula densa. When
too much fluid is filtered through the glomerular capillaries into the tubular
system, feedback signals from the macula densa constrict the afferent and
7328_C003.fm Page 104 Monday, September 18, 2006 10:57 AM

104 Biofluid Mechanics: The Human Circulation

efferent arterioles in the kidneys to reduce the renal blood flow and the
glomerular filtrate. In the brain, the concentration of carbon dioxide and
hydrogen ions also influences the blood flow rate to the organ in addition
to the effect of changes in tissue oxygen concentration.
Long-term control of local blood flow (i.e., effecting changes over a period
of weeks or months) in proportion to the needs of the tissue is the result of
increase or decrease of the physical size and number of blood vessels sup-
plying the tissue. If the metabolic rate in a given tissue is increased over a
prolonged period, the vascularity is also increased. Growth of new blood
vessels to increase vascularity results from the presence of various peptides,
such as vascular endothelial growth factor (VEGF), fibroblast growth factor
(FGF), and angiogenin, among others. Another mechanism for increasing
the vascularity is by development of a collateral circulation. When a blood
vessel is blocked, collateral vessels develop in the region around the blocked
vessel over a period of time in order to compensate for the decrease in blood
supply. For example, collateral blood supply is an important mechanism for
keeping the cardiac muscles viable and possibly preventing a heart attack
after thrombosis of a coronary vessel.
The global nervous control of circulation and cardiac function is performed
by the autonomic sympathetic and parasympathetic nervous systems. The
control of both these nervous systems is affected by the vasomotor center in
the brain. The sympathetic vasomotor nerve fibers pass through the spinal
cord and innervate the visceral organs and the heart as well as the vascula-
ture in the peripheral region. Parasympathetic nerve fibers do not play a
major role in the control of circulation, but control the heart by the nerve
fibers contained in the vagus nerve. The sympathetic nerve fibers carry a
large number of vasoconstrictor fibers that secrete norepinephrine. Constric-
tion of the blood vessels by norepinephrine increases the resistance to blood
flow (due to a reduction in the radius of the vessels), resulting in an increase
in the blood pressure. The large veins are also constricted and, hence, blood
from these highly distensible vessels is displaced toward the heart. Thus,
the volume of ventricular filling increases, resulting in increased cardiac
output through the Frank–Starling mechanism. With increase in the filling
volume, the heart muscles contract with increased force and, hence, the
stroke volume (and the cardiac output) increases together with an increase
in blood pressure. The sympathetic nervous system also directly acts on the
heart by increasing both the heart rate and the force of contraction. Parasym-
pathetic nerve fibers, on the other hand, decrease the heart rate and cause a
slight decrease in heart contractility resulting in a reduction in cardiac output
and blood pressure.
Baroreceptors, or pressoreceptors, are stretch receptors in the nervous
control system that help to control the arterial blood pressure. These recep-
tors are located particularly in the wall of the aortic arch and in the walls of
the internal carotid artery slightly above the bifurcation in the carotid sinus.
The baroreceptors respond rapidly to changes in blood pressure (rather than
to the mean arterial pressure). Should the blood pressure rapidly rise, the
7328_C003.fm Page 105 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 105

signals from the baroreceptors inhibit the vasoconstrictor nerves and excite
the vagal parasympathetic nerves. The net result is the dilation of the periph-
eral blood vessels, while the heart rate and force of contraction of the cardiac
muscles is reduced. Should the blood pressure rapidly decrease, the opposite
effects will take place resulting in an increase in blood pressure. The nervous
control system can act to rapidly control the blood pressure for a short time
on the order of minutes.
The kidneys play an important role in the long-term control of blood
pressure and in the development of hypertension (i.e., high blood pressure).
When the body contains too much extracellular fluid, the blood volume and
mean arterial pressure rises. The kidneys react to the increased blood pres-
sure by excreting excess body fluid and return the blood pressure to its
normal value. This phenomenon of increasing urinary output to control the
blood pressure is known as pressure diuresis. Aside from using changes in
the extracellular fluid in the kidneys to control blood pressure, the
Renin–Angiotensin system in the kidneys also plays an important role in
blood pressure control. With a decrease in the blood pressure, renin is
released in the kidneys. This enzyme reacts chemically with plasma proteins
to release Angiotensin I and II. Intense vasoconstriction caused by the effect
of Angiotensin II in the artery results in an increase in blood pressure. Mild
constriction of the veins displaces additional blood to the heart resulting in
increase in ventricular filling and cardiac output and an increase in blood
pressure as well. When the mean arterial pressure rises to magnitudes above
110 mmHg (along with corresponding increases in systolic and diastolic
blood pressures), this condition is referred to as hypertension. Hypertension
can be a result of release of large quantities of renin in the kidneys due to
malfunction of the renin-secreting juxtaglomerular cells. Chronic high blood
pressure will result in excess workload and eventual failure of the heart,
rupture of blood vessels, particularly in the brain, renal hemorrhage, and
renal failure.

3.10 Atherosclerosis
Diseases in the blood vessels and in the heart, such as heart attack and stroke,
are responsible for nearly half of the deaths in the U.S. The underlying cause
for these events is the formation of lesions, known as atherosclerotic plaques,
in the large- and medium-sized arteries in the human circulation. These
plaques can grow and occlude the artery and, hence, prevent blood supply
to its distal bed. Plaques containing calcium can also rupture and initiate the
formation of blood clots (thrombus). The clots can form emboli, which are
carried in the bloodstream and occlude smaller vessels, resulting in interrup-
tion of blood supply to the distal bed. Plaques formed in coronary arteries
can lead to heart attacks, while those formed in the cerebral circulation can
7328_C003.fm Page 106 Monday, September 18, 2006 10:57 AM

106 Biofluid Mechanics: The Human Circulation

result in stroke. There are a number of risk factors for the presence of athero-
sclerotic lesions. Those for which we have no control include gender, age,
and familial history of early cardiovascular disease. Risk factors that can be
controlled to minimize the occurrence of atherosclerosis include smoking,
obesity, sedentary lifestyle, elevated serum cholesterol and triglycerides,
untreated high blood pressure, and diabetes mellitus.

3.10.1 Morphology of Atherosclerosis

Atherosclerosis is associated with lipid deposition in the subendothelial space,
intimal thickening, SMC (Smooth Muscle Cell) proliferation, and plaque forma-
tion (Figure 3.23). It is an inflammatory response of the arterial wall to injury
resulting from fluid dynamics, local oxidation of lipoproteins, or toxins (carbon
monoxide). Increased accumulation of lipid and lipoprotein particles under-
neath the endothelium is made possible by locally enhanced permeability. Oxi-
dation of lipoproteins within the intima leads to activation of endothelial cells
and expression of receptors for monocytes and lymphocytes (immune response).
In the presence of elevated plasma cholesterol and high levels of low
density lipoproteins [LDL], oxidation within the intima, formation of foam

Diseases of the heart and blood vessels

FIGURE 3.23
Classical light micrograph of a cross section of a coronary artery that contains a large atheroscle-
rotic lesion. (Redrawn from Schlant, RC. and Alexander, R.W., Eds. (1994) Hurst’s The Heart, Arteries
and Veins, 8th ed., McGraw-Hill, Health Professions Division, New York. With permission.)
7328_C003.fm Page 107 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 107

cells from monocytes and mast cells, and their continued accumulation in
the intima leads to an atherosclerotic lesion, or fatty streak. With the con-
tinuing effect of hypercholesterolemia, endothelial cell injury, and abnormal
hemodynamics, the fatty streak will grow into a fibro-fatty lesion containing
multiple layers of SMC, connective tissue, macrophages, and T-lymphocytes.
With further growth, a fibrous cap will form to create an advanced lesion.
The role of elevated blood cholesterol levels in the development of athero-
sclerosis is well established. Cholesterol is a lipid that is absorbed in the
digestive tract and is not soluble in aqueous solutions. Cholesterol combines
with a lipoprotein molecule so that it can dissolve in the plasma. Cholesterol
that combines with high density lipoproteins (HDL) is associated with
lower risk of heart attacks, whereas elevated levels of low density lipoprotein
(LDL) cholesterol is associated with arterial disease. The LDL carrier is
necessary for cholesterol uptake into all of the body’s cells. The LDL portion
of the complex combines with an LDL receptor in the cell membrane and
the receptor–LDL–cholesterol complex enters into the cell by endocytosis.
However, if LDL-cholesterol is not absorbed into the cells and remains in
the blood, the excess LDL-cholesterol will move into the walls of the arteries.

3.10.2 Sequence of Growth of the Lesion

The initiation and development of atherosclerotic plaques is schematically
depicted in Figure 3.24. In the early stages of the lesion formation, the excess
LDL-cholesterol accumulates in the acellular layer between the endothelium
and the connective tissue. There it is oxidized and absorbed by the macro-
phages through phagocytosis. When the macrophages are filled with oxi-
dized LDL-cholesterol, they produce paracrines — substances that attract
smooth muscle cells to the region, resulting in the formation of fatty streaks.
When these events occur, newly formed cells in the media begin migrating
into the intima and lay down additional fibrous extracellular matrix material.
As cholesterol continues to accumulate, fibrous scar tissue forms around the
cholesterol. The migrating smooth muscle cells also divide and, as a result,
the intima begins to thicken, or become hyperplastic. The resultant intimal
hyperplasia (IH) is a serious chronic response of arterial tissue to changes
in local blood flow, which may result in arterial occlusion. When IH is
combined with the abnormal accumulation of excess fatty material, as well
as calcium, macrophages from the blood stream, and necrotic tissue, an
atherosclerotic plaque is said to form (Figure 3.25). The enlarging atherosclerotic
plaque then starts to protrude into the lumen until it eventually becomes a
flow obstruction.

3.10.3 Physiological Implications

To compensate for expanding atheroma, the arteries initially remodel by
distending and becoming thinner to maintain the arterial lumen cross section
7328_C003.fm Page 108 Monday, September 18, 2006 10:57 AM

108 Biofluid Mechanics: The Human Circulation

Endothelial cells
Elastic lamella

Smooth
(a) muscle cells
Monocytes
Platelets

(b)

Lipid arootets

(c)
Monocytes smooth
muscle cells

(d)

Cholesterol crystals

Blood vessel

(e)

FIGURE 3.24
Postulated sequence of events in the pathogenesis of atherosclerosis. (From White, R.A. and
Cavaye, D.M. (1993), in A Text and Atlas of Arterial Imaging – Modern and Developing Technology,
Cavaye, D.M. and White, R.A., Eds., W.W. Lippincott, Philadelphia. With permission.)

for normal blood flow. Once the remodeling limit is reached, however, the
lesion protrudes into the lumen forming a stenosis. In the advanced stages
of the lesion, the cells in the plaque may die and form calcified scar tissue.
The deposits of calcium create the hardening of the arteries, which reduces
the distensibility in the region of the atherosclerotic plaques. If the plaques
have a rough surface or if they rupture, platelets will stick to the damaged
area and to each other to form a thrombus on the arterial wall. Figure 3.26
7328_C003.fm Page 109 Tuesday, September 19, 2006 1:53 PM

Cardiovascular Physiology 109

The Clogging of an Artery Artery

Area of
detail
Lipoprotein
Monocyte
Fatty
streak
Endothelial
cells
Subendothelial
space
Smooth
muscle
cells

Lipoprotein
with oxidized Foam cells
cholesterol Macrophage

FIGURE 3.25
Schematic of the components involved in the development of atherosclerosis. (From Kleinstreu-
er et al. (2001). A computational analysis of particle-hemodynamics and prediction of the onset
of arterial disease, in Cardiovascular Techniques, Leondes, C., Ed., CRC Press, Boca Raton, FL.
With permission.)

shows the progression of initial fatty lesions to calcified plaques in the intimal
layer of a blood vessel and the gradual decrease in the lumen cross section
available for blood flow.
The common sites of predilection of atherosclerosis in the human circulation
are shown in Figure 3.27. These include the coronary arteries, the branching
of the subclavian and common carotids in the aortic arch, the bifurcation of
the common carotid to the internal and external carotids (especially in the
carotid sinus region distal to the bifurcation), the renal arterial branching in
the descending aorta, and in the ileo-femoral bifurcations of the descending
aorta. The common feature for the development of the lesion is the presence
of curvature, branching, and bifurcation present in these sites. Besides the
geometry, the fluid dynamics at these sites can be anticipated to be consid-
erably different from that at other segments of arteries that are relatively
straight and devoid of any branching segments. Hence, a number of inves-
tigators have attempted to link fluid dynamically induced stresses with the
formation of atherosclerotic lesions in the human circulation. Hemodynamic
theories for the genesis of atheromatous plaques and detailed descriptions
of the fluid dynamics at corresponding sites will be the topic of discussion
in subsequent chapters, where steady and unsteady flow models in the
human circulation will be considered.
7328_C003.fm Page 110 Monday, September 18, 2006 10:57 AM

110 Biofluid Mechanics: The Human Circulation

Liceration Normal

Atheroma Intimal thickening

Medial sclerosis Intimal lipid accumulate

Medial fibrosis
Medial Intimal or
Thrombus
muscle fibers medial fibrosis

Cholesterol crystals Medial calcification Lipid deposits

FIGURE 3.26
Schematic of the cross section of a blood vessel illustrating the growth of atherosclerotic lesions
progressively reducing the lumen area available for blood flow to the distal beds supplied by the
vessel. (Redrawn from Engeler et al., (1991) Intravascular sonography in the detection of arterioscle-
rosis and evaluation vascular interventional procedures, in AJR 156: 1087–1090. With permission.)

3.11 Problems
3.1 Under normal conditions, the mean systemic arterial pressure at the aortic
root is 95 mmHg and the corresponding central venous pressure is 0 mmHg.
The corresponding pressures in the pulmonary artery and the left atrium are
13 mmHg and 4 mmHg, respectively. If the cardiac output is 5.2 l/min, compute
the peripheral resistance in the systemic and pulmonary circulation. Express
your answers in PRUs as well as in SI units. In the case of moderate exercise,
the mean aortic pressure rose to 150 mmHg and the cardiac output increased
by three times the normal value. Compute the percentage change in the sys-
temic peripheral resistance. (Conversion factor: 1 mmHg = 133.3 Pa.)
3.2 In a clinical situation, the stroke volume for a patient was measured at
45 ml/stroke and the heart rate was recorded to be 85 beats per minute.
Compute the cardiac output for the patient. Cardiac Index is expressed as
the ratio of cardiac output over the body surface area. Assuming that the
7328_C003.fm Page 111 Monday, September 18, 2006 10:57 AM

Cardiovascular Physiology 111

I II
Proximal

Mid-proximal

Distal

IV

III

FIGURE 3.27
Common sites for the presence of atherosclerotic plaques in the human circulation. (Redrawn
from DeBakey et al., (1985) Patterns of artherosclerosis and their surgical significance, in Ann.
Surg. 201: 115–131. With permission.)

human body can be approximated to be a cylinder, compute the cardiac

index (lpm/m2) for the patient if the height and average diameter for the
patient were measured to be 5.6 ft and 0.95 ft., respectively.
3.3 The resting coronary blood flow in a normal human is 225 ml/min.
Assuming that 19.4 ml of oxygen is transported by the blood for every
100 ml. of blood and 65% of oxygen is absorbed by the myocardium as the
blood passes through the coronary arteries, compute the amount of oxygen
consumed by the heart muscles each minute.
3.4 In a typical human aortic segment of 10 cm length and an average
diameter of 2.5 cm, an 8% increase in diameter is measured for a pulse
pressure of 40 mmHg. Compute the compliance of the aortic vessel.
3.5 Estimate the work done by the left ventricle of the heart in one day,
in lbf /ft using the P-V diagram for the cardiac cycle shown in Figure 3.9.
7328_C003.fm Page 112 Monday, September 18, 2006 10:57 AM

112 Biofluid Mechanics: The Human Circulation

This estimate is smaller than the actual work done by the heart. Why? How
far could one raise a 100-lb weight with this amount of work?
3.6 The amount of work done per cycle computed in Problem 3.5 is the left
ventricular work output. For a heart rate of 70 bpm, the left ventricular
output per minute can be computed. The amount of oxygen consumed by
the myocardial muscle per minute computed in Problem 3.3 can be converted
as work input to the heart (1 ml of O2 consumed = 20 Joules = 14.75 lbf-ft).
Compute the efficiency of the heart as the ratio of work output to the work
input.
7328_S002.fm Page 113 Friday, July 28, 2006 4:36 PM

Part II

Biomechanics of the Human

Circulation
7328_S002.fm Page 114 Friday, July 28, 2006 4:36 PM
7328_C004.fm Page 115 Friday, October 6, 2006 12:25 PM

4
Rheology of Blood and Blood Vessel
Mechanics

CONTENTS
4.1 Rheology of Blood ..................................................................................... 116
4.1.1 Viscometry and Theory for Capillary, Coaxial
Cylindrical, and Cone and Plate Viscometers.......................... 116
4.1.1.1 Capillary Viscometer ...................................................... 116
4.1.1.2 Coaxial Cylinder Viscometer ........................................ 119
4.1.1.3 Cone and Plate Viscometer ...........................................122
4.1.2 Physical Properties of Blood .......................................................123
4.1.3 Viscous Behavior of Blood...........................................................124
4.1.3.1 Plasma...............................................................................124
4.1.3.2 Whole Blood ....................................................................126
4.1.3.3 Effect of Hematocrit .......................................................130
4.1.3.4 Effect of Temperature.....................................................130
4.1.3.5 Effect of Protein Content in Plasma ............................131
4.1.3.6 Yield Stress for Blood.....................................................131
4.1.3.7 Effect of Tube Diameter .................................................135
4.1.4 Pressure-Flow Relationship for Non-Newtonian Fluids ........139
4.1.4.1 Power Law Fluid ............................................................140
4.1.4.2 Bingham Plastic...............................................................141
4.1.4.3 Casson’s Fluid .................................................................144
4.1.5 Hemolysis and Platelet Activation with Fluid
Dynamically Induced Stresses ....................................................150
4.2 Blood Vessel Mechanics............................................................................152
4.2.1 Structural Components of the Blood Vessel Wall
and their Contribution to Material Behavior ...........................152
4.2.2 Material Behavior of Blood Vessels............................................155
4.2.2.1 Assumptions in the Analysis ........................................158
4.2.2.2 Experimental Results......................................................158
4.2.3 Residual Stress on Blood Vessels................................................162
4.2.4 Material Characterization of Cardiac Muscle...........................164
4.3 Summary .....................................................................................................165
4.4 Problems......................................................................................................165

115
7328_C004.fm Page 116 Friday, October 6, 2006 12:25 PM

116 Biofluid Mechanics: The Human Circulation

4.1 Rheology of Blood

Rheology is the study of deformation and flow of material. An object or
body is said to have deformed if its shape or size is altered due to the
action of appropriate forces on the same. If the degree of deformation
changes continuously with respect to time, the body is considered to be
flowing. This chapter will deal with the behavior of blood as it flows in
the blood vessels. A basic understanding of the rheological or flow char-
acteristics of blood in vivo under normal conditions will help us delineate
the effect of abnormal flow conditions, such as stenosed valves and
obstructed arteries on the formed elements of blood. The understanding
of the fluid dynamics past prosthetic devices, such as heart valves, vas-
cular grafts, and artificial hearts, and the effect of such flow conditions
on blood will also help us in improving the designs of the implants.
Several extracorporeal flow devices, such as blood oxygenators, dialyzers,
and ventricular-assist devices, are commonly used in modern medicine. A
thorough knowledge of blood flow behavior through such devices will
also aid in design improvement so as to minimize trauma to blood and
optimize the functioning of the devices.

4.1.1 Viscometry and Theory for Capillary, Coaxial Cylindrical,

and Cone and Plate Viscometers
Discussed and defined in Chapter 1 (Fundamentals of Fluid Mechanics) were
the viscosity coefficient as an intrinsic property of any fluid and the basic
constitutive relationship between the shear stress and rate of shear for a
Newtonian fluid as well as various classes of non-Newtonian fluids.
Before discussing the viscous properties of blood, we will briefly look
into various methods for measurement of viscosity. Detailed treatment
of viscometry can be found in several textbooks, including that of Dinsdale
and Moore (1962). Since rates of shear and shear stresses are quantities
that are difficult to measure directly, relationships are derived with them
and other quantities that can be measured relatively easily and accurately.
Three basic types of devices for measurement of fluid viscosity are the
(1) capillary viscometer, (2) coaxial cylinder viscometer, and (3) cone and
plate viscometer. The principles behind each type of viscometer are
outlined below.

4.1.1.1 Capillary Viscometer

Capillary viscometry is based on the Poiseuille flow relationship for steady
flow in a long, circular, cylindrical tube that was derived from the equations
of motion, or the Navier–Stokes equations, in Chapter 1 (Equation 1.43a to
7328_C004.fm Page 117 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 117

P1 > P2 ␶
R
P1 P2
r Direction
of flow

FIGURE 4.1
Steady flow of Newtonian fluid through a straight tube of circular cross section.

Equation 1.43c). There, the relationship between the flow rate and the pres-
sure gradient was found to be:

Vmax
Q = Vave ( πR 2 ) = ( πR 2 )
2
(1.72)
 ∆pR 2  ∆pπR 4
=  ( πR 2 ) =
 8µL  8µL

Keep in mind that one of the fundamental assumptions behind this equa-
tion is that the flow is fully developed or, L >> 2R. The nature of flow devel-
oping in the entrance region of a tube is discussed in Chapter 1 (section 1.8:
Fluid Mechanics in a Straight Tube) where an approximate relationship for
the entry length prior to fully developed flow is derived. One way to achieve
this condition is to use a very narrow, or “capillary,” tube.
Thus, in a capillary viscometer of radius R and length L, if the flow rate
and pressure drop of a fluid can be accurately measured, the viscosity coef-
ficient can be determined as

∆pπ R 4
µ= (4.1)
8QL

Referring to Figure 4.1, the net force pushing the fluid to the right is:

Fp = p1π r 2 − p2 π r 2 = ( p1 − p2 )π r 2

while the shear force, which retards the motion acts on the circumferential
surface of the fluid element, is given by

Fτ = τ 2 π rL
7328_C004.fm Page 118 Friday, October 6, 2006 12:25 PM

118 Biofluid Mechanics: The Human Circulation

In fully developed, steady flow, these two forces balance each other and
we obtain

( p1 − p2 )π r 2 = τ(2 π rL)

which can be rearranged to give

( p1 − p2 ) r ∆pr
τ= 2L =−
2L

From this expression for the shear stress, the magnitude of the shear stress
at the wall will be:

( ∆p)r ∆pR
τw = = (4.2)
2L r=R 2L

Furthermore, the constitutive law for a Newtonian fluid (Chapter 1, section

1.2: Intrinsic Fluid Properties) is given in cylindrical coordinates by

du
τ = −µ (1.1)
dr

where u is the velocity of the fluid. Hence, the shear rate at the wall is given
by

∆pR
γ
w =
2µL

Using the relationship for ∆p from Equation 1.74, wall shear rate becomes

4Q
γ
w = (4.3)
πR 3

Thus, having obtained values for wall shear stress and wall shear rate
(Equation 4.2 and Equation 4.3), respectively, one can calculate fluid viscosity
from their ratio.
Capillary viscometers include devices that measure the absolute viscosity
coefficient as well as the viscosity relative to fluids of known viscosity. The
flow through the capillary may be induced by an externally applied pressure
gradient or by gravitational forces. There are several precautions that need
to be observed in using these instruments. The derivation of the relationship
given in Equation 4.3 included several assumptions, such as laminar
7328_C004.fm Page 119 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 119

(streamlined), steady, fully developed flow, and that the fluid is Newtonian.
To avoid the effects of turbulent flow in deviating from the derived relation-
ship, experiments must be performed at lower Reynolds numbers where
streamlined flow will be assured. The assumptions of steady and fully devel-
oped flow imply that the velocity profile within the capillary is parabolic and
no acceleration of the fluid is occurring within the viscometer. In most types
of capillary viscometers, part of the applied pressure is used to induce kinetic
energy to the fluid. Moreover, a small amount of energy is also expended in
overcoming the viscous forces at the converging and diverging streamlines
at the entrance and exit to the capillary. Kinetic energy and Couette correc-
tions, respectively, need to be taken into account to correct for the errors due
to these two effects and are detailed in Dinsdale and Moore (1962).
In viscometers with externally applied pressure, the effective pressure
gradient includes the externally applied pressure gradient as well as the
hydrostatic head of the fluid in the viscometer. Hydrostatic corrections are
also needed to account for the constantly decreasing hydrostatic head (Dins-
dale and Moore, 1962). Use of this device on Newtonian fluids is not only
required for the Poiseuille relation to hold, but it is also essential due to the
variable shear rate within the tube. Differentiation of the parabolic velocity
profile shows that the rate of shear is not constant, but is a function of radial
position, r. Thus, a fluid particle near the wall would be exposed to a high
shear rate while one at the tube center would experience zero shear rate. If
the fluid viscosity were shear rate dependent, i.e., µ( γ
) , then a range of
viscosities would be seen and no meaningful value would be obtained. The
relationships between pressure gradient and flow rate for non-Newtonian
fluids in straight circular tubes are discussed later in the chapter.

4.1.1.2 Coaxial Cylinder Viscometer

Rotational viscometers are particularly useful in analyzing the rheological
characteristics of non-Newtonian fluids because one sample of fluid can be
used to study the viscous behavior at different rates of shear or when con-
tinuous measurements are required to study the effect of time-varying
conditions on the flow properties. A simple coaxial cylinder rotational vis-
cometer is shown schematically in Figure 4.2.
In the case of the coaxial cylinder viscometer, the inner cylinder, also
referred to as the “bob,” with a radius R1 and height h remains stationary.
The outer cylinder, also referred to as the “cup,” with a radius R2 , which
contains the test fluid, is rotated at a constant speed, Ω ( rad /sec ), and the
resultant torque, T (dyne cm), is measured by the angular deflection of the
inner cylinder, which is suspended by fine wire. It is assumed that the flow
of the fluid between the cylinders is steady and laminar and that the end
effects are negligible.
Fluid viscosity measurements are derived from the following analysis.
Consider a volume of fluid between the stationary inner cylinder and an
7328_C004.fm Page 120 Friday, October 6, 2006 12:25 PM

120 Biofluid Mechanics: The Human Circulation

Torsion wire

Graduated scale

bob
Cup

h R
R2

FIGURE 4.2
Schematic of a coaxial (Couette) cylinder viscometer.

arbitrary radius r. If ω is the angular velocity of the fluid at radius r, then

the torque exerted on the fluid at this radius is:

T = Ft r

or

T = τ r (2 πrh)r

The torque, T, measured at the surface of the inner cylinder must be the
same as the torque at any arbitrary radius r since the motion is steady.
Rewriting in terms of the shear stress, we get

T
τr = (4.4)
2 πr 2 h

It can be observed from the above relationship that the shearing stress is
inversely proportional to the square of the distance from the axis of rotation.
It is interesting to note, however, that by using an annular gap that is small
compared to the cylinder radii (i.e., dr << R1, R2), the shearing stress on the
fluid will be almost constant throughout the volume of the testing fluid and,
thus, approaches the case of Couette flow. (Couette flow is a classic config-
uration of flow between two infinitely large flat plates where either one plate
is in motion [as in this device] or a pressure gradient is present or both.)
7328_C004.fm Page 121 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 121

The velocity of the fluid at the radius r is:

V = rω

and, thus, the gradient of velocity (or rate of shear) will be given by

dV dω
=r +ω
dr dr

In this expression, the first term on the right-hand side represents the radial
velocity gradient and the second term is the angular velocity due to rigid
body rotation. Since we are interested in the tangential velocity gradient in
the fluid, the second term with the constant angular velocity can be
neglected. For a Newtonian fluid, τ r = µγ
, or

dV dω
τ r = µγ
= µ = µr
dr dr

Combining this with the expression for τr in Equation 4.4 results in

T dω
= µr
2 πr 2 h dr

Consequently,

dω T
= (4.5)
dr 2 πµhr 3

Integrating Equation 4.5 with appropriate boundary conditions results in

the expression

 4 πhR 2 R 2 
T =  2 1 22  µΩ (4.6)
 ( R2 − R1 ) 

or T = CµΩ , where C is the instrumentation geometric constant. The same

expression will be applicable if the inner cylinder is rotated while the outer
cylinder is held stationary. It is also applicable if the torque is applied exter-
nally and the resulting rotation of the cylinder is measured. It is to be noted
that the relationship derived in Equation 4.6 ignores the effects due to forces
on the top and bottom free surfaces of the space between the cylinders.
Correction due to the “end effects” must be taken into account especially in
fluids with a viscosity lower than 1 P. Various attempts to minimize the
errors due to end effects are discussed in Dinsdale and Moore (1962).
7328_C004.fm Page 122 Friday, October 6, 2006 12:25 PM

122 Biofluid Mechanics: The Human Circulation

s
ψ h
r

FIGURE 4.3
Schematic of a cone and plate viscometer.

4.1.1.3 Cone and Plate Viscometer

In the case of the cone and plate viscometer (Figure 4.3), the fluid sample is
contained between a cone of large apical angle and a flat surface (plate)
normal to its axis. If the angle between the cone and the plate, ψ, is small,
the rate of shear is essentially constant, as shown below.
Let Ω be the angular velocity at which the cone is rotated. The velocity of
the fluid at any radial distance r is then

V = Ωr

The rate of shear will be given by the ratio of the linear velocity and the
gap between the cone and plate h at that radius. From the figure, it can be
observed that h = r tan ψ.
Ωr Ω
Thus, the shear rate γ
= r tan ψ = ψ , provided that the cone angle ψ is small.
Furthermore, the shearing stress is also independent of the radius in this
configuration.
The total torque, T, can be obtained by the expression,

T= ∫ rτ dA
A
r

where dA = 2πrdS. However, from the figure, it can be observed that

r = s cos ψ . Therefore, ds = cosdrψ or ds = dr if ψ is small.
Hence,

∫
T = τ r 2 π r 2 dr
0

and, thus,

3T
τr =
2 πR 3
7328_C004.fm Page 123 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 123

For a Newtonian fluid,

τr 3T ψ
µ= = (4.7)
γ
2 πR 3Ω

In summary, the capillary tube viscometers require only a small volume

of fluid, are easy to use, and the results are reproducible. However, for non-
Newtonian fluids, measurements must be made over a wide range of rates
of shear, which are uniformly applied to the entire fluid. Coaxial cylinder,
or Couette, viscometers, therefore, are more appropriate because they subject
the fluid sample to a relatively constant rate of shear and multiple measure-
ments at various rates of shear can be made with a given blood sample.
However, the end effects of these devices discussed earlier will introduce
large errors particularly with non-Newtonian fluids and the instruments
are expensive. The experiments are also more complicated to perform com-
pared with the tube viscometers. The cone and plate viscometers have the
same advantages as those of the Couette viscometers and need a smaller
sample of blood for measurements. Although tube viscometers have been
used extensively in the past to study the rheology of blood, Couette and
cone and plate viscometers are widely used because of their greater accuracy
and flexibility.

4.1.2 Physical Properties of Blood

Whole blood consists of formed elements that are suspended in plasma. The
plasma is a dilute electrolyte solution (0.15N) containing about 8% by weight
of three major types of proteins — fibrinogen, globulin, and albumin in
water. When blood clots in the absence of an anticoagulant, the fibrinogen
polymerizes into fibrin. If the clot of fibrin and blood cells are left alone, the
clot will contract. The portion of the blood without the fibrin is called the
serum. Globulin, which may be subdivided into groups, is a carrier of lipids
and other water-soluble substances. Globulin also contains antibodies that
resist infection from bacteria and viruses. Albumin, with the lowest molec-
ular weight of the three proteins, is the main contributor to the total colloid
osmotic pressure of the plasma proteins and is important in the balance of
the water metabolism. Plasma is identical to the interstitial fluid in compo-
sition except for the presence of plasma proteins. The higher osmotic pressure
due to the presence of proteins in the plasma results in the absorption of
water from the interstitial fluid into the capillary. When the whole blood is
centrifuged with suitable anticoagulants added to prevent clotting, the
formed elements will settle to the bottom of the tube. About 45% by volume
will consist of formed elements and about 55% will consist of plasma in the
normal human blood. The formed elements in blood consist of 95% red
blood cells, 0.13% white blood cells, and about 4.9% platelets. The white
blood cells, also known as leucocytes, consist of monocytes, lymphocytes,
7328_C004.fm Page 124 Friday, October 6, 2006 12:25 PM

124 Biofluid Mechanics: The Human Circulation

neutrophils, eosinophils, and basophils. Monocytes that leave the circulation

and enter the tissues develop into macrophages. Neutrophils, monocytes,
and macrophages are collectively known as phagocytes because they can
engulf and ingest bacteria and other foreign particles. Platelets, which are
smaller than red or white blood cells, are anucleated cell fragments produced
in the bone marrow from huge cells known as megakaryocytes.
The average life of a platelet is about 10 days and the inactivated platelets
are discoid in shape. The cytoplasm of platelets contains organelles (a cyto-
skeleton consisting of microtubules and microfilaments) and granules. The
surface membrane expresses various glycoprotein molecules. The platelets
can be activated by prolonged exposure to high shear stress or by rapid
increases in shear stress. Platelets can also stick to exposed collagen (as a
result of vascular injury) and become activated. Upon activation, changes
occur in the internal cytoskeleton, which causes the platelets to become
spheroid in shape and to release the contents of their internal granules, such
as serotonin, ADP (adenosine diphosphate), and platelet-activating factors
(PAF). The release of PAF can cause additional platelets to become activated,
which promote platelet aggregation and collagen-induced thrombus forma-
tion. The initiation of the coagulation cascade results in fibrinogen being
converted to fibrin and then fibrin stabilizing the adhered platelets to form
a blood clot.
The majority of the formed elements are red blood cells and the measure
of the volume percent of formed elements in blood is referred to as the
hematocrit. The typical shape of a red blood cell is shown in Figure 4.4a and
an actual human red blood cell as viewed through an electron microscope
is shown in Figure 4.4b. The biconcave-shaped red blood cell consists of
hemoglobin surrounded by a flexible red cell membrane. The primary func-
tion of hemoglobin in the red blood cell is to transport oxygen from the
lungs to the living tissues of the body. The diameter of the red blood cell
is about 8.5µm with transverse dimensions of 2.5µm at the thickest portion
and about 1µm at the thinnest portion. The volume of the intracellular fluid
is about 87µm3 and the surface area of the cell is about 163µm2. It can be
easily shown that a spherical-shaped red cell with the same intracellular
fluid volume will have about 40% smaller surface area compared to the
biconcave shape. Thus, the advantage of the biconcave shape with a larger
surface area for efficient gas exchange in the capillaries can be noted. The
membrane is flexible and the cell can pass through capillaries with diameters
as small as 5µm by assuming a bent shape.

4.1.3 Viscous Behavior of Blood

4.1.3.1 Plasma
The viscosity of plasma can be measured by obtaining a sample of blood
and separating out the formed elements by centrifuging. However, as soon
as blood is removed from the body, the clotting mechanism is initiated and
7328_C004.fm Page 125 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 125

8.5 µ

1.0 µ 2.5 µ

(a)

(b)

FIGURE 4.4
(a) Schematic of the biconcave shape of the red blood cell; (b) a scanning electron microscope
image of a human red blood cell magnified × 13,000.

an anticoagulant must be added to prevent clotting. Typical anticoagulants

are heparin, sequestrate, oxalate, or citrate in isotonic concentration. The
addition of the anticoagulant is generally assumed to reduce the actual blood
or plasma viscosity even though the specific effect of the various compounds
is not clearly established.
Human plasma has a density of about 1.035 gm/cc and its viscosity coef-
ficient ranges between 0.011 and 0.016 Poise compared to 1.0 gm/cc and 0.01
Poise, respectively, for water. The presence of plasma proteins results in the
higher viscosity compared to that of water. Experiments using capillary as
well as rotating viscometers have suggested that mammalian plasma
behaves like a non-Newtonian fluid, while other published works have
suggested that plasma is a Newtonian liquid (Whitmore, 1968). Variability in
the protein concentrations in the samples within the species, across species,
and also in the experimental techniques may be contributing factors for such
contradictory results. The rheological characteristics of plasma in the presence
of pathological conditions may also contribute to non-Newtonian behavior.
Temperature also has a significant influence on the plasma viscosity causing
7328_C004.fm Page 126 Friday, October 6, 2006 12:25 PM

126 Biofluid Mechanics: The Human Circulation

it to decrease with an increase in temperature. A fall of 2 to 3% in the viscosity

coefficient per °C rise in temperature has been reported in the range of
25 to 37 °C. For our purposes, we will assume that the plasma behaves as a
Newtonian fluid with a constant viscosity at 37 °C, which is the human body
temperature. We will, for modeling the flows in the arterial system, assume
that the plasma viscosity is 1.2 cP (centiPoise).

4.1.3.2 Whole Blood

Samples of whole blood, with the addition of suitable anticoagulants, have
been tested in capillary, co-axial cylinder, or cone and plate viscometers over
a wide range of rates of shear. A description of the instruments used, test
samples taken, and experimental conditions for measuring the viscosity of
blood from some of the published works are included in Table 4.1. In order
to identify the constitutive relationship for whole blood, a plot of shear stress
vs. rate of shear can be obtained from the data. However, as pointed out
before, the viscosity of blood and plasma also differs from sample to sample
due to the variations in species and constituents such as protein content. To
avoid difficulties in interpreting the data due to the variability in the blood
samples, the shear stress is normalized with respect to the plasma viscosity
of the sample. A plot of shear stress/plasma viscosity and rate of shear data
is shown in Figure 4.5. As can be seen, the data suggest a nonlinear behavior,
particularly at low shear rates. It can also be noted that the best-fit curve
does not pass through the origin and exhibits a positive y– intercept indi-
cating that the induced shear stress must exceed a minimum yield stress
before blood will flow.
If the constitutive relationship for whole blood follows the power
law (τ = K pl γ
n : n ≠ 1) , then the data can be represented as a straight line in

TABLE 4.1
Summary of Selected Studies on Blood Viscosity Measurement Reported
in Literature of the 1960s
Viscometer Blood Anticoag. HCT Temp. Comments
Coaxial cylinder Human ACD 0–40% 20–40°C 6–200 sec1
Capillary Human Heparin 12.5–70% 25–32°C 12.8–335 sec1
power law
Cone and plate Dog Heparin 37°C 11.45–230 sec1
Cone and plate Human ACD Normal 37°C 2–100,000 sec1
Plate–coaxial Casson’s Eqn.
Coaxial cylinder Human Heparin 0–95% 37°C 0.052–52
Coaxial Human Heparin 43–48% 37°C 23–230 sec1
Cone and plate Human None/Heparin 33–57% 37°C 10–250 sec1
Oxalate
Coaxial cylinder Human ACD 42.5% 25–37°C 0.1–20 sec1
Cone and plate Human Heparin 0.80% 22–37°C 2.12 sec1
7328_C004.fm Page 127 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 127

800

Shear stress/plasma viscosity µ (sec–1)

p
␶

600

400

200

0
50 100 150 200 250
◊
Rate of shear γ (sec–1)

FIGURE 4.5
Shear stress normalized to the plasma viscosity vs. rate of shear plot from experimental data.
(Redrawn from Whitmore, R.L. (1968) Rheology of Circulation, Pergamon Press, New York. With
permission.)

the logarithmic shear stress vs. rate of shear plot (Figure 4.6). Even though
the data points exhibit a linear behavior over small ranges of shear rates,
a curvilinear relationship is observed over the entire range of shear rates
for which data have been obtained. Tests conducted with rotational vis-
cometers for shear rates ranging from 5 to 200 sec−1 follow the Power Law

1000
Shear stress/plasma viscosity µ (sec–1)
p
␶

100

n = 0.75
10
n = 1.0
(Newtonian)

1
0.01 0.1 1 10 100 1000
◊
Rate of shear g (sec–1)

FIGURE 4.6
Shear stress vs. rate of shear plot in logarithmic scale. (Redrawn from Whitmore, R.L. (1968)
Rheology of Circulation, Pergamon Press, New York. With permission.)
7328_C004.fm Page 128 Friday, October 6, 2006 12:25 PM

128 Biofluid Mechanics: The Human Circulation

30

␶ (sec–1/2)
µp

20
Shear stress/plasma viscosity

10

0 5 10 15
Rate of shear γ◊ (sec –1/2)

FIGURE 4.7
Square root of shear stress adjusted to plasma viscosity vs. square root of shear rate from
viscometric data. (Redrawn from Whitmore, R. L. (1968) Rheology of Circulation, Pergamon Press,
New York. With permission.)

reasonably well for values of the exponent n between 0.68 and 0.8. Values
of n exceeding 0.9 have been reported from results obtained with capillary
viscometers.
Inspection of Figure 4.5 once again reveals that even though blood exhibits
a yield stress (y– intercept on the plot), the data does not suggest a linear
relationship between shear stress and rate of shear particularly at low shear
rates, a relationship indicative of an ideal Bingham plastic (τ = τ y + Kb γ
). As
an alternative, we could consider the constitutive relationship for the
Casson’s fluid ( τ = τ y + K c γ
) where there is a linear relationship between
the square root of shear stress vs. the square root of rate of shear. A plot of
such a curve together with the data obtained for whole blood is shown in
Figure 4.7.
A least square fit of the data yields the relationship

τ
= 1.53 γ
+ 2.0 (4.8)
µp

The apparent viscosity coefficient for whole blood at any rate of shear,
thus, can be computed from the above relationship. For example, at a rate
of shear of 230 sec−1, the apparent viscosity of whole blood is about 3.3 cP
using a value of 1.2 cP for plasma viscosity. This value compares favorably
7328_C004.fm Page 129 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 129

0.6

Apparent viscosity µapp (poise)

0.4

0.2
Experimental data

Asymptotic
Casson values
equation
(4.8)
0
0.1 1 10 100 1000
◊
Rate of shear γ (sec–1)

FIGURE 4.8
A plot of apparent viscosity as a function of rate of shear. (Redrawn from Whitmore, R.L. (1968)
Rheology of Circulation, Pergamon Press, New York. With permission.)

with the experimentally determined values ranging from 3.01 to 5.53 cP.
Apparent viscosity data from experiments by various investigators for a
range of shear rates is shown in Figure 4.8. A plot of the constitutive rela-
tionship in Equation 4.8 is also included in the figure. As can be observed,
the apparent viscosity increases to relatively large magnitudes at low rates
of shear. This is because the red blood cells tend to aggregate into Rouleaux
formations and, thus, exhibit an increase in viscosity at low rates of shear.
Aggregates of red blood cells are shown in Figure 4.9. As the rate of shear
increases, the aggregates gradually break up if the rate of shear exceeds about
50 sec−1, and the viscosity coefficient approaches an asymptotic value of about
3.5 cP. Thus, when specifying the viscosity coefficient of blood, it is important
to specify the rate of shear at which the measurements were obtained.
As discussed earlier, apart from the fact that the rate of shear has a pro-
found effect on the measured apparent viscosity of whole blood, the vari-
ability of samples, the effect of the addition of anticoagulants, and the
experimental conditions must also be taken into account. Let us now look
into the effect of other variables on the viscosity coefficient of blood.
7328_C004.fm Page 130 Friday, October 6, 2006 12:25 PM

130 Biofluid Mechanics: The Human Circulation

FIGURE 4.9
A scanning electron microscope view of aggregation of red blood cells magnified × 2000.

4.1.3.3 Effect of Hematocrit

The hematocrit for normal human blood varies between 40 and 45%. To study
the effect of hematocrit on blood viscosity, the red blood cells are separated
out by centrifuging and reconstituted in appropriate proportions with serum,
saline, or Ringer’s solution before viscometry experiments. A plot of the
relative viscosity of blood (with respect to plasma) as a function of hematocrit
is shown in Figure 4.10a. (Note that the relative viscosity is plotted in loga-
rithmic scale in this figure.) The viscosity increases with increasing hematocrit
and also decreases with increasing rates of shear as observed earlier. The
viscosity of rigid spheres suspended in a solution is seen to rise faster than
that of blood, thus indicating the effect red blood cell flexibility has in mini-
mizing the viscosity. Figure 4.10b compares the viscosity as a function of
hematocrit for normal cells, rigid spheres, and red cells, which have been
hardened by treating in an aldehyde solution. Once again, the effect of red
cell flexibility on reducing the viscosity is clearly demonstrated.

4.1.3.4 Effect of Temperature

As pointed out earlier for the case of plasma, the effect of temperature on
the viscosity of blood is still not clearly established, so it is the usual practice
to keep the temperature at a constant value (preferably at body temperature)
when measurements are obtained.
7328_C004.fm Page 131 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 131

10

Spheres
8
52 sec–1
42 sec–1

6 64 sec–1
×

212 sec–1

×
Relative viscosity µr

3
×

2
×
×

1
0 20 40 60 80
Hematocrit H
(a)

FIGURE 4.10
(a) Effect of hematocrit on the relative viscosity of blood; (b) Effect of hardening of the red
blood cells on the relative viscosity. (Redrawn from Whitmore, R.L. (1968) Rheology of Circulation,
Pergamon Press, New York. With permission.)

4.1.3.5 Effect of Protein Content in Plasma

Figure 4.11 shows the plot of viscosity vs. rate of shear (at lower shear rates)
for whole blood as well as for red blood cells suspended in isotonic solution
with one protein of physiological concentration included. The results are
presented for 37°C and a hematocrit of 49%. It can be observed that globulin
appears to have the dominant effect on increasing the viscosity of blood,
whereas albumin moderates the rise in viscosity. Since the concentration
level of fibrinogen is very low, its effect on viscosity is minimal.

4.1.3.6 Yield Stress for Blood

A review of the Casson’s relationship (Figure 4.7) shows that the curve
exhibits a y– intercept, which is the square root of the yield stress. Studies
7328_C004.fm Page 132 Friday, October 6, 2006 12:25 PM

132 Biofluid Mechanics: The Human Circulation

10
Cells
Spheres hardened
8 normal

Relative viscosity µr 6

1
0 20 40 60 80
Hematocrit H
(b)

FIGURE 4.10 (CONTINUED).

have shown that the yield stress depends strongly on fibrinogen concentra-
tion and is also dependent on the hematocrit (Figure 4.12 a and Figure 4.12b).
Several empirical formulae relating the yield stress to hematocrit and
fibrinogen concentration in blood have been reported in the literature
(Cooney, 1976). An empirical relationship for the yield stress as a function
of hematocrit can be given by

1 A ( H − Hm )
τy3 = (4.9)
100

where A is a constant {(0.008 + 0.002 dynes/cm2)1/3}, H is the hematocrit

and Hm is the hematocrit below which yield stress vanishes (5 to 8%).
Assuming H = 45% and Hm = 5%, the yield stress values for normal human
7328_C004.fm Page 133 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 133

40

Temp = 37°C
Hematocrit = 49%

Viscosity, cP 30

20

14
12 Globulin 2.2 wt%

10
8 Whole blood
6 Albumin 3.5 wt%

4 Fibrinogen 0.6 wt%

2

0 10 20 40 60 80 100 120
Shear rate, sec–1

FIGURE 4.11
Effect of plasma proteins on the viscosity of whole blood. (Redrawn from Copley, A.L. and
Stainsby, G., Eds. (1960) Flow Properties of Blood, Pergamon Press, Oxford. With permission.)

blood should be between 0.01 to 0.06 dynes/cm2. Another empirical for-

mula for the yield stress as a function of hematocrit and fibrinogen
concentration is:

τ y = ( H − 0.1)(CF + 0.5) (4.10)

In this relationship, CF is the fibrinogen content in grams per 100 ml and

H is the hematocrit expressed as a fraction. This correlation fits the experi-
mental data satisfactorily and is valid for 0.21 < CF < 0.46 and H > 0.1. As
pointed out earlier, the yield stress appears to vanish below a hematocrit
fraction of 0.1.
An estimate can be obtained on the effect of yield stress on blood flow in
capillaries by considering the pressure gradient required to maintain steady
flow of blood in cylindrical tubes. Assuming Poiseuille flow for this situation,
the shear stress will vary linearly with the radius having a magnitude of
7328_C004.fm Page 134 Friday, October 6, 2006 12:25 PM

134 Biofluid Mechanics: The Human Circulation

0.5
Hct
25%
30%
0.4 35%
40%

(dyn/cm2)
× 45% ×
××
0.3 ××
× ×
×
×

0.2
τy,

0.1

0
0 0.1 0.2 0.3 0.4 0.5
Fibrinogen conc., g/100 ml
(a)

0.3
(dyn/cm2)

0.2

0.1
τy,

0
25 35 45
Hematocrit (%)
(b)

FIGURE 4.12
Effect of (a) fibrinogen concentration and (b) hematocrit on the yield stress for blood. (Redrawn
from Merrill et al. (1965), The Casson equation and rheology of blood near zero shear, in
Proceedings of the Fourth International Congress on Rheology, Copley, A.L. (Ed.). With permission.)

zero at the center and a maximum at the wall. To initiate flow, the imposed
pressure drop must be large enough so that the wall shear stress,

R∆p
τw =
2L

will exceed the yield stress for blood. Equating the wall shear stress to the
value for the yield stress for blood (0.07 dynes/cm2), the pressure gradient
7328_C004.fm Page 135 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 135

4.5

4.0
Apparent viscosity, cP

3.5

3.0

2.5
0 0.5 1.0 1.5 2.0 2.5
Tube radius, mm

FIGURE 4.13
Plot of apparent viscosity as a function of tube radius.

required to initiate flow can be computed. Assuming a length of 20 cm for

a typical capillary viscometer and a radius of 0.1 cm, the required pressure
drop can be computed as 0.021 mmHg. Similarly, for a typical systemic
capillary with a length of 1 mm and a radius of 5µm, the pressure drop
required is computed to be the same value. In the normal circulation, the
magnitude of pressure drop across the systemic capillaries is about 16 mmHg
and, hence, the effect of yield stress for blood appears to be negligible under
normal flow conditions.

4.1.3.7 Effect of Tube Diameter

A plot of the apparent viscosity of blood as a function of tube radius at a
hematocrit of 40% is shown in Figure 4.13. As can be observed, the apparent
viscosity for blood has a very low value in very small diameter tubes. The
viscosity increases with the increase in tube diameter and approaches an
asymptotic value at tube diameters larger than about 0.5 mm. This phenom-
enon of a decrease in the apparent viscosity in small diameter tubes is
referred to as the Fahraeus–Lindqvist effect. This effect can be explained on
a physical basis as detailed below. As the blood flows through a tube, the
blood cells tend to rotate as shown in Figure 4.14. Due to their spinning, red
blood cells tend to move toward the center of the tube and, hence, a cell-
free layer, referred to as the plasma-skimming layer, exists near the wall. In
tubes with small diameters, the cross-sectional area of the cell-free zone is
comparable to the central core. Thus, the net effect of the cell-free zone with
a lower viscosity (that of plasma alone) is to reduce the apparent viscosity
of flow through the tube. As the tube diameter increases, the effect of the
7328_C004.fm Page 136 Friday, October 6, 2006 12:25 PM

136 Biofluid Mechanics: The Human Circulation

Axis of tube

1
V1 Bernoulli force

V2
2

Tube wall
Velocity profile

FIGURE 4.14
Schematic describing the rotation of red blood cells under the influence of a velocity gradient.

cell-free zone reduces and, hence, the viscosity coefficient approaches an

asymptotic value.
Two mathematical models have been developed to describe the Fahraeus–
Lindqvist effect and they are described below.

4.1.3.7.1 Cell-Free Marginal Layer Model

Consider steady flow of blood through a circular tube of radius a as shown
in Figure 4.15. The tube cross section can be divided into a core region and
a cell-free plasma region near the wall. The governing equations for both
regions can be given by the following equations:

∆p 1 d  du 
− = µ r c 0≤r ≤R−δ
L r dr  c dr 
(4.11)

∆p 1 d  dup 
− = µ r  R−δ ≤r ≤R (4.12)
L r dr  p dr 

where µ c and µ p represent the viscosities of the core and periphery, respec-
tively, while uc and up represent the velocities of the core and periphery,
respectively. The boundary conditions used to obtain a solution for the two
differential equations are (1) the velocity gradient is zero at the center of the

a a–δ
Core
region

Cell free zone

FIGURE 4.15
Schematic of a cell-free zone layer model.
7328_C004.fm Page 137 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 137

tube, (2) no slip occurs at the tube wall, and (3) the velocity and the shear
stress is continuous at the interface between the two zones. The equations
can be integrated with the above boundary conditions to solve for the veloc-
ity components and to obtain an expression for the flow rate as given below:

π R 4 ∆p   δ  µ 
4
Q= 1 −  1 −   1 − p   (4.13)
8µ p L   R  µc  
 

By comparison, the expression for the flow rate for a homogeneous fluid
(Poiseuille’s Equation) was derived earlier as

π R 4 ∆p
Q=
8µ L

Comparing these two equations, if Poiseuille’s law was applied to calculate

the viscosity of blood, the apparent viscosity could be written as

µ p
µ app =
)( )
(4.14)

(
1 − 1 −
δ 4
R 1−
µp
µc



In the limit as δ R → 0, µ app → µ c, as would be expected. When the ratio δ R is

small, the higher order terms containing the ratio can be neglected and the
relationship for the apparent viscosity reduces to

µc
µ app =  
(4.15)
1 + 4 Rδ  µµc −1
 p 

4.1.3.7.2 The Sigma Effect

In deriving the Poiseuille formula, the homogeneous fluid was assumed to
be a continuum, allowing integration to be performed to obtain the flow rate
from the expression for the velocity. The Sigma effect theory, however, states
that when blood flows through a small diameter tube, the assumption of a
continuum is not valid. For example, assume that the tube diameter is so
small that there is only room for five red blood cells to move abreast. Then,
the ensuing velocity profile would not be continuous and would consist of
concentric laminae moving axially as demonstrated in Figure 4.16.
An apparent viscosity can be derived for this case by the following anal-
ysis. The expression for flow rate from the Poiseuille solution

∫
Q = 2 π u(r )rdr
0
7328_C004.fm Page 138 Friday, October 6, 2006 12:25 PM

138 Biofluid Mechanics: The Human Circulation

Umax
Umax

Parabolic velocity profile

FIGURE 4.16
Concentric layer model for flow in small tubes.

can be rewritten as

R R

∫ ∫
du
Q = π d[u(r )r ] − π r 2
2 dr
dr
0 0

Applying the “no slip” condition at the wall, the first integral on the right
side will be identically equal to zero. From the parabolic velocity profile
characteristic of Poiseuille flow, the expression for the velocity gradient will
be given by

du ∆pr
=−
dr 2µ L

Substituting this velocity gradient into the flow integral, we obtain

R
π ∆p 3
Q=
2µ L
r dr ∫
0
(4.16)

Now, if we assume that flow occurs in N concentric laminae, each of

thickness ε, then the radius R of the tube will be equal to N ε and any general
radius r can be replaced by n ε , where n = 1,2 ... N. If r = n ε , then dr = ε∆n
= ε since ∆ n is equal to unity.
Hence, the integration in Equation 4.16 can be replaced by a summation
as

N
π ∆p
Q=
2µ L ∑ (nε) ε
n= 1
3 (4.17)
7328_C004.fm Page 139 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 139

Employing the mathematical identity

N
N 2 ( N + 1)2
∑n
n= 1
3 =
4

one can show that

N N
N 2 ( N + 1)2
∑n= 1
(nε)3ε = ε 4 ∑n
n= 1
3 = ε4
4

Thus, the flow rate can be written as

2
π ∆pR 4  ε
Q= 1+  (4.18)
8µ L  R

and the apparent viscosity can be given by the formula

µ
µ app = (4.19)
 ε 2
 1+ R 

In this relationship, as ε R → 0, the expression reduces to that for Poiseuille

flow. It can also be seen that the Sigma effect theory and the cell-free zone
theory are equivalent if

εµ p
δ= (4.20)
2(µ c − µ p )

Even though the Fahraeus–Lindqvist effect is not important in large arteries

from the viscosity coefficient point of view, the cell-free zone is still present.
If blood samples are obtained from large arteries at the wall, the measured
hematocrit will be lower than the actual value due to the presence of the
plasma-skimming layer.

4.1.4 Pressure-Flow Relationship for Non-Newtonian Fluids

For any fluid flowing through a circular tube, the force balance between the
pressure gradient and the viscous drag forces yields the relationship

( ∆p)r ∆pR
τw = = (4.2)
2L r=R 2L
7328_C004.fm Page 140 Friday, October 6, 2006 12:25 PM

140 Biofluid Mechanics: The Human Circulation

The expression for the velocity profile and the flow rate for steady flow of
a Newtonian fluid through a tube was derived in Chapter 1, section 1.8: Fluid
Mechanics in a Straight Tube, Equation 1.70 and Equation 1.72, respectively,
as well as in the description of the capillary viscometer in this chapter. We
will now derive the corresponding relationships for fluids that follow the
Power Law, Bingham Plastic, and Casson’s constitutive relationships.

4.1.4.1 Power Law Fluid

The constitutive relationship for a Power Law fluid is given by

n
 du 
τ = K pl γ
n = K pl   (4.21)
 dr 

Substituting this term for shear stress yields

n
 du  r ∆p
K pl   = −
 dr  2L

Rewriting

1
 ∆p  n

∫ ∫r
1
du = −   n dr
 2 K pl L 

and integrating, we get

1
 ∆p  n
n n+ 1 n
u = −  r +C
 2 K pl L  n+1

Employing the boundary condition that u = 0 at r = R, we have

1
 ∆p  n
n n+ 1
C=  R n
 2 K pl L  n+1

Substituting this value for C, the expression for the velocity profile becomes

1
n  ∆p  n
 R n+ 1 n − r n+ 1 n 
u=
n + 1  2 K pl L 
(4.22)
 
7328_C004.fm Page 141 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 141

Finally, the flow rate, Q, through the tube is given by

Q= ∫ 2πrudr
0

n  ∆p 
1
n  R n+ 1 R 
= 2π
n + 1  2 K pl L  
0
∫
 R nrdr − r n+1 nrdr 
0


∫

n  ∆p 
1
n  R n+ 1 a 
= 2π
n + 1  2 K pl L  
0
∫
 R nrdr − r 2 n+1 ndr 
0


∫
1
  n  R 3 n+ 1 R
= 2π
n  ∆p   R n+ 1 n r 2 − r n 
n + 1  2K L   2 3 n+ 1
n

 pl   0 0


1
n  ∆p  n  R 3 n+ 1 n R 3 n+ 1 n 
= 2π  − 3 n+ 1 
n + 1  2 K pl L   2 n 
 

1
 n+ 1 R 3 n+ 1 n 
n  ∆p  n
 n 
= 2π
n + 1  2 K pl L   2  3 n+1 
  n  

which can be reduced to an expression for the flow rate as

1
nπ R 3  R∆p  n
Q=
3n + 1  2 K pl L 
(4.23)

The flow distribution in the tube given by Equation 4.23 depends upon
the value for n. The value for n for human whole blood can be approximated
to ~ 0.75 over the range of rates of shear from 10 to 1000 1/s.

4.1.4.2 Bingham Plastic

For the Bingham plastic, the constitutive relationship will be given by

 du 
τ = τ y + Kb γ
= τ y − Kb   (4.24)
 dr 
7328_C004.fm Page 142 Friday, October 6, 2006 12:25 PM

142 Biofluid Mechanics: The Human Circulation

where τ y is the yield stress and Kb is the viscosity coefficient for the Bingham
plastic. Again, substituting τ into Equation 4.2, we get

du r ∆p τ y
=− + (4.25)
dr 2 Kb L Kb

or

∆p τy
du = − rdr + dr
2 Kb L Kb

Integrating, we obtain

∆p r 2 τ y
u=− + r+C
2 Kb L 2 Kb

By applying the boundary condition that u = 0 at r = R, we can evaluate

the constant

∆p R 2 τ y
C= − R
2 Kb L 2 Kb

The velocity profile is then given by

∆p r 2 τ y r ∆pR 2 τ y R
u=− + + −
2 K b L 2 K b 4K b L µ b

Rearranging terms

∆p τy
u= (R 2 − r 2 ) − (R − r ) (4.26)
4K b L Kb

From the expression for the wall shear stress at the inner surface of the
tube (r = R) given by

R∆p
τw =
2L

until the shear stress at the wall, τ w , exceeds the Bingham Plastic yield stress,
there is no flow possible through the tube. However, once the wall shear
7328_C004.fm Page 143 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 143

stress exceeds the yield stress, flow is induced through the tube. We can also
determine a radius Rc at which the shear stress equals the yield stress

2 Lτ y
Rc =
∆p

The fluid will have a constant velocity uc between the center of the tube
and Rc and the velocity profile will follow Equation 4.26 from Rc to R. Using
Equation 4.26, the core velocity can be written as

∆p τy
uc =
4K b L
( R 2 − Rc 2 ) −
Kb
(R − Rc )

Rc ∆p
Substituting τ y = 2L , the core velocity is:

∆p
uc =
4K b L
( R 2 − Rc 2 ) − R2c ∆Lp K1 (R − Rc ) (4.27)
b

The flow rate through the tube is equal to the sum of the flow through the
core and the peripheral regions:

Q = Qcore + Qper

where

Qcore = uc π Rc 2

Thus,

∆pπ  ( R 2 − Rc 2 ) Rc 2 Rc 3 
Qcore =  − (R − Rc )  (4.28)
Kb L  4 2 

Flow in the peripheral region is given by

Qper = ∫ 2πrudr
Rc
7328_C004.fm Page 144 Friday, October 6, 2006 12:25 PM

144 Biofluid Mechanics: The Human Circulation

Substituting for τ y in terms of ∆p, the velocity in the periphery will be

expressed as

∆p  ( R 2 − r 2 ) Rc 
uper =  − ( R − r )
Kb L  4 2 

and the peripheral flow rate can be expanded as

 ( R 2 − r 2 ) Rc
R
∆p 
Qper =
Kb L ∫ 
Rc
4
− ( R − r ) 2πrdr
2 

Integrating

π∆p  R 4 R 2 Rc 2 5 4 Rc R 3 RRc 3 
Qper = − − R − +
Kb L  8 2 
(4.29)
4 24 c 6

Adding equations 4.28 and 4.29

π∆p  R 4 Rc 4 Rc R 3 
Q= + −
Kb L  8 6 
(4.30)
24

and substituting for Rc as

2τ y
Rc = ∆p
L

The relationship for the flow is then given by

 4
πR 4 ∆p  4  2 τ y  1  2 τ y  
Q= 1 −  ∆p  +  ∆p  (4.31)
8K b L  3  R ( L )  3  R ( L )  
 

If the yield stress becomes zero, the flow relationship will reduce to that
of Poiseuille flow for a Newtonian fluid. Equation 4.31 is known as the
Buckingham’s equation for an idealized Bingham plastic.

4.1.4.3 Casson’s Fluid

The constitutive relationship for a Casson’s fluid is given by the relationship

τ = τ y + K c γ
(4.32)
7328_C004.fm Page 145 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 145

Replacing γ
by − du
dr and the expression for the yield stress, we obtain

du 1 r ∆p τy
− = −
dr K c 2L Kc

Squaring the above equation, we get

du 1 r ∆p τ y 2 r τ y ∆p
− = 2 + 2− 2
dr K c 2 L K c Kc 2L

or

1 ∆p τy 2 τ y ∆p
du = − rdr − 2 dr + 2 rdr
Kc2 2 L Kc Kc 2L

Integrating

3
1 ∆p r 2 τ y 2 τ y ∆p r 2
u=− − 2 r+ 2 +C
2
Kc 2 L 2 Kc Kc 2L 3 2

applying the boundary condition that u = 0 at r = R

3
1 ∆p a2 τ y 2 τ y ∆p R 2
C= + 2 R+ 2
2
Kc 2 L 2 Kc Kc 2L 3 2

and substituting for C and rearranging terms, we obtain the relationship for
the velocity profile as

u=
1 ∆p 2 2
Kc 2 4L
τy
(R − r ) + 2 (R − r ) −
Kc
4
3K c 2 2L (
τ y ∆p 3 3
R 2 −r 2 ) (4.33)

As we discussed with a Bingham plastic, there will be no flow if ∆p/L <

2τy/R, while, if ∆p/L > 2τy/R, flow will exist. Similarly, the velocity profile
will be flat in the core region with a Casson velocity profile in the peripheral
region.
The radius of the core region is given by

2 Lτ y
Rc =
∆p
7328_C004.fm Page 146 Friday, October 6, 2006 12:25 PM

146 Biofluid Mechanics: The Human Circulation

Substituting for the yield stress from above in Equation 4.33, we get the
core velocity as

uc =
1 ∆p 2
Kc 2 4L
( R − Rc 2 ) + K1 2 R2c ∆Lp ( R − r ) − 3K4 2
c c 2L 2L (
Rc ∆p ∆p 3 2 3
R − Rc 2 )
Rewriting the equation, we get

uc =
( ) R
∆p
L
−
8 3 R2
Rc R 2 + 2 Rc R − c 
4K c 2 
2 (4.34)
3 3 

For r > Rc , the expression for the velocity profile will be:

uper =
( ) R
∆p
L 2 − r 2 + 2 Rc R − 2 Rc R −
8 3
Rc R 2 +
8 3 

4K c 2 
Rc R 2  (4.35)
3 3 

The flow rate through the tube is:

Q = Qcore + Qper

Qcore = uc πRc 2 = πRc 2

( ) R
∆p
L
−
8 3 R2
Rc R 2 + 2 Rc R − c 
4K c 
2
2 3 3 

and the expression for the flow in the core region can be simplified as

Qcore = π
( ) R R
∆p
L
−
8 52 32 R4
Rc R + 2 Rc 3R − c 
4K c 
2 2 (4.36)
2 c
3 3 

The flow through the peripheral region can be obtained as

Qper = ∫u
Rc
per 2πrdr

( )
∆p R
 3 

4K 2 ∫ 
L 8 3 8
= 2π R 2 − r 2 + 2R R − 2R r − R R 2+ R r 2  rdr
c c 3 c 3 c 
c Rc
7328_C004.fm Page 147 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 147

Integrating the above expression and simplifying, we get

Qper = π
( ) R
∆p 4
L
− R 2 Rc 2 +
13 4 2
R + R R 3 − 2 RRc 3 −
8 7 8 3 5 
Rc R 2 + R 2 Rc 2 

4K c  2 2 42 c 3 c 7 3 
(4.37)

Adding Equation 4.36 and Equation 4.37, we get the expression for the
flow through the tube as

Q=
( ∆p L ) π  R 4 − Rc 4 + 2 R R3 − 8 7 

4K c 2  2 c Rc R 2 
42 3 7 

Substituting for Rc in terms of the yield stress, the flow rate is given by

 1 
4  2 τ y  16  2 τ y  
4
1  2τ y 
2
πR 4 ∆p 
Q= 1 −   +   −   (4.38)
8K c 2 L  21  ( ∆p L ) R  3  ( L )  7  ( ∆p L ) R  
∆p
 

Assuming steady flow, one can compute the flow dynamics though a
typical artery using the relationships derived above for Newtonian and non-
Newtonian fluids. For a typical human peripheral vessel, the pressure gra-
dient ∆p/L can be computed as 6 dynes/cm3 from the Poiseuille relationship
assuming a flow rate of 140 ml/min and a radius of 0.43 cm. Using the
Poiseuille relationship, we are assuming a Newtonian behavior for human
whole blood with a viscosity coefficient of 3.5 cP. The wall shear stress can
be computed to be 1.29 dynes/cm2 and the corresponding wall shear rate is
36.86/sec as shown below.

R = 0.43 cm, Q = 140 ml/min, µ = 3.5 cP,

∆p = 8µQ = 6.0dynes / cm 3 , τ w = R∆p = 1.29dynes/cm 2 ,

 πa4 2

and

γ
= τ w µ = 36.9/sec.

Applying the same wall shear stress and the shear rate in the constitutive
relationships for the other fluids, the consistency indices and the correspond-
ing flow rate through the artery can be computed with the magnitudes
shown in Table 4.2. The computed velocity profiles for the four fluids
described above are included in Figure 4.17.
7328_C004.fm Page 148 Friday, October 6, 2006 12:25 PM

148 Biofluid Mechanics: The Human Circulation

TABLE 4.2
Computed Consistency Index Values and the Flow Rates through an
Artery Assuming a Constant Pressure Drop for the Various Fluids
Consistency Index Flow Rate Q
Newtonian fluid µ = 3.5cP 140 ml/sec

Power law fluid 127

(n = 0.75) k pl = τ w γ
0.75
w

= 0.0862
Bingham plastic (τ w − τ y ) 136
µb =
τ y = 0.07 γ

(dynes/cm2 ) = 0.0331
Casson’s fluid τw − τ y
117
k c=
τ y = 0.07 γ

(dynes/cm2 ) = 0.1435

Conversely, by assuming a constant flow rate and the consistency index

values computed above, one could also compute the pressure drop required
for the various fluids and the corresponding wall shear rates. Assuming a
yield stress for whole blood of 0.07 dynes/cm2, the core radius where plug
flow exists is very small (0.02 cm). Hence, the core region with a constant
velocity is confined to a very small region near the center of the tube in the
above plots. Using the assumed blood flow rate and the radius for the vessel
as above and the consistency indices computed for the various fluids, the

Velocity profiles – Newtonian/non-Newtonian fluids

9
8 Newtonian fluid
Power law fluid
7
××× ×× ××× × × × ××
××××
Bingham plastic
Velocity (cm/s)

6 ×× × × Cassons fluid
××
××
5 ××
××
××
4 ×
××
3 ××
×
×
2 ×
×
×
1 ×
×
×
0 ×
0 0.1 0.2 0.3 0.4 0.5
Radius (cm)

FIGURE 4.17
Velocity profiles computed with the radius and consistency index values shown in Table 4.2
for the various fluids. A constant pressure drop was assumed for each case.
7328_C004.fm Page 149 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 149

TABLE 4.3
The Pressure Gradient and the Wall Shear Stress Computed for the
Various Fluids with the Assumption of a Constant Flow Rate
Bingham
Newtonian Power Law Plastic Casson’s Fluid
Consistency 0.035 0.086 0.033 0.144
Index
6.0 6.37 6.10 6.71
∆p

dynes/cm3
1.29 1.37 1.31 1.44
τw
dynes/cm 2

computed pressure gradients and the corresponding wall shear stresses for
each of the fluids are shown in Table 4.3, when the flow rate through the
vessel is assumed to be the same for each case. The corresponding velocity
profiles are plotted in Figure 4.18.
It can be observed that there is an increase in pressure gradient of about
12% between the Newtonian fluid and the Casson’s fluid with the flow rate
and tube radius maintained constant and the wall shear stress increased by
about 10%. In simulation studies on blood flow in an artery, it is a common
practice to assume that blood is Newtonian since the shear rates are
expected to be larger than 50 sec−1 for most of the cardiac cycle. The above

Velocity profiles for Newtonian/non-Newtonian fluids

9
8 × ××××× × Poiseuille
×× ×× ×× ×××
7 ××× Power law
××
×× Bingham
×
Velocity (cm/s)

6 ×× × Cassons
××
5 ××
××
4 ××
××
3 ×
×
2
×
1 ×
×
×
0 ×
0 0.1 0.2 0.3 0.4 0.5
Radius (cm)

FIGURE 4.18
Velocity profiles for the various fluids assuming a constant flow rate and the computed pressure
drops (and wall shear stress) shown in Table 4.3.
7328_C004.fm Page 150 Friday, October 6, 2006 12:25 PM

150 Biofluid Mechanics: The Human Circulation

computations demonstrate that the wall shear stress magnitudes can be

expected to differ by about 10% if the Casson’s relationship rather than the
Newtonian law is employed in the simulations.

4.1.5 Hemolysis and Platelet Activation with Fluid

Dynamically Induced Stresses
We discussed above the rheological behavior of blood and the experimental
methods to determine the viscous behavior of whole human blood. Whole
blood follows a nonlinear constitutive relationship between the shear stress
and strain in the form of Casson’s equation. Furthermore, whole human blood
exhibits a yield stress, although its magnitude is too small to be of physiolog-
ical importance. The apparent viscosity for blood is relatively high at low rates
of shear because the red blood cells tend to aggregate, resulting in increased
viscosity. As the rate of shear increases, the apparent viscosity reduces to an
asymptotic value and, hence, whole blood exhibits a shear-thinning behavior.
At relatively large rates of shear, we can approximate the whole blood consti-
tutive relationship by that of a Newtonian fluid. A thorough knowledge of
the rheological properties of blood is important for many applications. In order
to analyze the blood flow dynamics in the human circulation from the large
arteries to the capillary level, and then to the venous circulation, it will be
necessary to apply the constitutive equations. In addition, blood analog fluids
with rheological behavior similar to that of whole blood are necessary for
employing in vitro experimental analysis of flow behavior in the various
regions of the human circulation. In Chapters 5 and 6, steady and unsteady
flow models of blood in large arteries and their applications to the under-
standing of physiological flow behavior will be discussed.
Comprehensive analysis of blood flow characteristics in the normal circu-
lation will also help in delineating changes in flow behavior under various
circulatory diseases. For example, atherosclerosis, or the formation and
development of plaques in the large arteries, is known to develop at specific
sites within the human circulation (Chapter 3, Figure 3.27) where the arteries
exhibit curvatures, bifurcations, or branches. Since these geometrical features
induce complex flow dynamics and fluid stresses compared to those in
arteries with a relatively straight geometry, various theories for the initiation
and development of atherosclerotic plaques due to hemodynamic factors
have been suggested. Numerous theoretical and experimental studies eluci-
dating flow behavior in curved, bifurcating, or branching geometries are
reported in the literature to support the hemodynamic theories of atheroma
development (Chandran, 1993; Ku, 1998). These studies have employed
either Newtonian or non-Newtonian flow behavior with the appropriate
constitutive relationships discussed in this chapter. Results of these studies
will also be discussed in detail in Chapter 6.
In employing the constitutive relationships for Newtonian or non-Newtonian
fluid behavior in blood flow models, blood is treated as a homogeneous
7328_C004.fm Page 151 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 151

106

RBC destruction
Platelet destruction
104

102
Exposure time (s)

100 Stress effects

dominate
Platelet serotonin
release
10–2

Surface effects
10–4 dominate

10–6

10–8
100 101 102 103 104 105 106
Shear stress (dynes/cm2)

FIGURE 4.19
Shear stress — exposure time plot for threshold hemolysis of red blood cells and destruction
of platelets.

fluid, thus ignoring the significant fraction of formed elements (red and
white blood cells and platelets) in it. The flow-induced stresses, predomi-
nantly shear stresses, can activate or even destroy the blood cells and platelets.
Many studies have been performed on the effect of high shear stresses on
blood cells. The magnitudes of the shear stresses and the time of exposure
of the cells to stresses are important determinants of the destruction or
activation of the blood cells. A shear stress vs. time of exposure plot for red
blood cells hemolysis threshold and for activation or destruction of platelets
is shown in Figure 4.19. The data included in the plot are a compilation of
experimental data in which the blood was subjected to a range of shear
stresses and exposure times collected by various investigators. As illustrated
in the figure, hemolysis and platelet activation can result even with relatively
lower magnitudes of shear stresses if the formed elements are exposed to
foreign surfaces during flow.
Hemolysis of red blood cells will result in the release of hemoglobin into
the plasma. Hemoglobin within the red blood cells is the principal material
for transport of oxygen from the lungs to the body tissues. Therefore,
hemolysis at an elevated rate will result in anemia so that the blood cannot
7328_C004.fm Page 152 Friday, October 6, 2006 12:25 PM

152 Biofluid Mechanics: The Human Circulation

transport sufficient oxygen to the tissues. Platelet activation, on the other

hand, can trigger the formation of thrombus within the vessel. Thrombus
formation is initiated by the adherence of activated platelets to the suben-
dothelial structures (e.g., the extra cellular matrix, collagen, and the vascular
smooth muscle cells), which have been exposed due to local vascular injury.
The thrombus then grows with additional platelet activation and aggrega-
tion. With the coagulation of blood, the adhered platelets combine with
fibrin to form thrombus or blood clots. If the thrombus breaks apart and
forms emboli that are transported downstream, they may lodge in smaller
vessels and prevent blood supply to the region downstream of those sites.
Should emboli lodge in the smaller vessels in the brain, neurological deficits
will occur. Heart failure can occur if the emboli affect the smaller vessels in
the coronary circulation.

4.2 Blood Vessel Mechanics

Pulsatile blood flow in the human arterial circulation is the result of constant
interaction between the flowing blood and the arterial wall. In the analysis
of the fluid mechanics of blood flow in human arteries, we need to under-
stand the behavior of the arterial wall in addition to knowledge about the
flow behavior of blood. Therefore, we will briefly review the structural
components and the mechanics of the walls of blood vessels.

4.2.1 Structural Components of the Blood Vessel Wall

and Their Contribution to Material Behavior
The arterial wall is a composite of three layers, each of which contain varying
amounts of elastin, collagen, vascular smooth muscle cells, and extracellular
matrix (Figure 4.20). The inner most layer of the artery, called the intima,
consists of a single layer of endothelial cells and a very thin lamina of elastin.
The role of the endothelial cells lining the inner surface of the artery is to
provide a smooth wall and selective permeability to water, electrolytes,
sugars, and other substances between the blood stream and the tissues. The
outer layer, the adventitia, consists of connective tissue that merges with the
surrounding structures. The middle layer, known as the media, consists of
elastin, collagen, and vascular smooth muscles embedded in an extracellular
matrix. Wall diameters, thicknesses, and compositions of various blood ves-
sels are shown in Figure 4.21. Elastin fibers are present in all the vessels
except the capillaries and the venules, and are very easily stretched. Collagen
fibers, which form a network in both the media and the adventitia, are much
stiffer than the elastin fibers. However, the collagen fibers are normally found
7328_C004.fm Page 153 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 153

Composite reinforced by
coilagen fibers arranged
in helical structures

Helically arranged fiber

reinforced medial layers

Bundles of collagen fibrils

External elastic lamina
Elastic lamina
Elastic fibrils

Collagenfibrils

Smooth muscle cell Futima

Internal elastic lamina
Endothelial cell Media

Adve
ntitia

FIGURE 4.20
Schematic of the components of a healthy artery. (From Holzapfel, Gasser, and Ogden (2000)
A new constitutive framework for arterial wall mechanics and a comparative study of material
models, in J. Elasticity 61: 1–48. With permission.)

with a degree of slackness and, thus, their full stiffness is not apparent until
after the vessels are stretched to the extent where the slackness is removed.
Even though the individual components behave like linear elastic materi-
als, the combination of taut elastin and relaxed collagen fibers results in a
nonlinear behavior as illustrated in Figure 4.22. A schematic illustration of
the taut elastin and unstretched collagen fibers is shown in Figure 4.22a. A
schematic of the force-deformation test for a single elastin fiber is shown in
Figure 4.22b whose elastic modulus is about 106 dynes/cm2. A similar sche-
matic for the collagen fiber behavior with an elastic modulus of about 109
dynes/cm2 is shown in Figure 4.22c. For a combination of elastin and col-
lagen fibers, the effect of increasing distension where the collagen fiber
becomes taut and reacts to the load will result in a bilinear curve as shown
7328_C004.fm Page 154 Friday, October 6, 2006 12:25 PM

154 Biofluid Mechanics: The Human Circulation

Aorta Medium Small art. Sphincter

art. arteriole pre-cap. AVA
0.4 cm 30 µ 35 µ
1 mm 20 µ 30 µ

End.
Ela.
Mus.
Fib.

True Vena cava

Venule
capillary 20 µ Vein
8µ 2µ
1µ
0.5 cm 3 cm
0.5 mm 1.5 mm

End.
Ela. > 200 µ
Mus. > 45 µ
Fib.

FIGURE 4.21
The diameter, wall thickness, and components of the wall of the various blood vessels in the
circulatory system. (Redrawn from Burton, A.C. (1971) Physiology and Biophysics of the Circulation.
Year Book Medical Publishers, Chicago. With permission.)

in Figure 4.22d. Figure 4.22e illustrates the continuous recruitment of more

collagen fibers with increased load results in a nonlinear force deformation
behavior. The function of the elastin and collagen fibers is to maintain a
steady tension within the vessels to act against the transmural pressure
exerted on them. On the other hand, the function of the vascular smooth
muscle is to provide an active tension by means of contraction under phys-
iological control. Thus, more smooth muscle content is observed in the
smaller arteries and arterioles where vascular resistance can be controlled
by the change in the radius of the vessel produced by the tension of the
smooth muscles. The elastic modulus for smooth muscles depends upon the
relaxed or contracted state of the muscle. The endothelial cells act more as
sensors and mediators of physiologic changes and offer negligible resistance
to mechanical load.
7328_C004.fm Page 155 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 155

σ
Collagen alone

Combined

Elastin Collagen
(a) Elastin alone

σ ε

Collagen Elastin

Elastin
(d)
ε
Single elastin fiber σ

Recruitment of 3rd
collagen fiber
(b)
Recruitment of 2nd
σ collagen fiber
Collagen
Recruitment of 1st
collagen fiber

ε
ε Collagen Elastin
Single collagen fiber

(c) (e)

FIGURE 4.22
(a) Schematic of the elastin–collagen complex in an unstretched arterial wall; (b) stress–strain
relationship for an elastin; and (c) for a collagen fiber; (d) stress–strain behavior of an elastin
and collagen fiber combination; and (e) behavior with additional collagen fiber recruitment
with increasing load.

4.2.2 Material Behavior of Blood Vessels

Uniaxial extension tests can be performed in vitro on strips of arteries cut in
various orientations from the artery (such as circumferential or longitudinal
strips) to determine the force deformation behavior. When biological tissues
are subjected to load-deformation tests cyclically, the tests will result in a
hysteresis loop. In other words, the load-deformation curve during loading
will take a different path from the unloading curve, indicating an energy
loss during the loading and unloading process. With repeated loading and
7328_C004.fm Page 156 Friday, October 6, 2006 12:25 PM

156 Biofluid Mechanics: The Human Circulation

FIGURE 4.23
Images of a pig’s thoracic aorta being subjected to a uniaxial test (bottom row), and the
corresponding nonlinear stress–strain relationship.

unloading, the area under the loop will decrease rapidly at first and reach
a steady state after a number of cycles. For this reason, data on the load-
deformation behavior for biological tissue are obtained only after precondi-
tioning the tissue by loading and unloading for several cycles. Figure 4.23
shows a photograph of a strip of a pig aorta gripped between clamps and
attached to a testing apparatus. The corresponding measured stress–strain
behavior for the specimen is also shown in the figure. Such tests will yield
information about the directional dependence of the force-deformation
behavior, if any, of the arterial wall. These in vitro tests should be performed
in a physiological environment with controlled ionic content, moisture, tem-
perature, and other variables. Figure 4.24 represents a schematic comparison
of the stress–strain relationship for soft tissue, such as skin, and blood vessels
with dry bone and steel used in engineering applications.
More useful information on the actual behavior of blood vessels can be
obtained by performing the tests with the vessels in their natural geometry.
Segments of blood vessels in their original configuration can be used in in
vitro or in situ testing under an applied transmural pressure. For in vitro
testing, relatively long segments need to be obtained, so that making the
measurements sufficiently away from the supports can eliminate “end
7328_C004.fm Page 157 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 157

σ Steel

Dry bone

Blood vessel

FIGURE 4.24
A comparison of stress–strain relationship for steel, bone, blood vessel, and skin.

effects.” Once again, the physiological environment needs to be maintained

during the testing to obtain any meaningful data. Consideration must be
given to the changing state of the smooth muscle with in vitro specimens. In
the absence of stimuli, the smooth muscles tend to relax while they also stiffen
after death (‘‘rigor mortis’’). When a segment of artery is cut, the length can
decrease by as much as 40% of the initial length. Hence, the usual practice
with in vitro testing is to mark the length of the artery before dissecting it and
then stretching the segment back to its original length before performing the
tests, so that the mechanical tethering will be similar to that in vivo.
Either a static pressure or a dynamic pressure pulse is applied to the arterial
segment in both the in vitro and in situ tests and the changes in the radius
and the wall thickness are measured to determine the load-deformation
behavior. The applied pressure can be measured relatively easily with sen-
sitive pressure transducers. However, measurement of change in radius and,
especially, the wall thickness is more complicated. Several techniques using
strain gauges, differential transformers, or electro-optical methods have been
used to measure the change in the outer radius. Measurement of the change
in the internal diameter and, thereby, the change in the wall thickness
requires data from angiographic, ultrasonic, or other imaging modalities. In
static tests, a change in radius for a corresponding change in a step pressure
increase can be measured to determine the incremental modulus as described
earlier. Sufficient time must be allowed before the measurement of the diam-
eter change after the application of pressure to allow for creep effects. In
dynamic tests, a sinusoidal pressure pulse is applied and the resulting
change in diameter is continuously recorded so that viscoelastic properties
of the arterial segments can be determined.
In vivo tests can also be performed either by surgically exposing the vessel
under study and making direct measurements or by applying noninvasive
7328_C004.fm Page 158 Friday, October 6, 2006 12:25 PM

158 Biofluid Mechanics: The Human Circulation

imaging techniques and inserting a pressure transducer by catheterization to

record changes in diameter and the pulse pressure, respectively. A pressure–
strain modulus can be computed using the relationship given in Chapter 2,
Equation 2.46. If the wall thickness change can also be measured, a static
elastic modulus can be computed from the measured data using the
relationship given in Equation 2.45. On the other hand, if a continuous
recording of the diameter changes can be obtained along with the physiological
pressure pulse, the signals can be separated into harmonics by the use of
the Fourier series analysis. From this, the dynamic elastic modulus values
over a range of frequencies can be obtained.

4.2.2.1 Assumptions in the Analysis

The data on changes in pressure, radii, and wall thickness, obtained from in
vitro or in vivo experiments, can be reduced to the elastic modulus based on
the expressions derived earlier. However, it should be kept in mind that
certain assumptions have been made in deriving those expressions. For
example, an expression for the elastic modulus was derived based on the
thin-walled elastic tube model (Chapter 2, Equation 2.29). This expression
is applicable only in cases where t/R < 0.1. As can be observed from Figure
4.20, the t/R ratio for most blood vessels ranges from 0.105 to 0.15. Hence,
the thick-walled formulation may be more appropriate in analyzing the
material properties of these vessels. The other important assumptions made
in the formulations described above are the homogeneity, incompressibility,
and isotropic behavior of the material. As already discussed, blood vessels
are not homogeneous and actually consist of several layers and components.
Expressions for the elastic moduli derived above for thick-walled models
include information on the Poisson’s ratio.
It is common to assume that blood vessels, similar to other biological soft tissue,
are incompressible and, hence, one can assign a value of 0.5 for v. This is sup-
ported by the observation that volumetric strain computations of the vessel walls
of excised dogs indicate that blood vessels are essentially incompressible. Due
to the tethering of the blood vessels, it can be expected that the elastic modulus
in the longitudinal direction will be larger than those in the circumferential and
radial directions. Numerous studies have been reported on the computed elastic
modulus in the radial, circumferential, and axial directions with the vessel gen-
erally being increasingly stiffer along those directions, respectively.

4.2.2.2 Experimental Results

Mechanical properties of blood vessels were assessed in vivo by the simultaneous
measurement of intra-arterial pressure and the arterial diameter in dogs in the
1960s. The arteries were assumed to be thin walled and incompressible and
the viscoelastic behavior of the vessel was determined by the relationship

pr ∆r d  ∆r 
σθ = =E + Eν   (4.39)
t R2 dt  R2 
7328_C004.fm Page 159 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 159

A simple viscoelastic model has been assumed in deriving the above

relationship with the circumferential stress σθ, the circumferential
 
strain ∆r/R2,
and the rate of strain in the circumferential direction dtd  R∆ r  . Data for various
 2
arterial segments in the systemic circulation from very young to old dogs
were presented in this study Peterson et al. (1960). The results showed that
the viscous component was relatively small compared to the elastic modulus.
The in vivo elastic moduli of various arterial segments from the thoracic aorta
to the femoral arteries ranged from 1953 to 34,944 gmf/cm2.
Bergel (1961a and 1961b) has presented data for static and dynamic elastic
moduli for the blood vessels based on the thick-walled theory for a homo-
geneous and isotropic material. The in vitro experiments were performed on
excised blood vessels from dogs with transmural pressure load of up to
240 mmHg. A nonlinear behavior characterized by stiffening of the vessels
with increasing pressure is apparent from the static study results shown in
Figure 4.25. Incremental elastic moduli determined for a dog’s aorta, as well
as the femoral and carotid arteries in this study, are included in Table 4.4.
Bergel (1961b) also performed dynamic studies by superimposing a sinu-
soidal variation over a static pressure of 100 mmHg and measuring the

25

20
Einc (dynes/cm2 × 106)

15

10

0
20 40 60 80 100 120 140 160 180 200 220 240
Pressure (mm Hg)

FIGURE 4.25
Incremental static elastic modulus as a function of transmural pressure for dog’s arteries ∆
thoracic aorta,
– abdominal aorta,
– femoral artery, and O – carotid artery. (Redrawn from
Bergel, D.H. (1961a) J. Physiol. 156: 445–457. With permission.)
7328_C004.fm Page 160 Friday, October 6, 2006 12:25 PM

160 Biofluid Mechanics: The Human Circulation

TABLE 4.4
The Static Elastic Properties of the Arterial Wall
Pressure Thoracic Abdominal Femoral Carotid
(mmHg) Aorta Aorta Artery Artery
40 1.2 + 0.1 (6) 1.6 + 0.4 (4) 1.2 + 0.2 (6) 1.0 + 0.2 (7)
100 4.3 + 0.4 (12) 8.9 + 3.5 (8) 6.9 + 1.0 (9) 6.4 + 1.0 (12)
160 9.9 + 0.5 (6) 12.4 + 2.2 (4) 12.1 + 2.4 (6) 12.2 + 2.7 (7)
220 18.2 + 2.8 (5) 18.0 + 5.5 (3) 20.4 + 4.4 (6) 12.2 + 1.5 (7)
Note: Figures in parentheses refer to the number of specimens in the experiments.
Source: Bergel, D.H. (1961a) J. Physiol., 156: 445–457. With permission.

diameter changes employing a photo-optical technique. The dynamic elas-

tic moduli computed for the canine arteries at various frequencies are
compared with the corresponding static values (0 Hz) in Table 4.5 and the
results show that the dynamic incremental modulus is significantly
increased even at 2 Hz compared to the corresponding static values. At
higher frequencies, the dynamic modulus does not show further changes.
The dynamic elastic moduli computed at mean arterial pressures ranging
from 87 to 130 mmHg and pulse frequencies ranging from 1.1 to 2.8 Hz in
the dog arteries in vivo showed a monotonic increase from the thoracic aorta
to the femoral arteries. The viscous modulus ranged from 9 to 12% of the
corresponding dynamic modulus. Effect of age on the viscoelastic proper-
ties of the human arterial wall has also been investigated and the results
showed that the viscous component of the elastic modulus was larger in
the femoral arteries and can be attributed to higher muscular content in
the wall. In the “young” group, the wall stiffness was observed to increase
toward the peripheral vessels and in the ‘‘old’’ group, the opposite trend
was observed. Such studies indicate the presence of regional variation in
the mechanical properties of arteries.
Studies have also been reported on the anisotropic behavior of the arterial
wall. Mean values for the incremental moduli in the radial, circumferential,
and longitudinal directions of 5480, 7510, and 10,100 Gm/cm2, respectively,
at a mean arterial pressure of 154 cm H2O have been reported from data
obtained in vivo in dogs. These studies also showed that the elastic moduli

TABLE 4.5
The Dynamic Elastic Properties of the Arterial Wall
Vessel/Frequency 0 Hz 2 Hz 5 Hz 18 Hz
Thoracic aorta 4.4 + 0.40 (10) 4.7 + 0.42 (10) 4.9 + 0.45 (10) 5.3 + 0.80 (4)
Abdominal aorta 9.2 + 0.94 (7) 10.9 + 0.88 (7) 11.0 + 0.82 (7) 12.2 + 0.46 (4)
Femoral artery 9.0 + 1.15 (5) 12.0 + 0.81 (5) 12.0 + 0.82 (5) 10.6 + 1.39 (5)
Carotid artery 6.9 + 0.48 (6) 11.0 + 1.00 (6) 11.3 + 0.99 (6) 11.5 + 1.03 (6)
Note: Figures in parenthesis show the number of specimens used in the experiments.
Source: Bergel, D.H. (1961a) J. Physiol., 156: 458–469. With permission.
7328_C004.fm Page 161 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 161

increased with increased mean arterial pressure. Moreover, the longitudinal

elastic modulus measured on the same vessels in vitro showed a decrease
after the removal of tethering of the vessels in situ.
The viscoelastic behavior of arteries is usually assessed by experiments on
specimens obtained from healthy animals of various species and the results
obtained have suggested an anisotropic, nonlinear stress–strain relationship
under a transmural load from zero to about 240 mmHg. It can be anticipated
that vascular disease would alter the material behavior. Common vascular
diseases in humans include atherosclerosis, high blood pressure (hyperten-
sion), and diabetes. Atherosclerosis is the formation of plaques within the
artery wall and lumen and is associated with the accumulation of cholesterol
combined with low-density lipoproteins (LDL) that dissolve in plasma. The
formation and growth of these lesions and their relationship with the local
flow dynamics are further described in Chapter 6.
Briefly, early atheroma results from abnormal deposition of LDL cholesterol
within the intima, followed by a cellular response, composed of macrophages,
lymphocytes, and proliferating vascular smooth muscles. Collagen and elastin
fibers eventually replace the cellular constituents, resulting in a changing histo-
logical picture with the lesion development. Initially, the arteries respond by
remodeling or thinning of the media in order to accommodate the expanding
lesion within the arterial wall in order to maintain the lumen cross section for
blood flow. With further development of the plaque and as the contents become
calcified, the plaques protrude into the lumen and begin occluding the cross
section available for blood flow. It can be anticipated that the arterial wall
material property is continually altered with the various stages of lesion devel-
opment and several studies have been reported on the nature of such alterations.
The arterial wall also responds to increased blood pressure by the reorganization
and thickening (hypertrophy) of the medial wall in the artery. The wall
thickening will result in the circumferential stress due to the transmural blood
pressure returning to normal levels. Diabetes also affects the material behavior
of the arterial wall and studies on the alterations in arterial wall material behavior
with these diseases continue to be the subject of ongoing research.
The interaction between stress and strain for an elastic material is the
constitutive relationship describing the behavior of the material under
external loading. In order to mathematically describe the material behavior
for nonlinear material (such as blood vessels), exponential, polynomial, or
logarithmic relationships between stress and strain have been generally
employed. In developing mathematical relationships for nonlinear and aniso-
tropic materials, it is convenient to describe the material behavior in the form
of a strain energy function (work done on the material in the deformation
process) and to express the relationship in terms of material constants. A best
curve fit of the force-deformation experimental data is employed for deter-
mining the material constants for the mathematical relationship being
employed. Further details on such material behavior descriptions for biological
soft tissues, including blood vessels, can be found in Fung (1981) and other
similar publications.
7328_C004.fm Page 162 Friday, October 6, 2006 12:25 PM

162 Biofluid Mechanics: The Human Circulation

4.2.3 Residual Stress on Blood Vessels

In engineering materials, a component that is not subjected to an external
load will not have any deformation and no internal stresses will develop in
the material. The arteries are subject to external loads through the transmural
blood pressure and, hence, using the same analogy as for general engineering
materials, it can be expected that arteries will be stress-free at zero transmural
load. In the last couple of decades, several studies have demonstrated that
the arterial wall tissue is not stress-free at zero transmural loads To demon-
strate, a small segment of an artery is excised into a ring shaped element
and, subsequently, a radial cut is made on the ring-shaped segment. Should
the artery be stress-free, the artery will maintain its circular shape even after
the radial cut. However, it has been demonstrated that the arterial segment
opens to a sector (a horseshoe shape) within a few minutes after the longi-
tudinal cut, as schematically shown in Figure 4.26, with the arterial wall
being stress-free in this open segment. The artery is assumed to be stress-
free in the sector-shaped configuration because an additional radial cut will
result in two pieces that do not result in noticeable additional strains or
change in shape. A photograph of an excised pig’s femoral artery in the ring-
shaped configuration and in the sector-shaped, stress-free state after a radial
cut is shown in Figure 4.27.
It is evident, then, that the ring-shaped arterial segment is subjected to a
prestress (or residual stress) even with no load (zero transmural pressure).
The magnitude of the prestress has been generally described in terms of the
opening angle of the sector shape, as shown in Figure 4.26. Studies have
demonstrated that the residual stress is a function of the anatomical location
of the arterial segment (e.g., carotid artery vs. femoral artery), and the species
from which the segments are obtained. The opening angle also varies
depending on the location of the radial cut of the ring-shaped segment. By
measuring the circumferential lengths of the intimal and adventitial bound-
aries in the uncut (ring-shaped) configuration and the corresponding lengths
in the open sector configuration, the residual strains at zero load can be

Radial cut

Ring shaped segment Sector shaped

segment: Q- hearing angle

FIGURE 4.26
Schematic of a ring-shaped arterial segment and the open sector (horseshoe-shaped), stress-
free state of the segment after a radial cut.
7328_C004.fm Page 163 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 163

FIGURE 4.27
A photo of an excised, ring-shaped, pig’s femoral arterial segment and the open-sector-shaped,
stress-free state of the artery with an additional radial cut.

computed and have been shown to be compressive in the intimal layer and
tensile in the adventitial layer. The computed residual stress with an assumed
elastic modulus demonstrates that the magnitudes of the residual stress are
about 10% of the estimated circumferential stress in the arterial wall under
physiological transmural load. Under normal physiological loading neglecting
the residual stress, it has been shown that circumferential stress is generally
larger in the intimal border and decreases towards the adventitial border.
Thus, the inclusion of a residual stress that is compressive in the intimal
border and tensile in the adventitial border will result in a circumferential
stress distribution across the arterial wall thickness, which is more uniform
under physiological loading. The presence of residual strain and stress has
7328_C004.fm Page 164 Friday, October 6, 2006 12:25 PM

164 Biofluid Mechanics: The Human Circulation

been implicated in the remodeling of the arterial wall. Finite element studies
indicate that the distribution of the circumferential stress throughout the wall
thickness is more uniform if the residual strain effect is included in the
analysis.

4.2.4 Material Characterization of Cardiac Muscle

As was described in the chapter on cardiovascular physiology (Chapter 3),
the cardiac muscles constantly contract and relax in order to develop blood
pressure and maintain adequate circulation to the various organs in the body.
The basic structure of the cardiac muscle is a fiber that is arranged in columns
that are roughly cylindrical in shape with a diameter of 10 to 20 µm and a
length of 50 to 100 µm. The fibers are oriented nearly vertically in the inner
(endocardium) and outer (epicardium) layers of the muscle. The fiber ori-
entation undergoes a gradual change through the myocardium such that the
fibers in the middle third of the wall thickness are oriented circumferentially.
Individual muscle fibers are surrounded by a surface limiting membrane
called the sarcolemma. Within the sarcolemma, each fiber contains rod-like
structures called the fibrils. The fibrils run the length of the fiber and contain
serially repeating structures called the sarcomeres. Mitochondria are inter-
spersed within the fibrils while the sarcotubular system surrounds the fibrils
and connects with the sarcolemma. The sarcomeres are the fundamental
functional units of cardiac contraction.
In order to describe the behavior of cardiac muscles, both the passive
elastic characteristics (during the diastolic phase) and the active contractile
properties (during the early systolic phase) of the cardiac muscle, need to
be analyzed. The cardiac muscles are at least orthotropic with the force-
deformation behavior being stiffer along the direction of the fibers compared
to that in the transverse direction. Since the fiber orientation gradually
changes along the thickness of the muscle, passive elastic deformation tests
are usually performed on strips of papillary muscle since the fibers are ori-
ented along the longitudinal direction in these muscles. Just as with other
biological soft tissue, the passive stress–strain behavior of cardiac muscles is
also nonlinear and an exponential stress–strain relationship is generally
employed to describe the rheological properties of the cardiac muscle. The
contractile properties of the cardiac muscle (and skeletal muscle) are generally
described using the force-velocity relationship, which is typically obtained
using the following steps. Strips of isolated papillary muscles are stretched
with a preload and temporarily fixed at this length. An after load is then
attached in this configuration and the muscle is stimulated. The peak velocity
of contraction for the given total load (preload and after load) is measured
as the muscle contracts under the stimulation. By varying the initial length
of contraction as well as the preload, a family of force-velocity curves is
obtained. In the case of cardiac muscle, the data extrapolated to obtain the
velocity at no load from the set of curves converge to a single point referred
to as the maximum velocity of shortening. The passive elastic properties, as
7328_C004.fm Page 165 Friday, October 6, 2006 12:25 PM

Rheology of Blood and Blood Vessel Mechanics 165

well as the maximum velocity of shortening, can be altered if the cardiac

muscles are diseased due to myocardial infarction or cardiomyopathy.

4.3 Summary
In this chapter, the rheological behavior of blood was considered. The basic
relationships for viscometry were developed and the application of viscometry
to measure the viscosity of blood and plasma was discussed. The importance
of understanding the rheological behavior of blood and its changes with
diseased states was briefly discussed. A study of the arterial wall material
behavior is important for a basic understanding of the physiology of blood
vessels. The arteries serve as a pressure reservoir in the circulation. As blood
is ejected from the left ventricle during systole, the vessel walls distend as
the arterial pressure rises. Later, during diastole, the passive arterial wall
tension maintains a force to drive the blood through the peripheral capillaries.
Assessment of mechanical properties of the blood vessels in various
segments of the peripheral circulation is important in the understanding of
the propagation of pressure pulses and the pressure-flow relationship in the
circulation under normal and diseased states. It is also important to under-
stand the interaction between blood flowing through the lumen and the
vessel wall. Knowledge of the behavior of the normal artery will also be
helpful in early detection of any diseases of the blood vessels, such as
atherosclerosis. Study of the material characterization of the cardiac muscles
is also important in our understanding of normal physiological behavior and
various alterations, which occur with the onset of diseases.

4.4 Problems

4.1 Derive the relationship for the torque developed in a Couette vis-
cometer (Equation 4.6). A typical Couette viscometer with a 4.0 cm
inner diameter, 4.2 cm outer diameter and 6 cm height is used to
measure the viscosity of blood. If the angular velocity is 50 rpm,
compute the torque required to rotate the outer cylinder. Also plot
the distribution of the rate of shear between the inner and the outer
cylinder. Use a whole blood viscosity of 3.5 cP.
4.2 For a typical red blood cell, the intracellular fluid volume is 87 µm3
and the surface area of the cell is 163 µm3 . If the red blood cell had
a spherical shape with the same intracellular fluid volume, compute
the surface area for the same. What is the advantage of the biconcave
shape for the red blood cell?
7328_C004.fm Page 166 Friday, October 6, 2006 12:25 PM

166 Biofluid Mechanics: The Human Circulation

4.3 Plot the apparent viscosity as a function of shear rate for the rela-
tionship given by Equation 4.8 for a range of shear rates from 1 to
500 sec−1. Compute the apparent viscosity of blood at a rate of shear
of 230 sec−1. Use a plasma viscosity of 1.2 cP.
4.4 For a normal hematocrit of 43% and fibrinogen concentration of 0.3
g/100 ml, compute the yield stress for blood employing the empir-
ical formulae given in Equation 4.10. In a capillary with a diameter
of 5 µm and a length of 1 mm, estimate the minimal pressure gradient
required for the blood to overcome the computed yield stress and
induce flow through the vessel.
4.5 Derive expressions for the flow rate and the apparent viscosity for
the cell-free boundary layer model described in the text.
4.6 For human blood, assuming values of plasma viscosity of 1.2 cP and
whole blood viscosity of 3.3 cP at 37°C, compute the apparent vis-
cosity of blood flowing through a tube of 100 µm diameter using
both the cell free marginal layer and the Sigma effect theories. Assume
a cell free layer thickness of 3 µm and a mean thickness of unsheared
laminae of 15 µm . Compare the values from both theories.
4.7 Assume that whole blood with a hematocrit of 45% flows through
a small diameter tube. The total flow rate is 16 cc/sec, although in
the core region it is 12 cc/sec and in the peripheral region it is 4 cc/sec.
The blood cells accumulate in the core region with a volume of
10 mm3 and there are no blood cells present in the cell-free peripheral
region which has a volume of 6 mm3. By performing a mass balance
on the red blood cells, determine the hematocrit in the core region
and the average hematocrit in the whole tube. What is the effect on
apparent viscosity of the blood in the tube?
4.8 An artery of an animal has an internal diameter of 1 cm and a wall
thickness of 0.75 mm at an end diastolic pressure of 85 mm Hg. An
8 percent increase in the diameter was measured for a pulse pressure
of 45 mm Hg. Compute the circumferential stress in the wall of the
artery as well as the elastic modulus of the vessel at the mean arterial
pressure assuming that the arterial wall is thin and made of linear
isotropic elastic material.
4.9 In an experiment with a canine femoral artery, the following stress-
strain relationship was obtained. Using the data, compute the
incremental elastic modulus at a strain of 0.15.

Strain 0.00 0.03 0.05 0.07 0.09 0.10 0.12 0.15 0.17 0.19 0.21 0.23 0.25 0.27 0.28
Stress (× 105 0 0.3 1 2 5 10 15 31 48 66 83 102 121 119 102
dynes/cm2)
7328_C005.fm Page 167 Monday, September 18, 2006 12:20 PM

5
Static and Steady Flow Models

CONTENTS
5.1 Introduction ................................................................................................167
5.2 Hydrostatics in the Circulation ...............................................................168
5.3 Applications of the Bernoulli Equation .................................................169
5.3.1 Total vs. Hydrostatic Pressure Measurement ...........................169
5.3.2 Arterial Stenoses and Aneurysms ..............................................170
5.3.3 Cardiac Valve Stenoses.................................................................172
5.4 Rigid Tube Flow Models ..........................................................................178
5.4.1 Vascular Resistance .......................................................................180
5.5 Estimation of Entrance Length and Its Effect
on Flow Development in Arteries ..........................................................182
5.6 Flow in Collapsible Vessels......................................................................185
5.7 Summary .....................................................................................................189
5.8 Problems......................................................................................................189

5.1 Introduction
In this chapter, we will discuss some applications of hydrostatics and steady
flow models to describe blood flow in arteries. Even though the flow in the
human circulatory system is unsteady, particularly at the precapillary level,
steady flow models do provide some insight into the aspects of flow through
the arteries and some useful applications can be found using the steady
flow models. As would be expected, steady flow models are also simpler
to use because of the absence of time variations in the governing equations
(cf. the Equations of Motion, Equation 1.43a to Equation1.43c). Steady flow
models also avoid the complexity of having a moving interface between the
blood and the vessel wall as the artery distends in response to pulsatile
pressure.

167
7328_C005.fm Page 168 Monday, September 18, 2006 12:20 PM

168 Biofluid Mechanics: The Human Circulation

5.2 Hydrostatics in the Circulation

For a typical adult, the mean arterial pressure at the level of the heart is about
100 mmHg. In the supine (i.e., lying down) position, pressures in the blood
vessels of the head and legs would be approximately 95 mmHg (Figure 5.1).
However, when a person is standing up, the hydrostatic effects that were
discussed in Chapter 1, section 1.3: Hydrostatics, must be taken into account.
In order to assess the pressure differences due to gravitational effects, the
force balance of Equation 1.12 can be applied as

∆p = ρgh (1.12)

Using the above equation and knowing the density of blood, the pressures
in the blood vessels can be estimated as shown in the figure. Since arteries
are relatively stiff vessels, the increase in pressure due to hydrostatic effects
will only cause minimal alterations in the blood volume since the vessel
cross section remains nearly constant. However, in the veins, the blood
volume will be greatly affected because they are much thinner and can
expand significantly (see Chapter 3, section 3.4: Systemic Circulation). In
fact, if the venous tone is low and a person suddenly stands up, they may
actually faint because of the increased pooling of blood in the lower extrem-
ities that reduces blood flow back to the heart and, thus, flow to the brain.
Moreover, due to the corresponding decrease of pressure in the head due to
further increases in elevation, the veins in that region may be in a partially

Arterial pressure Arterial

Venous
39
51
(–44 mm Hg)

100 2
95 100 95
Venous pressure (+88 mm Hg)

5 2 5 183

FIGURE 5.1
Hydrostatic pressure differences in the circulation. (Redrawn from Burton, A. C. (1971) Physi-
ology and Biophysics of the Circulation. Year Book Medical Publishers, Chicago. With permission.)
7328_C005.fm Page 169 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 169

collapsed state. Typically, the veins take on an elliptical, or dumbbell shape,

in which flow of blood still persists, but against a much higher resistance.
A good way to prevent this is by flexing the muscles in the leg calf so as to
effectively pump the blood out of the lower veins and back to the heart. In
the event of fainting, however, the resultant change in posture from standing
to supine is the ideal “remedy” for the effects of cerebral hypotension.

5.3 Applications of the Bernoulli Equation

The Bernoulli equation, derived in Chapter 1, section 1.6, was based on the
assumption of steady flow along a streamline in an incompressible, inviscid
fluid (Figure 1.9) and is given by the relationship

V2
p+ρ + ρgz = H (1.52)
2

where H is a constant and is referred to as the “total head” or total energy

per unit volume of fluid.
This equation can be applied to several disease conditions in the circulatory
system, such as flow through constrictions and across orifices, keeping in
mind the assumptions used in deriving the equation.

5.3.1 Total vs. Hydrostatic Pressure Measurement

The most common method for measuring intravascular blood pressure clini-
cally is to insert a fluid-filled catheter into a vessel, then externally connecting
the catheter to a pressure transducer. The pressure measured then is that which
is sensed at the tip of the catheter and transmitted along the fluid channel to
the manometer (known as a “Fluid-Filled Catheter Manometer”). In addition
to the pressure energy, however, the effect of kinetic energy on the pressure
measurement must also be taken into account, as seen in Equation 1.52.
The hydrostatic pressure, p, would be best measured by the catheter shown
in the configuration with a lateral port (Figure 5.2a). However, when the
catheter opening faces the flow, as shown in the configuration in Figure 5.2b,
the fluid that impinges against the catheter opening slows to zero velocity.
When that happens, the kinetic energy in the fluid will be converted into
pressure and will introduce a difference between the end pressure and the
lateral pressure

ρV 2
pe − p1 = pk = (5.1)
2
7328_C005.fm Page 170 Monday, September 18, 2006 12:20 PM

170 Biofluid Mechanics: The Human Circulation

To pressure To pressure
transducer transducer

Flow Side ports Flow

‘‘Lateral pressure” ‘‘End pressure”

(a) (b)

FIGURE 5.2
Schematic for the measurement of lateral and end pressure.

In this equation and in subsequent discussions on the application of the

Bernoulli equation in biological flows, V will refer to the mean velocity over
the cross section of the artery/conduit. This difference can be significant,
especially in segments of narrowed vessels where velocities, and, thus, the
kinetic energy become considerable. Thus, for normal physiological pressure
measurements, openings are located at the sides of the catheter such that
lateral pressures are measured. In the case of catheter-tipped pressure trans-
ducers, where a sensor is directly placed on the catheter, these pressure
sensing elements are also mounted on the side of the catheter.

5.3.2 Arterial Stenoses and Aneurysms

In patients with vascular disease, segments of an artery may become nar-
rowed due to fatty deposits and atherosclerosis, resulting in a stenosis. Here,
the Bernoulli equation can be applied to the flow conditions to estimate the
effects of this narrowing. If, for example, a cross section of an artery is
stenosed such that the diameter at that cross section is less than its normal
value (Figure 5.3), then from the principle of Conservation of Mass

A1V1 = A2V2

2
a1
P1 a2 P2
V1 V2

FIGURE 5.3
Application of Bernoulli equation to an arterial stenosis.
7328_C005.fm Page 171 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 171

where subscript 1 denotes a cross section upstream to the stenosis and

subscript 2 at the narrowest site of the stenosis. If the cross sections under
consideration are relatively close to each other, we can neglect the effect of
viscous dissipation and apply the Bernoulli equation. Over a short distance,
the gravitational changes will also be negligible and the relationship in
Equation 1.52 will reduce to

ρV12 ρV 2
p1 + = p2 + 2 (5.2)
2 2

Since p1, V1, and V2 are known, the pressure at the site of constriction, p2,
can be computed. To express the kinetic energy per unit volume (also in
terms of mmHg), the conversion factor 1 g/cm-s2 = 7.5 × 10−4 mmHg can be
used.

Example:
Consider a case where there is a focal stenosis of a 6 mm diameter femoral
artery in which its cross-sectional diameter is reduced to one-third of normal.
Then, the velocity at the stenosis, V2, will be nine times the upstream velocity,
V1 . Furthermore, if the flow rate through the artery were 50 cc/min, then
the velocities, V1 and V2, are:

V1 = (50 cc/min)/(60 * π (0.6 cm)2/4) = 2.95 cm/s

V2 = 9 * V1 = 26.5 cm/s

Therefore, if the pressure at the upstream was 100 mmHg, then the pres-
sure at the stenosis will be reduced by

ρ/2 [V22 – V12] = 9300 g/(cm-s2) = 0.29 mmHg

One result of a lower pressure at a constriction is that under severe con-

ditions, the vessel walls may actually cave in under sufficient external pres-
sure, even to the point of complete occlusion. In this case, the flow velocity
will slow down due to frictional resistance and the kinetic energy will be
converted to pressure. Then, the pressure at the constriction will once again
increase, reopening the artery. This phenomenon will repeat itself cyclically,
resulting in the phenomenon of arterial flutter. This is actually the mecha-
nism behind the Korotkov sounds that are used with the traditional blood
pressure measurement made with an inflated cuff around the arm and sensed
by a stethoscope.
A converse condition can occur where there is a weakening of the arterial
wall and a corresponding increase in the lumen cross section, which causes
a bulge, or an aneurysm, to occur. While it is not certain as to the exact cause
of this defect (there is some evidence that it results from an excess of elastase
7328_C005.fm Page 172 Monday, September 18, 2006 12:20 PM

172 Biofluid Mechanics: The Human Circulation

in the blood, an enzyme, which breaks down the elastin in the artery wall),
it is commonly seen in the distal, or abdominal, aorta where it is referred to
as an Abdominal Aortic Aneurysm (AAA). Analysis performed on the
energy present in the flow similar to that above shows that the flow velocity
is reduced at the cross section of the aneurysm and that some of the kinetic
energy will be converted into pressure at that site. Although the increase in
pressure may not be significant in the resting state (< 5 mmHg), the corre-
sponding conversion from kinetic energy to pressure may be substantial
under exercise conditions when aortic velocity increases several fold. The
increase in pressure will lead to a corresponding increase in wall stress
(see Chapter 2, section 2.2: Analysis of Thin-Walled Cylindrical Tubes), which
may be sufficient to cause further expansion of the aneurysm cross section.
This sequence of events acts as a positive feedback loop in that further
enlargement of the artery reduces velocity and, again, increases static pres-
sure. Ultimately, this process may result in the bursting of the vessel at that
site. Obviously, this is never a desirable outcome, but when it occurs in the
major outlet vessel from the heart, it is particularly critical, being fatal in
~ 50% of cases.

5.3.3 Cardiac Valve Stenoses

Another application of the Bernoulli equation is in the analysis of natural
and prosthetic heart valves. In traditional engineering applications, this
analysis is based on flow through an orifice that is used for flow meters
placed in internal flows. An example of this is flow through a nozzle as
shown schematically in Figure 5.4. As the fluid passes through the sharp
edges of the nozzle, flow separation causes a recirculation region imme-
diately downstream of the nozzle. The fluid in the core region continues
to accelerate to form a contracted cross section (denoted by 2 in the figure)
referred to as the vena contracta. At this cross section, the streamlines
are parallel to the axis of the tube and the pressure over the cross section
is uniform.

Flow D1 D0 V2 D2
V1

CV

1 2

FIGURE 5.4
Schematic of the flow through a nozzle.
7328_C005.fm Page 173 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 173

Keeping in mind the assumptions used in deriving the Bernoulli equa-

tion and also neglecting the effect of gravity, the Bernoulli equation can
be written as

ρ 2 ρV 2  V 2 
p1 − p2 = (V2 − V12 ) = 2  1 − 12 
2 2  V2 

Also, from the Continuity Equation, we have

2 2
 V1   A2 
 V  =  A 
2 1

Substituting this relation, we get

ρV22   A2  
2
p1 − p2 = 1 −   
2   A1  
 

Solving the above equation, we can express the ideal velocity, V2ideal, as

2( p1 − p2 )
V2 ideal =
  2 
ρ 1 −  AA2  
  1 

The velocity computed using the above expression is “ideal” because we

used the Bernoulli equation and, thus, have neglected the effect of fluid
viscosity. As can be observed from Figure 5.4, the area at position 2, where
the streamlines are essentially parallel to the axis, is smaller than that at the
throat of the orifice. The area A2 can be related to the throat area A0 by
introducing a contraction coefficient, Cc, such that

A2
Cc =
A0

Then the expression for V2ideal becomes

2( p1 − p2 )
V2 ideal =
  2 
ρ 1 − Cc2  AA0  
 1
 
7328_C005.fm Page 174 Monday, September 18, 2006 12:20 PM

174 Biofluid Mechanics: The Human Circulation

Due to frictional losses, the velocity at position 2 will be less than the ideal
value given by the above expression. Hence, a velocity coefficient, Cv , is
introduced such that

V2 actual
CV =
V2 ideal

Then, the actual volume flow rate can be computed as

Q = A2V2 actual = Cc A0CVV2 ideal

or

2( p1 − p2 )
Q = A0CcCV
  2 
ρ 1 − Cc2  AA0  
 1
 

Squaring the expression and rewriting, we obtain

  2 
ρQ  1 − C 2  A0 
 
 A1 
2 c
p1 − p2 =  
2  A0 Cc CV 
2 2 2

 

or

ρQ 2 1
p1 − p2 =
2 A02Cd2

where

CcCV
Cd =
 2
1 − CV2  AA0 
 1

is defined as the discharge coefficient.

The discharge coefficient Cd depends on the geometry of the nozzle as well
as the dimensions of the tube and the throat of the nozzle. In flow meter
applications, the discharge coefficient is determined experimentally. In clin-
ical practice, this expression is applied to determine the effective orifice area,
EOA, of heart valves in the fully open position. In terms of the area at the
throat, we get

Qm ρ
EOA = A0 = (5.3)
Cd 2 ∆p
7328_C005.fm Page 175 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 175

where Qm is the mean flow rate and ∆p = p1 − p2.

In the case of the effective orifice area for aortic valves, Qm is replaced by
the mean systolic flow rate (as opposed to the cardiac output). Equation 5.3
was employed in the 1950s to estimate the orifice area of human heart valves
and is commonly referred to as the Gorlin equation in the clinical literature.
For the aortic valve, the Gorlin equation is given by

MSF
AVA = (5.4)
44.5 ∆pm

where AVA is the aortic valve area, MSF is the mean systolic flow rate, and
∆pm is the mean pressure drop across the valve.
The constant, 44.5 (cm/s)/((mmHg)1/2) takes into account the conversion
of units between the mean systolic flow rate (cm3/s) and the mean systolic
pressure drop (mmHg) and it also includes an assumed discharge coefficient
to compute the orifice area in cm2. The corresponding Gorlin equation for
the mitral valve is given by

MDF
MVA = (5.5)
31.0 ∆pm

where the mean diastolic flow rate (cm3/s) and the mean diastolic pressure
gradient (mmHg) are used.
Measurements of the flow rate and the pressure drop across the valves
can be used to predict the effective valve orifice areas of the natural valves
suspected of being stenotic. Such information is useful for cardiac surgeons
in deciding when to replace diseased valves. For normal aortic and mitral
valves, a discharge coefficient close to unity is assumed. In the case of
natural heart valves with centralized flow and no obstructions, the analogy
with the flow through a nozzle may be reasonable. This formula is also
extensively used to predict the effective valve orifice area of prosthetic
valves. However, due to obstruction of the occluders, especially with
mechanical valves, the discharge coefficient can be expected to be signifi-
cantly different from that of natural valves. The common practice is to
perform in vitro experiments in which the actual effective orifice area can
be measured and to determine the discharge coefficient for each type of
prosthetic valve.
The expression for the effective orifice area derived above is based on
steady flow across a nozzle and, thus, is applied assuming constant systolic
flow conditions. In reality, however, even during this period when the valve
is fully open, the blood flow goes through acceleration and deceleration
phases so that the flow is time-dependent. Therefore, a more rigorous
analysis should include the acceleration of the fluid as well as viscous
7328_C005.fm Page 176 Monday, September 18, 2006 12:20 PM

176 Biofluid Mechanics: The Human Circulation

dissipation. In one-dimensional analysis, the pressure drop across an orifice

can be written as (Young, 1979)

dQ
∆p = A + BQ 2 + CQ (5.6)
dt

where Q is the flow rate, dQ/dt represents the temporal acceleration, BQ2
represents the convective acceleration and CQ represents the viscous
dissipation.
The inertial term (time dependence) can be eliminated by averaging
Equation 5.6 over the forward flow interval (time during which the valve
is open), resulting in

∆pm = BQm2 + CQm (5.7)

or by taking the measurements at the time of peak flow when dQ/dt will be
identically equal to zero, which results in

∆pp = BQp2 + CQp (5.8)

In this relationship, the subscript p denotes that the values are measured
at the instance of peak flow through the orifice. Performing a dimensional
analysis on the important parameters for flow through an orifice — pressure p,
flow rate Q, orifice area A, density ρ, and viscosity µ — shows that the
constants B and C can be related to

ρ
B=
A2

and

µ
C= 3
A 2

respectively. Hence, the relationships for the peak and mean pressure drops
can be rewritten as

ρ 2 µ
∆pm = k1 Q +k Q (5.9)
A2 m 2 A 3 2 m

and

ρ 2 µ
∆pp = k1 Q + k 2 3 Qp (5.10)
A2 p A 2
7328_C005.fm Page 177 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 177

Substituting typical physiological values for ρ, µ , A, and Q into the above

equations and performing an order of magnitude study, it can be observed
that the inertial term in both the equations (first term on the RHS) is three
orders of magnitude larger than the viscous effect (second term on the RHS).
Thus, the viscous effects can indeed be neglected in flow through the valves.
By computing the mean values during the forward flow duration, the effec-
tive orifice area (EOA) relationship Equation 5.3 will reduce to the form

ρ
EOA = KQrms (5.11)
∆pm

The difference between the above relationship and the Gorlin equation
derived earlier is the root mean square flow rate used here instead of the
mean flow rate used in the Gorlin equation. The relationship using the peak
flow will be given by

ρ
EOA = KQp (5.12)
∆pp

where Qp is the peak flow rate and ∆pp is the pressure drop across the valve
at the instant of peak flow rate. Use of the above two relationships rather than
the Gorlin equation provides more accurate values for the valve orifice area
from the rigorous fluid mechanical analysis discussed above.
A dimensional analysis of steady flow through an orifice has shown that

d 
Cd = f  , Re
D 

where d is the orifice diameter, D is the diameter of the pipe, and Re is the
Reynolds number = ρVD µ ( )
. For a given valve geometry, the only variable in
the above relationship is the flow rate and, hence, it has been proposed that

kQ
EOA = (5.13)
Cd (Q) ∆p

where EOA represents the effective orifice area and Cd(Q), the discharge
coefficient as a function of flow rate Q. Using an in vitro experimental setup
in which the orifice areas of prosthetic valves of various geometries were
measured using planimetry over a range of flow rates, studies have demon-
strated that when the discharge coefficient was expressed as a linear function
of flow rate, such as

Cd (Q) = C1Q + C2 (5.14)

7328_C005.fm Page 178 Monday, September 18, 2006 12:20 PM

178 Biofluid Mechanics: The Human Circulation

the prediction capability of the effective valve orifice areas was significantly
improved. Further experimental studies are necessary in order to be able to
afford a more accurate clinical prediction of the effective valve orifice areas
in order to improve diagnostic capabilities.

5.4 Rigid Tube Flow Models

The simplest model for blood flow through a vessel would be steady, fully
developed flow of a Newtonian fluid through a straight cylindrical tube of
constant circular cross section. Such flows are characterized as Poiseuille
flow in the honor of J.L.M. Poiseuille (1799–1869), who performed experi-
ments relating pressure gradient, flow, and tube geometry in such a model.
From his experiments, he empirically derived the relationship given below:

K ∆pD4
Q= (5.15)
L

where Q is the flow rate, ∆p is the drop in pressure in a tube of length L,

and diameter D. K denotes a constant, which was found to be independent
of the other variables. G.H.L. Hagen (1797–1884) working independently,
arrived at the theoretical solution for the above problem that introduced
fluid viscosity as follows:

πR 4 ( p1 − p2 )
Q= (5.16)
8µL

where µ is the coefficient of viscosity and a is the tube radius. From Equation
5.15 and Equation 5.16, it can be observed that K = π 128µ . Thus, the relation-
ship in Equation 5.2 is also referred to as the Hagen–Poiseuille law. We have
already derived the above relationship in Chapter 1 from the Navier–Stokes
equation (Equation 1.72). We also utilized this relationship considering the
forces acting on a volume element of the fluid in Chapter 4 in our discussions
on the principles of the capillary viscometer. Several assumptions were made
in deriving the relationship for the flow rate given in Equation 5.16 and we
should critically examine the validity of these assumptions in models
describing blood flow in arteries. The assumptions are:

1. Newtonian fluid: The governing equations used in deriving the

expression for the velocity profile assumed that the fluid is Newtonian
with a constant viscosity coefficient. When we discussed the vis-
cous behavior of blood in Chapter 4 (section 4.1.3: Viscous Behavior
of Blood), we concluded that the rheology of blood can best be
7328_C005.fm Page 179 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 179

described by Casson’s relationship and that blood exhibits nonlin-

ear shear stress vs. rate of shear characteristics, especially at low
rates of shear. However, it was also determined that at relatively
high rates of shear, the viscosity coefficient asymptotically
approaches a constant value. Thus, for flow in large blood vessels,
where relatively large shear rates can be expected during systole, a
Newtonian description appears to be reasonable.
2. Laminar flow: The governing equations also assume that the flow
regime is laminar. For steady flow in cylindrical pipes, we can define
a critical Reynold’s number, Rec, beyond which the flow can be
considered to be turbulent. In a typical human aorta, one can esti-
mate the Reynolds number assuming a diameter at the root of the
aorta of 2.5 cm and a mean time averaged flow velocity of 20 ml/s
(based on a cardiac output of 6 lpm) and compute a magnitude of
about 1500. Whole blood density of 1.056 gm/cc and a viscosity
coefficient of 0.035 Poise were used in this computation. Thus, the
time averaged Reynolds number is observed to be well below the
critical value of about 2100. If a peak flow rate of 20 l/m is assumed
during systole, then the Reynolds number during that part of the
cardiac cycle is about 5100. However, the human aorta is a disten-
sible vessel with a complex geometry, so the critical Reynolds num-
ber determined from experiments in rigid straight cylindrical pipes
is not applicable in this situation. In vivo velocity measurements
using hot film anemometry have indicated disturbed flow during
the deceleration phase of the cardiac cycle although there is no
experimental evidence of sustained turbulence in the human circula-
tion (in the absence of any diseased states such as valvular or arterial
stenoses). Thus, the assumption of laminar flow in the model also
appears to be reasonable.
3. No slip at the vascular wall: The innermost lining of the arterial wall
in contact with the blood is a layer of firmly attached endothelial
cells and it appears to be reasonable to assume no slip at the
interface.
4. Steady flow: As pointed out earlier, the steady flow model is the
simplest to deal with mathematically and, therefore, was used in
deriving the above relationship. Based on this assumption, we
neglected the inertial forces in the simplifications of the governing
equations. However, as discussed in the review of cardiovascular
physiology (Chapter 3), flow through human arteries is clearly pul-
satile, consisting of systolic and diastolic phases; therefore, the
assumption of steady flow is not valid in the major part of the
circulatory system.
5. Cylindrical shape: The tube model discussed above assumed a circular
cross section and axisymmetry. Even though this geometry may be a
good approximation for most of the arteries in the systemic circulation,
7328_C005.fm Page 180 Monday, September 18, 2006 12:20 PM

180 Biofluid Mechanics: The Human Circulation

the veins and the pulmonary arteries are more elliptical in shape. The
arteries also have a taper with their cross sections narrowing with
distance downstream. Thus, the general assumption of a circular cross
section without taper is a deviation from reality.
6. Rigid wall: As was discussed in the previous chapter (Chapter 3,
section 3.4: Systemic Circulation), the arterial walls are visco-elastic
and distend with the pulse pressure. The interaction between the
flowing blood and the distensible arterial wall is an important factor
in the description of the flow dynamics. Thus, the assumption of
rigid walls in the model is also not valid. However, for the special
case of steady flow models in the circulatory system, distensibility
of the vessels will not affect the solution.
7. Fully developed flow: The model described above also assumes that
the flow is fully developed, which implies that the velocity profile
remains the same at any cross section with distance downstream.
However, as the blood leaves the ventricle through the aortic valve,
the velocity profile in the aorta is relatively flat and a finite length
is needed before the flow may become fully developed (as
described in the next section). Similarly, even in the distal arteries,
flow passes through several branching points and curved arterial
sections. At each location where the cross-sectional geometry devi-
ates from a straight arterial segment, the flow will also be appro-
priately altered. Thus, the assumption of fully developed flow is
also not valid.

It can be observed that in applying the steady flow models to describe

blood flow in the circulation, the assumptions of steady flow, wall rigidity,
cylindrically shaped lumens, and fully developed flow are all clearly violated.
To deal with this, we will discuss unsteady flow models in distensible vessels
in the next chapter. In order to analyze the effect of complex noncylindrical
geometry as well as the developing nature of flow, experimental measure-
ments or use of numerical solution methods, such as a finite element tech-
nique, are necessary. At this point, however, we will use the basic relationship
obtained with this simplified steady flow model in order to introduce some
applications in the flow through the arterial system.

5.4.1 Vascular Resistance

We introduced the concept of resistance to flow in the chapter on cardio-
vascular physiology (Chapter 3). The vascular resistance is given by the
relationship

∆p
R= (3.3)
Q
7328_C005.fm Page 181 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 181

R1 R2 R3
RT = R1 + R2 + R3

R1

R2
1/RT = 1/R1 + 1/R2 + 1/R3

R3

FIGURE 5.5
Total resistance with tubes in series or parallel configuration.

This expression is analogous to the electrical resistance given by

E
R= (5.17)
I

where I is the current and E the voltage across a segment of a circuit. If the
pressure drop is measured in terms of mmHg and the flow rate in terms of
cc/s, then the resistance is expressed as mmHg-s/cm3 or a peripheral resis-
tance unit (PRU) and it is used in physiological literature. From the Poiseuille
expression for flow rate through a tube, we obtained the relationship

8µL
R= (3.4)
πR 4

When the vessels are in series, the total resistance will be the sum of the
individual resistances. When the vessels are in parallel, the total resistance
across the vessels can be computed using the relationship shown in Figure 5.5.
This formula for the vascular resistance, derived from the steady, fully
developed flow relationship, can be used to estimate the resistance in seg-
ments of the vascular system. For example, the mean pressures at the aortic
root and at the terminal end of the vena cavae can be measured along with
the time averaged flow rate through the systemic circulation. These data
can be used to compute the resistance in the systemic circulation as a mea-
sure of the functioning of the circulatory system. Similarly, the pulmonary
vascular resistance can also be determined by measurement of the corre-
sponding pressures. The expression for the resistance (Equation 3.4) shows
that it is inversely proportional to the fourth power of the radius and, thus,
a small change in the radius of the vessel will considerably affect the resis-
tance to flow. The implication of this is that the autonomic nervous system
in the body controls the tension of the smooth muscles in the vessel wall in
7328_C005.fm Page 182 Monday, September 18, 2006 12:20 PM

182 Biofluid Mechanics: The Human Circulation

the arterioles. Thus, with the alteration of the muscle tension, the arterioles
can be distended or contracted selectively to control the amount of blood
flow into the various segments of the body.
Even though the measurement of vascular resistance yields information
about the state of the circulatory system and, hence, can be used as a
diagnostic parameter, several limitations must be kept in mind on the
information provided: (1) It does not indicate which pathway between the
two points of measurement is constricted or dilated; (2) it does not indicate
the cause for the change (due to nervous stimulation, increased transmural
pressure or other causes); (3) it yields information only on net changes and
does not indicate local changes; and (4) it does not provide information to
distinguish between dilation of the vessels or the opening of new vessels
(Angiogenesis).

5.5 Estimation of Entrance Length and Its Effect

on Flow Development in Arteries
As pointed out in Chapter 1 (section 1.8.1.3: Flow Development), as fluid
enters a pipe from a reservoir, the velocity profile will be relatively flat
and the fluid must pass through a finite length of the tube before the
velocity profile attains a final shape. This observation is represented in
Figure 5.6.
At the tube entrance, the fluid coming in contact with the tube wall will
be forced to have zero velocity (i.e., the “no slip” condition) due to frictional
forces and a gradient in velocity is established in the radial direction. Further
downstream, more and more fluid is retarded due to the radial penetration
of shearing effects of fluid adjacent to the wall. At the same time, the fluid
in the core region is actually accelerated by the unbalanced pressure and
viscous forces, which then maintains a constant flow rate at all cross sections
in the tube (i.e., satisfies the Continuity requirement). Consequently, the

U
d
x
Boundary-layer edge

FIGURE 5.6
Concept of entry length before the flow becomes fully developed.
7328_C005.fm Page 183 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 183

U
A2 du
u m
dy 2
δ du A1
m
dy 1 y2
y1

FIGURE 5.7
A force balance on a fluid element in the boundary layer.

initially blunt profile is progressively modified and, ultimately, a parabolic

velocity profile is established farther downstream. Thus, very near the
entrance, the radial distance in the fluid over which the viscous effects are
present is very small. However, as we proceed downstream, the viscous
effects have diffused further in the radial direction. The “thickness” within
which this diffusion has occurred is referred to as the boundary layer. At
some location downstream, the boundary layer grows to the point where it
reaches the centerline of the tube and, at this point, the flow is said to become
fully developed. In Chapter 1, we presented a relationship for the entrance
length in terms of tube diameter and the Reynolds number (Equation 1.78)
for laminar flow. We will now show the basis for such a relationship based
on the effect of inertial and viscous drag forces within the boundary layer.
A small fluid element within the boundary layer is considered in
Figure 5.7. Let the area A1 and A2 be equal to A. The tangential shear stresses
in those areas are µ du
dy
and µ du
dy
, respectively. The net viscous force on the
1 2
element will be the area times the change in stress with the distance y from
the wall and will be given by

d  du 
µ A( y2 − y1 )
dy  dy 

In the absence of flow acceleration, inertial forces acting on the element

must balance the viscous forces. However, a solution of the governing equa-
tions in the boundary layer is too complicated to be solved analytically. By
assigning suitable scales to the variables representing the two forces, an
estimate of the entrance length can be obtained. To estimate the viscous
forces, we use the boundary layer thickness δ at a distance X from the
entrance and the free stream velocity U in the above equation. Then the
viscous force is proportional to

µU
A( y2 − y1 )
δ2
7328_C005.fm Page 184 Monday, September 18, 2006 12:20 PM

184 Biofluid Mechanics: The Human Circulation

The inertial force acting on the element is given by

ρ (Convective acceleration) (Volume of the element)

If the boundary layer thickness is considered at a distance X from the

entrance of the tube, then the time scale for the fluid to reach that distance
(the convection time) is X/U. The convective acceleration at this location is
then proportional to U/(X/U), or, U2/X. Thus, the inertial force on the ele-
ment will be given by

U2
ρ A( y2 − y1 )
X

Equating these two forces, we obtain

U2 U
ρ = kµ 2
X δ

where k is the proportionality constant that can be determined from exper-

iments. Thus, the boundary layer thickness at any axial location can be
written as

µX
δ∝
ρU

From the above relationship, we can see that the boundary layer grows in
proportion to the square root of the distance. Also, the boundary layer
thickness decreases at any given location as the flow rate through the tube
increases, producing a corresponding increase in the free stream velocity.
The boundary layer will fill the tube and the flow will become fully devel-
oped (i.e., with no further convective acceleration in the fluid) when the
boundary layer thickness equals the radius, or δ = D/2, where D is the
diameter of the tube. Then, the entrance length will be given by

U
X = kD2
ν

or

 UD 
X = kD  (5.18)
 ν 

where the terms in the parenthesis represent the Reynolds number. In this
relationship, ν = µρ is referred to as the kinematic viscosity. Equation 5.18 can
7328_C005.fm Page 185 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 185

be used to predict the entrance length for steady laminar flow through a
straight pipe of circular cross section. The magnitude for the constant k has
been experimentally determined to be approximately 0.06. Thus, if an artery
was assumed to be a straight cylindrical tube with steady and laminar flow,
we would be able to estimate the distance at which the flow would become
fully developed using the expression given in Equation 5.18. Note that the
relationship in Equation 5.18 is valid only for flows with Reynolds numbers
greater than 50. In cases where the Reynolds number is close to zero (e.g.,
in capillaries where the Reynolds number is ≤ 0.01), the entrance length
becomes a constant of 0.65D.
In arterial flow, flow development is affected by a number of factors. In
large arteries, for example, the entrance length is relatively long since it
depends not only on vessel diameter but also on the Reynolds number
(Equation 5.18), both of which are large. This produces a situation in which
a large portion of most major arteries is exposed to developing flow with
higher velocity gradients near the wall. Secondly, the axial velocity profile
becomes skewed at sites of curvature and bifurcation of the arteries
(see Chapter 6, section 6.6: Flow Through Curved Arteries and Bifurcations)
with higher velocity (and, thus, a higher velocity gradient) toward one wall
and a lower velocity toward the opposite wall. One of the implications of
this is that the intimal lining of arteries is exposed to higher shear forces
proximally and lower shear forces distally. Also, there is a general tendency
for higher shear forces on the outer wall of curvatures and branch inlets and
on the flow divider of bifurcations. In terms of the boundary layer, it will
be thinner in regions of high velocity compared to regions with a low veloc-
ity. The size of the boundary layer is also important in terms of mass trans-
port of molecules (i.e., gas and nutrient) between the blood and artery wall
since diffusive effects are more important than convective effects in large
boundary layer regions. This is the basis for one theory of atherosclerosis
(see Chapter 6, section 6.4: Hemodynamic Theories of Atherosclerosis) in
that certain molecules, such as low-density lipoproteins (LDL), may accu-
mulate within regions of thicker boundary layers and tend to stay in that
region for longer time, enhancing their diffusive transport to the subendo-
thelial region and initiating atherosclerotic lesions.

5.6 Flow in Collapsible Vessels

In the Hagen–Poiseuille relationship derived for flow through conduits
(Chapter 1, section 1.8: Flow Stability and Related Characteristics), we
observed that the flow through an individual conduit is proportional to the
pressure drop across the system when the flow is laminar. Even when
the flow is turbulent, the flow rate will monotonically increase with an
increase in pressure drop, but not in a linear fashion. In the circulatory
7328_C005.fm Page 186 Monday, September 18, 2006 12:20 PM

186 Biofluid Mechanics: The Human Circulation

h P′
(cm) (mm
Hg) Q = 1.0
15 0

(9) (3)
10 4 1 –5

5 8

Q = 1.0
Q = 2.0
0 12 12 9 3 0
(12) (9) (3) (0)

Arteries Capillaries Veins

FIGURE 5.8
Pressure and flow through rigid pipe segments. (Redrawn from Milnor, W.R. (1989) Hemo-
dynamics, Williams and Wilkins, Baltimore. With permission.)

system, however, the conduits are flexible and, in some instances, the trans-
mural (i.e., across the vessel wall) pressure can cause significant collapse of
the conduit. This is especially true downstream of a stenosis where the
pressure in the conduit can drop below that of the extramural (i.e., outside
the vessel wall) pressure. In such cases, the flow through the conduit is no
longer dependent upon the upstream, p1, and downstream, p2, pressure
difference, but rather on the difference between the upstream pressure and
the pressure surrounding the conduit, pe. When this happens, flow through
the conduit can remain relatively constant even though the downstream
pressure, p2, varies widely. This phenomenon is variously described as flow
limitation, Starling resistor phenomenon, sluice effect, and waterfall
effect. It has been observed in blood flow situations such as from extratho-
racic to intrathoracic veins, diseased coronary arteries, pulmonary blood
flow, flow through cerebral vessels, and in urine flow.
Let us consider a system of blood vessels as shown in Figure 5.8 where all
the vessels are assumed to be rigid. Pressure at the inlet of the channel is
assumed to be 12 mmHg and the pressure drop across the arteries, capillaries,
and veins is assumed to be 3 mmHg, 6 mmHg, and 3 mmHg, respectively.
These pressure drops are due to viscous dissipation as the fluid flows across
the vessel. A total flow rate of 2 ml/s is assumed to be evenly divided
between the two segments. The vertical distance between the vessels and
the corresponding hydrostatic pressure difference are also shown on the scale
on the left of the figure. The resulting pressures, which include the hydro-
static effect, are given inside the vessels and the pressures due to the effects
7328_C005.fm Page 187 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 187

H P′
(cm) (mm
Hg)
15 0 Q = 0.43

(10.7) (8.1)
–6.7
10 4 2.7

0.1
5 8

Q = 1.43
0 12 12 9 3 Q
(12) (9) (3) (0)

Q = 1.0

FIGURE 5.9
Effect of vessel collapsibility on the pressure and flow. (Redrawn from Milnor, W.R. (1989)
Hemodynamics, Williams and Wilkins, Baltimore. With permission.)

of viscous losses alone are given at the outside. If the vessel in the upper
segment is not rigid, but is made of a thin elastic material, it will collapse
due to the low transmural pressure and, thus, the flow would cease in the
upper segment. However, as soon as the flow ceases, the pressure would
rise to 4 mmHg due to the static conditions and the vessel would once again
open. Thus, this sequence of events will be repeated continuously and has
been demonstrated in in vitro experiments. However, in vivo, thin-walled
vessels will tend to attain an equilibrium state as shown in Figure 5.9 (Milnor,
1989). Assume here that the capillary segments collapse to a narrow lumen
at a transmural pressure of 0.2 mmHg and completely collapse at 0 mmHg.
Pressure and flow through the upper channel will reach a state of equilibrium
at about 0.1 mmHg transmural pressure with a lower flow through the upper
segment, as shown in Figure 5.9.
The phenomenon described above can be used to explain flow through
the lung capillaries. When a person is standing, little flow occurs through
the capillaries in the apices of the lungs due to the collapse of the blood
vessels. As discussed earlier, the flow rate through a rigid tube is dependent
on the distal pressure, which is not necessarily true in a collapsed vessel as
discussed below. Figure 5.10 shows two rigid vessels connected by a collaps-
ible vessel. The collapsible vessel is enclosed in a box so that the pressure
outside the vessel can be independently set. This arrangement is also referred
to as Starling’s resistor. The collapsible segment is assumed to be fully open
at a transmural pressure of 0.2 mmHg and to be completely closed at 0 mmHg.
The pressures at various points in the system (in mmHg) are included in the
7328_C005.fm Page 188 Monday, September 18, 2006 12:20 PM

188 Biofluid Mechanics: The Human Circulation

P′
h (mm
(cm) Hg) (Pa – Pv)(mm Hg)
15 0 20 15 10 5 0

10 4 15

Q 5 8

Flow (ml/min)
4 10
12 10 –6 0 12
a b c
5
–5 16
–8
d –10 20 0
–5 0 5 10
Pv (mm Hg)

FIGURE 5.10
Pressure and flow through a collapsible tube. (Redrawn from Milnor, W. R. (1989) Hemodynam-
ics, Williams and Wilkins, Baltimore. With permission.)

figure. The flow through the vessel will reach equilibrium when the pressure
inside the vessel is 4.1 mmHg representing a transmural pressure of 0.1
mmHg. Changing the distal pressure at point “c” by raising or lowering the
outflow beaker will not alter the flow through the vessel provided it is not
raised above 4.2 mmHg. However, changing the pressure within the box will
affect the flow. The interaction between the variables is represented in the
curve shown in the right half of the figure.
When the outlet pressure Pv is equal to the inlet pressure Pa, there isn’t
any flow through the vessel. Gradually decreasing the outlet pressure results
in a linear increase in the flow through the vessel until the outlet pressure
equals the pressure in the box surrounding the vessel. Further decrease in
the outlet pressure does not increase the flow through the vessel even though
the gradient Pa – Pv continues to increase. The model described above is
similar to that occurring in the lungs. Here, the pressure in the box would
depict the alveolar pressure and, due to the low pressure in the vessels in
the lung circulation combined with the hydrostatic effects, the capillary acts
like a sluice gate controlled by the arterial pressure and the alveolar pressure.
Such a phenomenon where the flow is independent of the downstream
pressure is also described as the “vascular waterfall.” The physics of this
phenomenon, in which the mechanism by which the flow becomes indepen-
dent of the variation in the downstream pressure, is explained by the
“inertial” and “frictional” mode of flow limitation. In the ‘inertial’ mode
explanation, the upstream and downstream pressure is assumed to be decou-
pled at a point in the converging (collapsed) conduit at which the flow has
accelerated to a velocity that exceeds the pressure-wave propagation velocity
of the system at that point. Pressure disturbances distal to the ‘‘choke point’’
7328_C005.fm Page 189 Monday, September 18, 2006 12:20 PM

Static and Steady Flow Models 189

cannot be propagated upstream since the medium is moving in the opposite

direction with a velocity greater than the pressure wave propagation velocity.
In the mechanism due to the frictional mode of flow limitation, it is suggested
that changes in the upstream pressure alter the pressure distribution in the
adjacent segment of the conduit to cause a compensatory change in the
flow resistance in a direction to maintain a constant flow through the
conduit.

5.7 Summary
In this chapter, we used simple steady flow models to derive some relation-
ships between the flow and pressure in the circulatory system and discussed
some clinical applications based on the simplified models. We will now
consider the effect of unsteady flow and the distensibility of the blood vessels
as we proceed on in following chapters to more realistic models to describe
blood flow dynamics in the arterial system.

5.8 Problems
5.1 For a typical human, the diameter at the root of the aorta is 2.5 cm, the
time averaged flow rate (Cardiac Output) is 5.5 lpm, and the peak flow rate
during systole is 20 lpm.

1. Calculate the Reynolds number based on the time-averaged flow

rate as well as the peak flow rate assuming a Poiseuille flow
relationship.
2. Compute the corresponding wall shear stress in the aorta.
3. Assuming time-averaged flow rates of 0.6 lpm in the carotid artery
(diameter ≈ 0.8 cm) and a flow rate of 0.3 lpm for a femoral artery
(diameter ≈ 0.5 cm), calculate the Reynolds number and wall shear
rate in these vessels also.
4. Do you think that the flow in the aorta becomes turbulent based on
the calculated Reynolds numbers?

5.2 Assume that the blood is flowing through an aorta of 1.0 cm in diameter
at an average velocity of 50 cm/s. Let the mean pressure in the aorta be
100 mmHg. If the blood were to enter a region of stenosis where the diameter
of the aorta is only 0.5 cm, what would be the approximate pressure at the
site of narrowing? (Assume blood density to be 1.056 g/cc: 1 g/cm-s2 = 7.5 ×
10−4 mmHg.)
7328_C005.fm Page 190 Monday, September 18, 2006 12:20 PM

190 Biofluid Mechanics: The Human Circulation

5.3 Reconsider the above problem where the blood enters an aneurysm
region with a diameter of 1.5 cm.

1. Determine the pressure in the aneurysm.

2. If for conditions of vigorous exercise, the velocity of blood upstream
of the aneurysm was four times the normal value (200 cm/s), what
pressure would develop at the aneurysm? (Assume blood density
to be 1.056 g/cc: 1 g/cm-s2 = 7.5 × 10-4 mmHg.)

5.4 Estimate the pressure difference between the inlet and outlet of a capillary
that would be needed if blood is flowing through the capillary with a velocity
of 0.2 cm/sec. Assume that (1) the capillary diameter is 8 mm, (2) the capillary
length is 200 mm, and (3) the apparent viscosity of blood is 2.5 cP. The friction
factor for laminar flow in tubes can be expressed as f = 16/Re, where Re is the
Reynolds number and f = ρRV∆2pL .
5.5 Assume that blood entering the aorta from the left ventricle has a steady
state mean velocity of 30 cm/s. If the diameter of the aorta at the entry is
1.2 cm, estimate the distance in the aorta that the blood has to travel before
the flow becomes fully developed. What are the assumptions made in the
formula used in computing this entry length, which is not realistic in the
human circulation?
7328_C006.fm Page 191 Monday, October 16, 2006 10:51 AM

6
Unsteady Flow and Nonuniform
Geometric Models

CONTENTS
6.1 Introduction ................................................................................................191
6.2 Windkessel Models for the Human Circulation...................................192
6.3 Continuum Models for Pulsatile Flow Dynamics ...............................195
6.3.1 Wave Propagation in the Arterial System.................................195
6.3.1.1 Moens–Korteweg Relationship.....................................196
6.3.1.2 Womersley Model for Blood Flow:
Including Viscosity Effects ............................................205
6.3.1.3 Wave Propagation in an Elastic
Tube with Viscous Flow ................................................213
6.4 Hemodynamic Theories of Atherogenesis ............................................222
6.4.1 Low Pressure, Low and High Wall Shear
Stress Theories ...............................................................................223
6.4.2 Time Varying Wall Shear Stress, Oscillatory
Shear Index, and Wall Shear Stress Gradients .........................227
6.5 Wall Shear Stress and Its Effect on Endothelial Cells .........................228
6.6 Flow through Curved Arteries and Bifurcations .................................229
6.6.1 Curved Vessels...............................................................................231
6.6.2 Bifurcations and Branches ...........................................................236
6.7 Flow through Arterial Stenoses and Aneurysms .................................241
6.8 Summary .....................................................................................................254
6.9 Problems......................................................................................................254

6.1 Introduction
Even though the steady flow models considered in the previous chapter
provided us with some insight on the flow through arteries and we consid-
ered some applications using the same models, more realistic ones need to
take into account the unsteady nature of flow through the circulatory system.

191
7328_C006.fm Page 192 Monday, October 16, 2006 10:51 AM

192 Biofluid Mechanics: The Human Circulation

In the unsteady models, the effect of pressure pulse propagation through

the arterial wall, the effect of inertial forces due to the acceleration and
deceleration of blood, and the effect of distensibility of the arteries on blood
flow will be considered. Some applications of the unsteady flow models on
the understanding of the blood flow dynamics in the circulatory system will
be discussed.

6.2 Windkessel Models for the Human Circulation

Early theories to describe the blood flow in circulation considered the arterial
system to be elastic storage vessels, which transformed the discontinuous
flow due to the pumping of the heart into steady flow in the peripheral
organs. The theory based on the concept that the blood vessels act as elastic
storage vessels is referred to as the Windkessel model. This name arose from
the analogy of early fire engines where the repeated strokes of the water
pump were smoothed out to provide a nearly continuous flow by an air
compression chamber (Windkessel, in German).
The Windkessel theory considers the arteries as a system of intercon-
nected tubes with a storage capacity. A schematic representation of the
Windkessel model for blood flow is shown in Figure 6.1. The fluid is
pumped into the Windkessel chamber intermittently by the ventricular
ejection and the outflow at the other end is based on the pressure gradient
and the resistance to flow based on the Poiseuille theory. The storage
capacity of the elastic blood vessels will be given by the distensibility, Di,
such that

dV
Di = (6.1)
dp

where V is the volume and p is the pressure.

Inflow P Outflow
Q (t) V per. res.
Rs

FIGURE 6.1
Schematic representation of blood flow in the circulation based on the Windkessel theory.
7328_C006.fm Page 193 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 193

The bulk modulus or the modulus of volume elasticity, k, (see Chapter 1)

is related to distensibility by the relationship

Vdp V
k= =
dV Di

The rate of storage of volume in the elastic chamber d∨ dt can be written as

dV  dV  dp
=
dt  dp  dt

or

dV dp
= Di
dt dt

A mass balance can be written for the fluid in the elastic chamber as

Inflow — Outflow = Rate of storage

i.e.,

p dp
Q (t ) − = Di
Rs dt

The outflow is represented by the drop in pressure ( p − pV ) over the periph-

eral resistance Rs and, when the venous pressure, pV , is assumed to be close
to zero, this reduces to p R .
s
The simplest assumption on the inflow is:

Q(t) = Q0 0 ≤ t ≤ ts

and

Q(t) = 0 ts ≤ t ≤ T

where ts is the time at end systole and T is the duration of the cardiac cycle
and Qo is a constant. Then the equation for systole can be written as

dp p Q
+ = 0
dt Rs Di Di

with the initial condition p = p0 at t = 0.

7328_C006.fm Page 194 Monday, October 16, 2006 10:51 AM

194 Biofluid Mechanics: The Human Circulation

Integrating the above equation, we obtain

 t 
−
p(t) = RsQ0 − (RsQ0 − p0 )e  RsDi  (6.2)

During diastole, the equation reduces to the form

dp p
=−
dt Rs Di

with the condition that p = pT at t = T.

Integration of the above equation yields

−( t − T )
p ( t ) = pT e RsDi (6.3)

In this relationship, pT is the pressure at the end of diastole and T is the

duration of the cycle. Figure 6.2 shows a typical pressure pulse curve using
the Windkessel theory and an arterial pressure pulse curve is shown in the
inset. With the assumption of constant inflow, the stroke volume Vs will be
given by

Vs = Q0ts

I II

FIGURE 6.2
Pressure pulse plot from the Windkessel theory. II represents the time interval in which the
inflow stops and the pressure decays exponentially, given by the relationship in Equation 6.9.
An actual pressure recording is displayed in the insert. (Redrawn from Noordergraaf, A. (1978)
Circulatory Systems Dynamics (1978) Elsevier, Philadelphia. With permission.)
7328_C006.fm Page 195 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 195

Other patterns of inflow conditions with Q as a function of time (e.g., a

half sine wave) have also been considered using this Windkessel theory.
Then, the stroke volume during a cardiac cycle will be given by

ts

∫
Vs = Q ( t )dt
0

The stroke volume determined using this theory could also be used to
estimate the cardiac output. Even though the Windkessel theory is able to
estimate the pressure changes in the arterial system, there are several draw-
backs in applying this theory for understanding the flow dynamics in
circulation. The Windkessel theory assumes that the pressure pulse wave
propagates instantly throughout the arterial system and neglects the finite
time needed for the pulse wave transmission, as we will see later. Several
improved models using the Windkessel theory, incorporating traveling
waves and wave reflections, have also been considered. However, even with
such models, details of the velocity profiles and mechanical forces due to
the blood flowing at various segments in the circulatory system cannot
be studied.

6.3 Continuum Models for Pulsatile Flow Dynamics

6.3.1 Wave Propagation in the Arterial System
In order to describe the mechanical events due to pulsatile blood flow
through a segment of an artery, the laws of classical mechanics need to be
applied and solutions obtained from the governing equations describing the
motion. A number of theoretical and experimental studies have appeared in
attempting to describe pulsatile flow through the arterial system. We will
attempt to develop some of the basic models describing the pulsatile flow
in elastic vessels and discuss some of the applications to understanding the
flow through the human circulatory system under normal states and in the
presence of arterial disease.
The pressure pulse, generated by the contraction of the left ventricle,
travels with a finite speed through the arterial wall and the speed of trans-
mission is dependent on the wall elastic properties as well as the interaction
between the wall and the blood contained within. The pressure pulse also
changes shape as it travels downstream due to the interaction between the
forward moving waves and the waves reflected at discontinuities in the
arterial system, such as branching and curvature sites. We will develop some
relationships for the speed of wave propagation in the arterial system.
7328_C006.fm Page 196 Monday, October 16, 2006 10:51 AM

196 Biofluid Mechanics: The Human Circulation

h
r u(r, z, t)
R

z Flow w(r, z, t)

FIGURE 6.3
Schematic diagram of a model for wave propagation in a thin-walled elastic tube.

6.3.1.1 Moens–Korteweg Relationship

Consider an infinitely long, thin-walled elastic tube of circular cross section
as shown in Figure 6.3. We will consider the governing equations for the
fluid and the elastic tube in deriving the relationship for the wave propaga-
tion. Since the flow is axisymmetric, the velocity component in the φ direc-
tion as well as the change of other variables in the φ direction will be equal
to zero. Then, the continuity equation in the cylindrical coordinate system
(Chapter 1, Table 1.1b) will reduce to the form

1 ∂
r ∂r
( rVr ) + ∂∂Vzz = ∂∂Vrr + Vrr + ∂∂Vzz = 0 (6.4)

The momentum equations in the r and z directions can be rewritten as

 ∂V ∂V ∂V 
ρ  r + Vr r + vz r 
 ∂t ∂r ∂z 

∂p  ∂2V 1 ∂Vr Vr ∂2Vr 

= Fr − + µ  2r + − + 2 
∂r  ∂r r ∂r r 2 ∂z 

 ∂V ∂V ∂V 
ρ  z + vr z + v z z 
 ∂t ∂r ∂z 

∂p  ∂2V 1 ∂Vz ∂2Vz 

= Fz − + µ  2z + + 2 
∂z  ∂r r ∂r ∂z 

In these equations, the dependent variables (velocity components and the

pressure) are a function of r, z, and t. We will neglect the body forces (grav-
itational force) in the horizontal segment of the artery under consideration.
Initially, we will assume that the fluid is inviscid and, hence, the viscous
force terms will drop out from the above equations. However, the momen-
tum equations will still contain the nonlinear convective acceleration terms
and we will perform an assessment of the contributions due to the nonlinear
7328_C006.fm Page 197 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 197

terms in our analysis. We will write the estimated magnitude of each term
below the corresponding term in the continuity and momentum equations.
In the momentum equations, we will assume that the leading temporal
acceleration term has the predominant effect and compare the relative mag-
nitudes of the nonlinear convective acceleration terms with respect to the
leading term. We will scale the magnitude of the dependent variables in
order to determine the magnitude of each term as follows:

Vr → u, Vz → w , r → R , z → λ , and t→τ

If λ is the wavelength and τ is the period of the pulse wave through the
elastic tube, then the wave speed c will be given by

λ
c=
τ

We will also assume that the ratio of fluid velocity to the pulse wave
velocity is much smaller than unity, i.e., w c
1 .
The magnitudes of the terms in the continuity equation with Vφ neglected
will be:

∂Vr Vr ∂Vz
+ +
∂r r ∂z

u u w
~
R R λ
R
Multiplying each term with , we have
w

u u R
~
w w λ

We observe that the ratio u/w, which is the same order as that of R/λ, can
also be assumed to be much less than unity. Hence, the radial velocity
induced by the tube displacement in the radial direction can be expected to
be much smaller than the axial velocity magnitudes.
Let us consider the magnitude of the inertial terms in the r-direction
momentum equation

∂Vr ∂V ∂V
+ Vr r + Vz r
∂t ∂r ∂z

u u2 wu
~
τ R λ
7328_C006.fm Page 198 Monday, October 16, 2006 10:51 AM

198 Biofluid Mechanics: The Human Circulation

τ
Multiplying each term with , we get
u

τu wτ
1
R λ

u w w
1 λ
w Rτ τ

u R R λ
Since ∼ and ∼ ∼ c , we have
w λ τ τ

Rw w
1
λ c c

Similarly, in the z-direction momentum equation, comparing the magnitudes

of the inertial terms

∂Vz ∂V ∂V
+ Vr z + Vz z
∂t ∂r ∂z

w uw w2
~
τ R λ
τ
Multiplying each term with , we get
w

u w w
1 λ
w Rτ τ

and, hence,

Rw w
1
λ c c

Comparing the order of magnitude of the convective acceleration terms

with the temporal acceleration term, we can neglect the nonlinear terms as
a first approximation since Rλ << 1 and wc << 1. Thus, we conclude that the
nonlinear convective acceleration terms have a smaller effect on the flow
dynamics compared to the temporal acceleration term and can be neglected
as a first approximation. Then, the momentum equations in the r- and z-
directions reduce to the form

∂u ∂p
ρ =− (6.5)
∂t ∂r
7328_C006.fm Page 199 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 199

and

∂w ∂p
ρ =− (6.6)
∂t ∂z

If we assume that the pressure gradient in the axial direction  ∂∂pz  is the
dominant driving force inducing the flow, we can assess the magnitude of
the radial pressure gradient, ∂∂pr , with respect to the axial pressure gradient.
( )
In Equation 6.4, the left-hand side term is ∼ ρ uτ and the right-hand side
term is ∼  Rp  . In the axial direction, the left-hand side term of Equation 6.5
is ∼ (ρ wτ ) and the right-hand side term is ∼ λp .
Taking the ratio of magnitudes of these two equations, we get

ρ uτ p
= p
R
ρ wτ λ

Since the axial pressure gradient is the most important term, we assign a
value of unity for that term and get

ρ uτ Rp
=
ρ wτ 1

or

p u R
= ∼
R w λ

Since Rλ
1, we can neglect the gradient of pressure in the radial direction.

In other words, the pressure is assumed to be a constant in a cross section
from the order of magnitude study. Thus, p is a function of z and t only.
The flow rate through the tube will be given by

∫
Aw ( z, t ) = πR 2 w ( z, t ) = 2 π wrdr
0

where w is the mean velocity over the cross section.

Integrating the continuity equation over the cross section, we have

R R
∂w 1 ∂
∫0
∂z
2 πrdr + ∫ r ∂r (ur)2πrdr = 0
0
7328_C006.fm Page 200 Monday, October 16, 2006 10:51 AM

200 Biofluid Mechanics: The Human Circulation

sθ sθ

dθ

FIGURE 6.4
An element of the tube wall with the forces acting on the same.

Combining the above two equations, we get

∂  2
πR w( z, t)  + 2 πur 0 = 0
R
∂z 

Because of axisymmetry, we have u r= 0 = 0 . Also u|r = R = uR, where uR is the

radial velocity magnitude at the wall. Thus, we obtain

R ∂w
uR ( z, t) = − (6.7)
2 ∂z

Thus, with the simplifying assumptions and the order of magnitude study,
the two reduced governing equations for the fluid motion are Equation 6.5
and Equation 5.6.
Let us now consider the equations of motion for the tube wall. We will
assume that the tube is thin walled and, hence, neglect the bending stresses
and the stress variation in the radial direction. Let η( z, t) be the displacement
of the tube in the radial direction. Figure 6.4 represents a segment of the
tube wall with the forces acting on the element shown. The length of the
segment along the axial direction is assumed to be unity.
The circumferential strain can be computed as

2 π ( R + η) − 2 πR η
εθ = =
2 πR R
7328_C006.fm Page 201 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 201

The hoop stress (stress in the wall along the circumferential direction) will
be given by

η
σ θ = Eεθ = E
R

Applying Newton’s law in the radial direction, we have

d2 η
ρt hRdθ = pRdθ − σ θ hdθ
dt 2

In this equation, the term on the left-hand side represents the inertial force
and the terms on the right-hand side represent the forces due to the internal
pressure and the hoop stress in the membrane, respectively and ρt is the
density of the tube wall material. As an initial approximation, if we neglect
the inertial forces using the same reasoning as that for neglecting ∂p/∂r, the
equation of motion for the tube reduces to

hEη
p ( z, t ) = (6.8)
R2

Finally, the boundary condition at the interface equates the fluid velocity
in the radial direction at the wall with the rate of the tube displacement in
the radial direction and is given by the equation

dη
= η = uR
dt

Using Equation 6.6, we have

R ∂w
η = −
2 ∂z

Differentiating Equation 6.7 and substituting

R 2 ∂p R ∂w
η = =−
hE ∂t 2 ∂z

and rewriting the same, we obtain

∂ w 2 R ∂p
− =
∂z hE ∂t
7328_C006.fm Page 202 Monday, October 16, 2006 10:51 AM

202 Biofluid Mechanics: The Human Circulation

Differentiating with respect to time, we get

∂ ∂ w 2R ∂2 p
− =
∂t ∂z hE ∂t 2

Since the pressure is a constant over a cross section and, hence, indepen-
dent of r-coordinate, integrating Equation 6.5 over the cross section, we get

∂w 1 ∂p
=−
∂t ρ ∂z

Differentiating with respect to z and substituting, we get

2R ∂2 p ∂  ∂ w  ∂  ∂ w  ∂  1 ∂p 
= =−  =
hE ∂t 2 ∂t  ∂z  ∂z  ∂t  ∂z  ρ ∂z 

and, hence,

∂2 p 1 ∂2 p 1 ∂2 p
= =
∂z 2 ( hE
2 Rρ ) ∂t 2 c02 ∂t 2

The above equation is the wave equation from which we obtain the rela-
tionship for the velocity of wave propagation, Co, as

hE
c02 = (6.9)
2 Rρ

The above equation is known as the Moens–Korteweg relationship and

expresses the speed of a pressure pulse wave propagation through a thin-
walled elastic tube containing an incompressible, inviscid fluid. The wave
speed is directly proportional to the modulus of elasticity of the tube and to
the ratio of tube thickness and the radius. However, this expression yields
a constant value for the wave speed, whereas experimental measurements
have shown that the pulse wave speed in the arteries is a function of the
frequency.
Therefore, going back to the derivation of the tube equations, we will retain
the inertial force term that we neglected earlier. Then the equation of motion
for the tube segment will be:

Eh d2 η
p= 2
η + ρt h 2
R dt
7328_C006.fm Page 203 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 203

where the inertial term has been retained in the last term. To obtain the
solution of this second-order differential equation, we will assume that the
pulse pressure is sinusoidal of the form

p = A sin( kz − ωt)

where k = λ1 , λ being the wave length, and ω the circular frequency of oscil-
lation. We will assume that the tube displacement will also follow the same
form and, hence,

η = B sin( kz − ωt)

Substituting the relationship for the pressure, radial displacement of the

tube, and its derivative into the differential equation, we obtain

 Eh 
A sin( kz − ωt) =  2 − ρt hω 2  B sin( kz − ωt)
R 

or

 Eh 
p =  2 − ρt hω 2  η
 R 

Differentiating with respect to time and using the continuity equation

relationship, we get

∂p  Eh  ∂η a  Eh  ∂w
=  2 − ρt hω 2  = −  2 − ρt hω 2 
∂t  R  ∂t 2 R  ∂z

This equation can be rewritten as

∂w 2 R ∂p
− =
∂z hE∗ ∂t

where

 Eh 
E∗ = E  2 − ρt hω 2 
R 
7328_C006.fm Page 204 Monday, October 16, 2006 10:51 AM

204 Biofluid Mechanics: The Human Circulation

Differentiating once again with respect to time, we get

2R ∂2 p ∂  ∂w  ∂  ∂w 
=−  =−  
hE ∂t
∗ 2 ∂t  ∂z  ∂z  ∂t 

Thus,

∂2 p 1 ∂2 p
= hE∗
∂z 2 2 Rρ ∂t 2

The above relationship is the wave equation and, thus, the wave speed is
given by

hE  ω 2ρt R 2 
c02 ( ω ) = 1−
2 Rρ  E 
(6.10)

Even though Equation 6.9 includes the frequency in the relationship, in

magnitude, the second term in parenthesis is much smaller than unity and,
hence, the computed wave speed will be essentially a constant for a wide
range of frequencies. A plot of Equation 6.8 and Equation 6.9, along with
experimentally measured wave speed as a function of frequency, are
included in Figure 6.5. As can be observed, the theoretical computation

22

20

18
Experimental data
16
Phase velocity (m/sec)

14
Eqn. 6.13
12

10

2 Eqn. 6.8 and 6.9

0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10
Frequency (␻/2p)

FIGURE 6.5
Pulse wave velocity plotted as a function of frequency from the theoretical models compared
with experimental data. (Experimental curve redrawn from Noordergraaf, A. (1978) Circulatory
Systems Dynamics, Elsevier, Philadelphia. With permission.)
7328_C006.fm Page 205 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 205

essentially predicts a constant wave speed in variation with the experimen-

tally observed values. We will now consider the effect of the inclusion of the
viscosity on our analysis.

6.3.1.2 Womersley Model for Blood Flow: Including Viscosity Effects

Womersley (1955a, 1955b; 1957a, 1957b) published an extensive treatise on
theoretical analysis of blood flow through arteries. Earlier, Morgan and Kiely
(1954) among others also reported on models of blood flow through the
arteries. These models considered unsteady flow of incompressible, New-
tonian fluid through elastic tubes and obtained expressions for the velocity
profiles in the cross section of the tube. Once again, we will consider an
axisymmetric elastic tube of circular cross section, as shown in Figure 6.3.
Due to axisymmetric flow, the tangential velocity component Vθ will be equal
to zero in the governing equations (Chapter 1, Table 1.1b: Cylindrical coor-
dinate system), and the variation with respect to θ can also be neglected. As
before, performing an order of magnitude study on the nonlinear terms in
the radial and axial momentum equations we can show that the nonlinear
terms can be neglected in comparison to the leading temporal acceleration
terms. Thus, the simplified continuity and momentum equations, including
viscous effects, will be:

∂Vr Vr ∂Vz
+ + =0
∂r r ∂z

∂Vr 1 ∂p µ  ∂2Vr 1 ∂Vr ∂2Vr Vr 

=− + + + 2 − 2
∂t ρ ∂r ρ  ∂r 2 r ∂r ∂z r 

∂Vz 1 ∂p µ  ∂2Vz 1 ∂Vz ∂2Vz 

=− + + + 2 
∂t ρ ∂z ρ  ∂r 2 r ∂r ∂z 

In the above equations, the gravitational forces have been neglected by

assuming that the artery is horizontal. For a solution for flow through elastic
tubes, we need to develop the governing equations for the elastic tube in
addition to the above equations and then simultaneously solve the equations
with the appropriate interface boundary conditions. Before we consider the
effect of wall elasticity on the flow, we will initially assume that the wall is
rigid and analyze for the velocity distribution in the fluid. With the assump-
tion of rigidity, there will be no radial motion of the wall and with symmetry;
the radial velocity component will be zero. Then the continuity equation will
reduce to

∂Vz ∂2Vz
= =0
∂z ∂z 2
7328_C006.fm Page 206 Monday, October 16, 2006 10:51 AM

206 Biofluid Mechanics: The Human Circulation

The momentum equation in r direction will reduce to

∂p
=0
∂r

From the above relationship, we observe that the pressure is a function of

axial distance and time alone (being constant over the cross section) and the
axial velocity component is a function of radius and time alone. Or, in other
words, p( z, t) and w(r , t) .
The momentum equation in the z direction will reduce to

∂Vz 1 ∂p µ  ∂2Vz 1 ∂Vz 

=− + +
∂t ρ ∂z ρ  ∂r 2 r ∂r 

Since p is a function of z and t, dp dz will be a function of time only. For a

sinusoidal variation in the pressure gradient, we will assume a relationship

dp
= Ae iωt
dz

where A is a constant representing the amplitude of the pressure gradient

pulse.
Then, it follows that the velocity will also be of similar form and, hence,
we have

Vz = w(r )e iωt

Substituting these equations into the governing equation in the z direction,

we have

A iωt µ  d 2 w 1 dw  iωt
iωwe iωt = − e +  2 +
r dr 
e
ρ ρ  dr

which can be rewritten as

d 2 w 1 dw iωρ A
+ − w=
dr 2 r dr µ µ

We will normalize the radius in the equation by the relationship

r′ = r R
7328_C006.fm Page 207 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 207

Then the differential equation will reduce to the form

d 2 w 1 dw A
+ − iα 2 w = R 2 (6.11)
dr ′ 2 r ′ dr ′ µ

where

ωρ
α 2 = R2
µ

or

d ωρ
α= (1.81)
2 µ

where d is the diameter of the tube.

The term α is a nondimensional parameter known as the unsteady
Reynolds number and also is referred to in the literature as the unsteadiness
parameter or the Womersley parameter. α can be rewritten as shown

ωρ ρRv τ
α 2 = R2 ∝ µv
µ R

The numerator represents the inertial forces and the denominator, the
viscous forces, similar to the Reynolds number defined earlier. Table 6.1 gives

TABLE 6.1
Radius of the Vessel, Heart Rate, and the Womersley Parameter for the Various
Species
Species Weight (kg) Vessel Radius (cm) Heart Rate αa (min-1)
Mouse 0.017 Aorta 0.035 500 1.4
Rat 0.6 Aorta 0.13 350 4.3
Cat 3.0 Aorta 0.21 140 4.4
Rabbit 4.0 Aorta 0.23 280 6.8
Dog 20.0 Aorta 0.78 90 13.1
Man 75.0 Aorta 1.5 70 22.2
Ox 500.0 Aorta 2.0 52 25.6
Elephant 2000.0 Aorta 4.5 38 49.2

Rat 0.6 Femoral 0.04 350 1.5

Rabbit 4.0 Femoral 0.08 280 2.4
Dog 20.0 Femoral 0.23 90 3.9
Man 75.0 Femoral 0.27 70 4.0
a At fundamental frequency.
Source: Milnor, W.R. (1989) Hemodynamics, 2nd ed., Williams and Wilkins, Baltimore. With
permission.
7328_C006.fm Page 208 Monday, October 16, 2006 10:51 AM

208 Biofluid Mechanics: The Human Circulation

the typical values for the radius of the aorta as well as the femoral artery
for various species and the corresponding Womersley parameter values. As
can be observed, even though the radius and heart rate vary over a wide
range among the species, the variation in the Womersley parameter for the
various species is within one order of magnitude.
The differential equation given in Equation 6.10 is of the form

d2w dw
z2 2
+z + ( z 2 − γ 2 )w = 0
dz dz

and is known as the Bessel’s equation of the first kind and γth order. The
solution for the equation is the Bessel function of the first kind and γth order
is J γ . With the appropriate boundary conditions, the solution for the Equation
6.10 is:

 J 0  αr′i 2  
 3 
AR 2 
w= 1 −  3  
iµα 2  J 0  αi 2  
 

and, hence,

 J 0  αr′i 2  
 3 
AR 2 
Vz = 1 −  3   e iωt (6.12)
iµα 2  J 0  αi 2  
 

In these equations, i is the imaginary number –1.

To obtain an expression for the instantaneous flow rate in any given cross
section, the above relationship can be integrated over the cross section to
obtain

1  2 J 1  i 2α 
 3 
πR 2 Ae iωt 
0
2
∫
Q = 2 πR wr ′dr ′ = 1−
iωρ  i 3 2αJ 0  i 3 2α 
  
(6.13)

The results given above for the axial velocity and the instantaneous flow
rate are complex functions and can be separated into the real and imaginary
parts. They can also be expressed in terms of the modulus and phase, and
the phase angle will represent the phase lag or lead with respect to the applied
harmonic pressure gradient. Womersley (1955a) expressed the complex
expressions in terms of modulus and phase angle by using the relationship

  J  i 3 2αr′  
 0 
M0′ e iε0 = 1 −   3   
  J 0  i 2α  
7328_C006.fm Page 209 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 209

and

  2 J  i 3 2α  
iε10 = 1 −   
1
′
M10 e   3 2  3 2  
  i J 0  i α  

In these expressions, M’ represents the modulus and ε represents the phase

angle. Thus, the expressions for the velocity and the instantaneous flow rate
were given by

 Ae iωt 
Vz =  ( M0′ e iε0 ) 
 i ωρ 

and

 πR 2 Ae iωt 
Q= ′ e iε10 ) 
( M10
 iωρ 

Corresponding to a real part of the applied complex pressure gradient, the

solution will be the real part of the above equation and can be expressed
using the trigonometric functions as

A R 2 M0′
Vz = − sin(ωt − φ + ε 0 )
µ α2

and

A πR 4 M10
′
Q= sin(ωt − φ + ε′10 )
µ α 2

The relationship for the instantaneous flow rate can be rewritten as

A πR 4 M10
′
Q= cos(ωt − φ + [90° − ε′10 ])
µ α2

Thus, for a given applied sinusoidal pressure gradient, the results shown
above describe the velocity profile over a cross section as well as the instan-
′
′ , M10
taneous flow rate. Figure 6.6 shows a plot of M10 α2
, and ε′10 as a function
′
of a. It can be observed from the figure that as α approaches 0, M10 α2
approaches 1/8, which is the constant for Poiseuille flow relationship in
steady flow (zero frequency) and ε′10 approaches 90. It can be observed that
7328_C006.fm Page 210 Monday, October 16, 2006 10:51 AM

210 Biofluid Mechanics: The Human Circulation

1.0
M′10

0.10

M′10/a2
M′10

0.5
0.05

M′10/a2

0 0
a
90°
⑀′10

60°

30°

FIGURE 6.6
' , M /α2, and ε'
M10 10 10 as a function of α. (Redrawn from Milnor, W.R. (1989) Hemodynamics,
2nd ed., Williams and Wilkins, Baltimore. With permission.)

the phase lag is zero when ε′10 is 90. Thus, with α equal to 0, we have steady
flow with maximum flow rate and zero phase lag between the applied
pressure gradient and flow. With increasing frequency of oscillation, the flow
rate decreases and the phase lag between the pressure gradient and the flow
increases. Figure 6.7 shows a measured pressure gradient from the main
pulmonary artery of a human subject. This pressure gradient pulse, which
repeats itself for every cycle, can be resolved into various harmonics using
the Fourier series expansion. The middle panel shows the first harmonic of
the pressure gradient pulse and the bottom panel shows the instantaneous
flow rate computed. The modulus and amplitude of the first harmonic of
the pressure gradient and the computed flow rate are also shown in
Figure 6.7. It can also be observed that the flow lags behind the pressure
gradient. Since the applied pressure gradient and the computed flow rate
are sinusoidal, averaging the flow rate over a cycle will yield a zero flow rate.
Figure 6.8 shows the synthesis of a measured pressure gradient (panel D)
into four harmonics. Panel A illustrates the first two harmonics and Panels
B, C, and D show the consecutive addition of the other two harmonics to
obtain the resulting pressure gradient. The computation of the corresponding
instantaneous flow rate for each of the harmonics is shown superimposed
with the corresponding pressure gradient in Figure 6.9. Note the phase
7328_C006.fm Page 211 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 211

0 p 2p

Phase shift (rad)

0.5
mm Hg/cm

(a)
0

–0.5 (b) 0.5

mm Hg/cm
A
0

f (1.57 – ⑀′) –0.5

50 (c)
Q
ml/sec

–50

0 0.3 1.0
Time(s)

FIGURE 6.7
Pressure gradient, p/z, measured from the main pulmonary artery of a human; (a) the first
harmonic of the pressure gradient, (b) and the first harmonic of the computed instantaneous
flow rate, and (c) curves. (Redrawn from Milnor, W.R. (1989) Hemodynamics, 2nd ed., Williams
and Wilkins, Baltimore. With permission.)

difference between the pressure gradient and the flow rate in each of the
harmonics.
As pointed out earlier, when the flow rate curves are time averaged over
a cardiac cycle, zero net flow will be the result. However, we know that the
arteries have a net positive flow (cardiac output) and this is incorporated by
the addition of a steady or d.c. component of flow as shown in Figure 6.10.
In this figure, the addition of the four harmonics of the flow is shown and
in the last panel, the x axis is shifted down to include a steady flow compo-
nent. The time average of this flow rate curve in a cycle will give the stroke
volume and multiplied by the heart rate will give the cardiac output.
Figure 6.11a shows a mean velocity curve computed from the theory. The
measured pressure gradient curve used in the computations is included in
Figure 6.11b. A comparison between the computed mean velocity curve and
the measured curve (McDonald, 1974) shows good agreement, in spite of
the assumptions used in deriving the expression for the theoretical flow rate.
Equation 6.11 gives the expression for the velocity profile in the lumen
using the above theory and, once again, the velocity profiles can be computed
for the various harmonics of the pressure gradient and plotted for various
times during the cardiac cycle. Figure 6.12 shows velocity profiles plotted
as a function of the nondimensional radius for the first four harmonics, the
7328_C006.fm Page 212 Monday, October 16, 2006 10:51 AM

212 Biofluid Mechanics: The Human Circulation

6 6
dp/dz (mm Hg/cm)

dp/dz (mm Hg/cm)

4 Harmonic 2 4
1+2
2 2

0 0

–2 –2
Harmonic 1 Harmonic 3
–4 –4
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(a) (b)

6 6
dp/dz (mm Hg/cm)

dp/dz (mm Hg/cm)

4 4
1+2+3 1+2+3+4
2 2

0 0

–2 –2
Harmonic 4
–4 –4
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(c) (d)

FIGURE 6.8
The first four harmonics of the pressure gradient measured in the dog’s femoral artery. The
pressure gradient data were obtained from McDonald’s Blood Flow in Arteries. (Nichols, W.W.
and O’Rourke, M.F. (1990) McDonald’s Blood Flow in Arteries, 3rd ed., Lea and Febiger, Phila-
delphia.)

angle in the y axis representing the time in the cardiac cycle (ωt). Panel A
represents the fundamental frequency with α = 3.34 and B, C, and D represent
the next three harmonics with values of 4.72, 5.78, and 6.67, respectively.
It can be noted that the profiles are plotted only for one-half of the cardiac
cycle. Since the velocity varies as a sinusoidal function, the profiles for the
other half will be a mirror image of the first half. Several interesting features
can be noted from Figure 6.12. It can be observed that the profiles do not
become completely parabolic at any time during the cardiac cycle. Since the
velocity of the fluid at the wall is zero, the laminae near the wall are slowed
down first due to the viscous forces and the fluid motion near the wall is
slower. Thus, when the pressure gradient reverses, the fluid near the wall
tends to reverse direction easily. Also, some time must elapse before the
viscous diffusion occurs towards the core region. Hence, as the frequency
increases in the subsequent harmonics, the viscous effects are not felt in the
core region and the velocity profile is relatively flat during most of the
7328_C006.fm Page 213 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 213

Pressure gradient (mm Hg/cm)

Pressure gradient (mm Hg/cm)

6 3 6 3
Harmonic 1 Harmonic 2
Flow rate (cm3/sec)

Flow rate (cm3/sec)

4 2 4 2
Pressure
2 Flow 1 2 1
0 0 0 0
–2 –1 –2 –1
–4 –2 –4 –2
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(a) (b)

Pressure gradient (mm Hg/cm)

Pressure gradient (mm Hg/cm)

6 3 6 3
Harmonic 3 Harmonic 4
Flow rate (cm3/sec)

Flow rate (cm3/sec)

4 2 4 2
2 1 2 1
0 0 0 0
–2 –1 –2 –1
–4 –2 –4 –2
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(c) (d)

FIGURE 6.9
The first four harmonics of the pressure gradient are plotted along with the corresponding flow
rate computed using Equation 6.12. The radius of the femoral artery was assumed to be 0.15
cm and the first harmonic was assumed to be 3.34.

cardiac cycle. Figure 6.13 shows the computed velocity profile in the femoral
artery of a dog from the measured pressure gradient and with an addition
of a steady flow component (a parabolic velocity profile with an axial velocity
of 22.65 cm/s). Once again, it can be observed that the velocity profile
reversal occurs first near the wall.

6.3.1.3 Wave Propagation in an Elastic Tube with Viscous Flow

In the relationships derived in the previous section, we assumed that the
tube is rigid and that assumption implies an infinite wave speed. However,
we know that the pulse wave propagates through the arteries with a finite
speed and so we will now introduce the effect of tube elasticity. Hence, we
need to derive the governing equations of motion for the elastic tube. We
will assume that the tube is homogeneous, isotropic, with Hookean elastic
material and is also thin walled so that we can neglect the bending stresses
in the tube wall. Figure 6.14 shows a small element of the tube with all the
stresses represented. σ represents the forces exerted on the tube due to the
7328_C006.fm Page 214 Monday, October 16, 2006 10:51 AM

214 Biofluid Mechanics: The Human Circulation

6 6
Flow rate (cm3/sec)

Flow rate (cm3/sec)

4 4 1+2
Harmonic 1
2 2 Harmonic 3

0 0

–2 –2

–4 Harmonic 2 –4
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(a) (b)

6 6

Flow rate (cm3/sec)

Flow rate (cm3/sec)

4 1+2+3 4 1+2+3+4

2 Harmonic 4 2

0 0

–2 –2

–4 –4
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec) Time (sec)
(c) (d)

FIGURE 6.10
Shown is the addition of the first four harmonics of the computed flow rate to yield the flow
rate through the femoral artery. The plot on the bottom right illustrates an offset in which a
steady flow component with a mean velocity of 22.65 cm/s has been added. The mean velocity
was computed based on the Poiseuille flow relationship with the steady flow component of
the measured pressure drop.

fluid and S represents the stresses within the tube membrane. Displacements
in the r, θ, z directions are respectively η, ζ, and ξ. ρt is the tube material
density and h is the thickness of the tube.
The equation of motion in the r direction will be:

∂2 η
ρt hRdθdz = σ rr Rdθdz − Sθ hdzdθ
∂t 2

and will simplify as

∂2 η h
ρt h = σ rr − Sθ
∂t 2 R
7328_C006.fm Page 215 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 215

100

80

Mean velocity (cm/sec)

60

40

20

–20

–40

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Time (sec)
(a)

4
dp/dz (mm Hg/cm)

–2

–4
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (sec)
(b)

FIGURE 6.11
Plots of the computed flow rate and measured pressure gradient (steady flow component and
the first four harmonics) through the femoral artery of a dog. The agreement between the
computed and the measured (Nichols, W.W. and O’Rourke, M.F. (1990) McDonald’s Blood Flow
in Arteries, 3rd ed., Lea and Febiger, Philadelphia) flow rate profiles is very good.

In the θ direction, the displacement is zero due to the axisymmetry. In the

z direction, we get

∂2 ξ ∂S
ρt hRdθdz = − σ rz Rdθdz + z hdzRdθ
∂t 2 ∂z
7328_C006.fm Page 216 Monday, October 16, 2006 10:51 AM

216 Biofluid Mechanics: The Human Circulation

(a) (b) (c) (d)

180°
165°
150°
135°
120°
Axial velocity (cm/sec)

105°
90°
75°
60°
45°
30°
100 cm/sec
15°
0°

–1.0 0 1.0 –1.0 0 1.0 –1.0 0 1.0 –1.0 0 1.0

FIGURE 6.12
Velocity profiles plotted as a function of nondimensional radius in the femoral artery of a dog
computed using the relationship given in Equation 6.11. A, B, C, and D refer to the first four
harmonics with a = 3.34, 4.72, 5.78, and 6.67, respectively.

and will simplify as

∂2 ξ ∂S
ρt h = − σ rz + z
∂t 2 ∂z

The strains in the θ and z directions will be given by

2 π(R + η) − 2 πR η
εθ = =
2 πR R

and

∂ξ
εz =
∂z

From the generalized Hooke’s law, we can derive the expressions for the
tube stresses as follows:

Eε z = Sz − νSθ
7328_C006.fm Page 217 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 217

360°
165° 345°
150° 330°
135° 315°
120° 300°
Axial velocity (cm/sec)

105° 285°
90° 270°
75° 255°
60° 240°
45° 225°
30° 210°
15° 195° 100 cm/sec
0° 180°

–1.0 –0.5 0 0.5 1.0 –1.0 –0.5 0 0.5 1.0

FIGURE 6.13
Computed velocity profiles plotted as a function of the nondimensional radius during a cardiac
cycle in the femoral artery of a dog. A parabolic velocity profile with a mean axial velocity of
22.65 cm/s has been superposed on the first four harmonics in the plots.

σrr, η

Sθ
δSz
Sz + d
δz z

Szθ

σrz, ξ σrθ, ζ
Sθz

Sz

Sθ
h

FIGURE 6.14
Stresses on an element of the tube.
7328_C006.fm Page 218 Monday, October 16, 2006 10:51 AM

218 Biofluid Mechanics: The Human Circulation

and

Eεθ = Sθ − νSz

Solving for the tube stresses in terms of strains, we get

(εθ + νε z )
Sθ = E
(1 − ν2 )

and

(ε z + νεθ )
Sz = E
(1 − ν2 )

Thus,

 η ∂ξ 
 +ν 
 R ∂z 
Sθ = E
(1 − ν )
2

and

 ∂ξ η 
 +ν 
 ∂z R 
Sz = E
(1 − ν )
2

or

 ∂2 ξ ν ∂η 

+ 
∂Sz  ∂z2 R ∂z 
=E
∂z (1 − ν2 )

The fluid stresses in the cylindrical coordinates are:

∂Vr  1 ∂Vr ∂Vθ Vθ 

σ rr = − p + 2µ ; σ rθ = µ  + −
∂r  r ∂θ ∂r 2 r 

 1 ∂Vθ Vr   ∂V 1 ∂Vz 
σ θθ = − p + 2µ  +  σ θz = µ  θ + 
 r ∂θ r   ∂z r ∂θ 

∂Vz  ∂V ∂V 
σ zz = − p + 2µ σ rz = µ  z + r 
∂z  ∂r ∂z 
7328_C006.fm Page 219 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 219

Substituting the stress relationships into the equations of motion, we get

∂2 η ∂V hE  η ν ∂ξ 
ρt h = − p + 2µ r − +
∂t 2 ∂r r=R
(1 − ν2 )  R 2 R ∂z 

and

∂2 ξ  ∂Vz ∂Vr  hE  ∂2 ξ ν ∂η 
ρt h = − µ + + +
∂t 2 
 ∂r ∂z  r = R (1 − ν2 )  ∂z 2 R ∂z 


The governing equations of motion for the fluid, along with the above two
equations must be solved simultaneously to obtain the solutions for the
dependent variables (i.e., the velocity components, pressure, and the tube
displacements). The boundary conditions at the interface between the fluid
and the tube are specified as

∂η
= u r=R
∂t

and

∂ξ
= w r=R
∂t

Morgan and Kiely (1954), as well as Womersley (1957a, 1957b), have

presented the solutions for the above-formulated problem. As before, by
performing an order of magnitude study, the equations can be linearized
and further simplifications can be made. To obtain the solutions for a
harmonic input pressure gradient, the pressure can be assumed to be of
the form

p = Pe i ( kz−ωt )

where k = k1 + ik2 in which k1 = λ1 is the wave number and k2 is the damping

constant. Morgan and Kiely (1954) obtained the solutions for the velocity
profiles in terms of Bessel functions. They also presented results for special-
ized cases of small viscosity and large viscosity.
For small viscosity, the wave speed and damping constant were given by
the relationship

 Eh 
1  1 
ω 1 − 1 − ν + ν  1  µ  
2 2 2
C = = ± (6.14)
k1  2 Rρ     
4  R  2ρω  
 
7328_C006.fm Page 220 Monday, October 16, 2006 10:51 AM

220 Biofluid Mechanics: The Human Circulation

and

1 1
 2 Rρ  2  µ  2
1 ν2 
k2 = ±ω  1− ν + 
 Eh   2ρω  
R 4

For large viscosity, the relationships are:

1 1
 Eh  2  ωρ  2
1
C = ± R 
 2 Rρ   µ  (5 − 4 ν)

and

1 1
 2 Rρ  2 1 µ  2
k2 = ±  (5 − 4 ν)
 Eh  R  ωρ 

The plot of the wave velocity for the low viscosity case as a function of
frequency is also shown in Figure 6.5. The theoretical results predict an
increase in phase velocity with increase in frequency in contrast to the exper-
imental results. Even though the theoretical models presented above appear
to yield reasonable results for blood velocity profiles, further improvements
are necessary in order to simulate the propagation of pulse waves in the
arterial wall as briefly described below. In deriving the above relationships
for flow of incompressible viscous fluid through an elastic tube, we made
several simplifying assumptions and following is a discussion on the validity
of those assumptions.

1. Laminar flow: This condition is generally satisfied in the circulation

for most of the time. During peak systole near the aortic root, some
disturbance to flow has been measured and also in vessels like the
main pulmonary artery. However, sustained turbulent flow may not
exist in the circulation under normal conditions. Turbulent or highly
disturbed flow may be present distal to the diseased valves, in the
presence of stenosis or with implanted prosthetic devices.
2. Newtonian fluid: In our discussion of blood rheology, we learned that
the blood exhibits a non-Newtonian behavior at low shear rates.
However, in relatively large blood vessels, the rates of shear are
expected to be high for most of the pulsatile flow cycle and the
assumption of a Newtonian fluid appears to be reasonable.
3. Uniform cylindrical tube: This assumption is an approximation to for-
mulate the governing equations in the cylindrical coordinate system.
The blood vessels are not perfectly cylindrical and also the cross
section tapers in the axial direction and, thus, the cross-sectional area
7328_C006.fm Page 221 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 221

reduces in the distal portions even though the tapering is very gradual.
For a single conduit, the reduction in cross section can be approxi-
mated by an exponential form (Li, 1987). For the aorta, the change
in cross section can be expressed as

kz
−
A( z) = A0e r

where the taper factor k can be expressed as

 r  A 
k =   ln  0 
 z  A 

In the equation above, Ao is the cross-sectional area at the proximal

site, r is the radius, and z is the distance along the axis. For the
peripheral vessels, Li (1998) expressed the change in cross sections
using a relationship

A = A0e − kz

where the magnitudes for k were computed for each of the vessels
from in vivo measurements.
4. Entrance effects: The derivations given above assume a fully devel-
oped flow, which is a condition that does not exist in the circulation.
In order to distribute blood throughout the body, blood vessels very
often branch out and their geometry is very irregular and includes
some curvature. Hence, fully developed flow cannot be expected in
the blood vessels and this assumption is also invalid.
5. Reflected waves: The elastic tube was assumed to be infinitely long
without the presence of any discontinuities, such as curvature or
branching where wave reflections can occur. The effect of wave
reflections has not been taken into account in our formulation.
6. Linearization of equations: Through an order of magnitude study, the
equations were linearized with the assumption that the contribu-
tions from the nonlinear terms are relatively small. Since the equa-
tions were linearized, it was also possible for us to superimpose the
solutions for the higher harmonics to obtain the solution for physi-
ological pulsatile flow. Hence, we have neglected the contributions
of the nonlinear terms and have also ignored the nonlinear interac-
tions between the harmonics. Womersley (1957a) has shown that
neglecting the nonlinear terms in the governing equations for the
fluid will result in an error of about 5% in the computed velocity
magnitudes.
7328_C006.fm Page 222 Monday, October 16, 2006 10:51 AM

222 Biofluid Mechanics: The Human Circulation

7. Thin-walled tube: As was discussed in a previous chapter, a thick-

walled tube model is more appropriate for the blood vessel wall and
the thin-walled formulation is again a simplifying approximation.
8. Homogeneous, isotropic, and Hookean material: The blood vessel has
several components and is not homogeneous. It also exhibits at least
a transversely orthotropic behavior and a nonlinear load deforma-
tion. Hence, this assumption is also an approximation.

In order to remove the restrictions on the assumptions discussed above in

the blood flow models, several theoretical investigations have been reported
in the literature. Typical studies have included wave propagation through
an initially stressed elastic tube containing viscous fluid and wave propaga-
tion in a tethered, initially stressed orthotropic elastic tube. Experimental
studies have included the transmission characteristics of axial waves in blood
vessels by superimposing sinusoidal waves in segments of arteries and
measuring the wave speed. The dissipation and dispersion properties of
small waves in arteries and veins with viscoelastic wall properties have also
been analyzed. The wave propagation in a thick-walled viscoelastic tube
containing Newtonian fluid as well as the effect of wall compressibility on
the transmission characteristics have been analyzed. Experimental measure-
ments of the pressure wave propagation in the canine femoral artery have
suggested that linearized wave propagation models are reasonably accurate
in predicting the flow and pressure wave propagation. With each complexity
introduced in the model for theoretical analysis, the solution procedure
becomes complicated because the nonlinear governing equations for the tube
and the fluid have to be simultaneously considered with the appropriate
interface boundary conditions. Moreover, verification of the results pre-
sented from the theoretical analysis is extremely difficult because detailed
measurements of wave propagation in vivo are exceedingly complex.

6.4 Hemodynamic Theories of Atherogenesis

As discussed earlier (Chapter 3, section 3.10: Atherosclerosis), the develop-
ment of atherosclerosis in large arteries is the leading cause of cardiovascular
mortality in the Western world. Atherosclerosis is a dangerous disease
because it not only obstructs the artery lumen, but it also may result in a
plaque rupture where the fibrous cap covering the plaque breaks open and
exposes a highly thrombogenic surface. The common sites of predilection of
atherosclerosis in the human circulation were shown earlier in Figure 3.25.
These sites include the coronary arteries; the branching of the subclavian
and common carotids in the aortic arch; the bifurcation of the common
carotid to internal and external carotids, especially in the carotid sinus region
7328_C006.fm Page 223 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 223

distal to the bifurcation; the renal arterial branching in the descending aorta;
and in the ileo-femoral bifurcations of the descending aorta. The common
feature in the location for the development of the lesion is the presence of
curvature, branching, and bifurcation present in these sites. The fluid dynam-
ics at these sites can be anticipated to be vastly different from other segments
of the arteries that are relatively straight and devoid of any branching
segments. Therefore, several investigators have attempted to link the fluid
dynamically induced stresses with the formation of atherosclerotic lesions
in the human circulation. The hemodynamic theories for the genesis of
atheromatous plaques, and the detailed description of the fluid dynamics
at corresponding sites in the human circulation will be considered in the
following sections.
It has been observed that, coincident with the geometric features, there is
a deviation of blood flow from simple antegrade, axial path lines to more
complex patterns (Figure 6.15). What all of these changes have in common,
fluid dynamically, is the presence of low, and potentially reversed, flow
velocities associated with locally altered pressures and shear stresses.

6.4.1 Low Pressure, Low and High Wall Shear Stress Theories
One of the earliest theories on the cause of atherosclerosis development was
the Low Pressure theory (Texon, 1957). This publication was one of the first
attempts to link fluid dynamics with atherogenesis. It was based on the
hypothesis that at sites of arterial curvature, such as the human aorta, as the
blood is forced to turn around the curve, pressure will increase on the outer
wall and correspondingly decrease at the inner wall of curvature. It was
proposed that the pressure at the inner wall would decrease significantly to
become negative and would literally pull the endothelium into the lumen,
initiating the disease process. It was later shown that the radial pressure
gradient is very small and that a significant decrease in pressure will not
occur; hence, the effect proposed in this theory is not probable. Nonetheless,
this publication was important in that it did contribute toward the thought
process on hemodynamic effects on the endothelial cells and initiation of
atheroma.
Some investigators have also suggested a role for elevated shear stress as
an initiating factor in atherosclerosis because it would produce actual phys-
ical damage to the endothelial cells and also impair the normal mass trans-
port process. In 1968, Fry published a significant study on the effect of wall
shear stresses on the endothelial cells, which formed the basis for the High
Shear Stress theory. It consisted of in vivo canine studies in which he intro-
duced a plug in the descending aorta with known dimensions of an orifice
in order to increase the flow rate and wall shear stresses to which the
endothelial cells would be exposed. The opening in the plug was eccentric
so that part of the luminal circumference was exposed to the endothelial
cells. By the analysis of the equations of motion, he justified neglecting the
7328_C006.fm Page 224 Monday, October 16, 2006 10:51 AM

224 Biofluid Mechanics: The Human Circulation

Q0
A

Re0 = 772 4.1 Left main

D0 = 4.13 mm coronary artery
Q0 = 334 ml/min

U0 = 416 mm/sec
Left circumflex
branch
3.5 2.4
S
Left anterior
descending branch
Q3
0.46

3.3 mm

1.3
Q1 3.2 Intermediate
0.49 branch
Q2
0.05
Left main coronary artery
3.7 Left anterior
B 3.9
descending branch

3.4 mm
Q0

S P
C 3.9 3.7
44 Q1
87 dynes/cm2
0.50
766
766
mm/sec 3.4
Q0 82
23
12
766
20
Left circumflex branch Q1
Aortic sinus 0.50

FIGURE 6.15
Steady flow path lines in models of human left coronary arteries. (Note the relative thickness
of the artery wall at each location.) (From Askura, T. and Karino, T. (1990) Flow patterns and
spatial distributions of atherosclerotic lesions in human coronary arteries, in Circ. Res. 66: 1054.
With permission.)

stresses due to local acceleration as well as turbulent stress fluctuations and

used the time averaged wall shear stress relationship

R ∂p
τ=
2 ∂z
7328_C006.fm Page 225 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 225

By measuring the time-averaged pressure at various axial distances along

the plug, he was able to calculate a range of wall shear stresses to which the
endothelial cells were subjected. The endothelial cells (EC) were exposed to
known stresses for 1 to 3 hours as Evans blue dye (which tags the diffusion
of albumin into the wall) was injected into the blood stream. The animals
were then sacrificed and the tissue subjected to histological study to look at
the morphology of the ECs. Fry concluded that ECs appeared to be normal
if the time-averaged shearing stresses were below 379 ± 85 dynes / cm2.
A more widely held hypothesis is that atherosclerosis is correlated with
low fluid wall shear stress. The fundamental mechanism cited for this
theory is that abnormally low wall shear stresses, τw, impair mass transport
between blood and the vessel wall, potentially impacting not only the
uptake of nutrients and oxygen by the vessel wall from the blood, but also
the release of waste products and carbon dioxide into the blood from the
wall. This latter effect would help explain how material components typi-
cally found in atherosclerotic lesions and plaques, such as cholesterol, may
build up in the vessel wall near regions with slow or stagnant flow. This
Shear-Dependent Mass Transport theory was proposed by Caro et al.
(1971) who, like Texon, observed differences in the hemodynamics of the
aortic arch. Their focus, however, was on wall shear stress, which is higher
on the outer wall of aortic curvature and lower on the inner wall of curva-
ture. And since atherosclerotic lesions occur along the inner wall of curva-
ture where there is low shear stress, their proposal was that cholesterol is
actually synthesized in the arterial wall and diffuses into the lumen where
it is washed away by the blood stream. In the regions of high wall shear
stresses (and, hence, velocity gradients), more cholesterol is washed away
by the blood flow. On the other hand, where the shear stress is low, excess
cholesterol is deposited on the surface of the lumen initiating atheroma
development. Thus, it is the shear-dependent mass transport that is respon-
sible for atheroma growth. Furthermore, the endothelium has been shown
to “switch” from an atheroprotective to an atherogenic state when exposed
to low wall shear stress (Figure 6.16). Under low and oscillatory shear
conditions, endothelial cells exist in a prothrombotic state and secrete more
of those agents that promote increased thrombogenicity, vasoconstriction,
and cellular proliferation.
Low and high wall shear stress theories were debated for several years.
Numerous model studies were performed to determine the velocity pro-
file and shear stress distribution at the curvature and branching sites in
the human circulation. Experimental studies included idealized models
(2-D and 3-D) and subsequently castings made from cadaveric specimens.
Studies included qualitative flow visualization and velocity measurements
using hot-film anemometry and then laser Doppler velocity measure-
ments. Theoretical models were also initially restricted to 2-D idealized
models and then 3-D models with regular geometries. Initially studies were
restricted to steady flow analysis to isolate regions of high and low wall
shear stresses and correlate the same with regions of lesion presence in vivo.
 Physiologic arterial 1 PGl2
Monocyte activation
226

1 NO Low arterial
shear stress 1 tPA
Anticoagulant shear stress
(τs > 15 dyne/cm2)
antithrombotic state (τs – ± 0 – 4 dyne/cm2) 1 PGl2
Procoagulant 1 NO
prothrombotic state 1 tPA
Slesdy 1 MCP–1
high flow Ose IVCAM
slow

τs
7328_C006.fm Page 226 Monday, October 16, 2006 10:51 AM

EC quiescence
1 Proliferation
1 Apoptosis Prooxidant state
Paracrine quiescent state 1 COX-1, 2
1 TGF-β 1 PDGF-B 1 Mn SOD
1 NO/eNOS 1 ACE 1 Cu/Zn SOD
1 PGl2/PGl2 Synthase 1 ET–1 Paracrine proliferative state
1 Adrenomedulin 1 ECE 1 ET–1 1 NO/aNOS
1 CNP 1 ACE 1 Adrenomedulin
High EC antioxidant activity
1 COX-1, 2 1 PDGF-B 1 PGl2/PGl2 Synthase
VCAM molecule Fibroblast 1 ECE 1 TGF-β
1 Mn SOD
1 Cu/zn SOD Activated monocyte Platelet aggregate 1 CNP
Smooth muscle cell Fibrin plug

FIGURE 6.16
Effect of shear stress on endothelial cells (EC) and smooth muscle cells (SMC). (From Malek, A.M., Alper, S.L., and Izumo, S., (1999) Hemodynamic shear
stress and its role in atherosclerosis, in JAMA 282: 2035. With permission.)
Biofluid Mechanics: The Human Circulation
7328_C006.fm Page 227 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 227

Studies were extended next to unsteady flow analysis for detailed analysis
of time-dependent velocity profiles and wall shear stress distribution.
With the development of more sophisticated measurement techniques, such
as laser Doppler and particle image velocimetry techniques, more detailed
data analysis became possible. Simultaneously, with computational fluid
dynamic (CFD) algorithm development and the advent of high-speed com-
puters, detailed CFD models were also developed. In addition, advent of
sophisticated imaging modalities, such as MRI, computed tomographic (CT),
and ultrasound imaging, morphologically realistic 3-D geometrical reconstruc-
tion of the segments of interest also became available. Studies on arterial sites
of interest resulted in additional theories as well. As a result of this work, the
high shear stress theory became less widely supported than the low shear
stress theory since the levels of wall stress necessary for cellular erosion to
occur were determined to be rarely seen under normal flow conditions.

6.4.2 Time Varying Wall Shear Stress, Oscillatory Shear Index,

and Wall Shear Stress Gradients
While several studies have shown modest correlations between anatomic
changes in the vessel wall and the mean wall shear stress acting on it, there
is insufficient evidence to exclude other factors as well. Thus, additional
theories have arisen based on the rate of change of wall shear (i.e., wall shear
stress gradients [WSSG]) with respect to either time or position. This allows
for consideration of the pulsatile nature of arterial flow and also the wide
range of geometric shapes, which exist in the arterial tree. Therefore, the
temporal wall shear stress gradient, ∂τ∂wt and the oscillatory shear index, OSI,
have been investigated to see if rapid changes in wall shear magnitude and/
or direction, respectively, correlate with development of atherosclerosis. This
could be due, for example, to greater stimulation of growth factors produced
by endothelial cells and the consequent thickening of the vessel wall. The
temporal wall shear stress gradient (TWSSG) is obtained by taking the time
derivative to the magnitude of τw , hence,

∂τ w
TWSSG =
∂t

with its mean value calculated as

∫ TWSSG dt
1
TWSSG =
T
0

In addition to regions experiencing pulsatile forward flow, there are many

regions where the flow is reversed during part of the cardiac cycle causing the
wall shear stress to vary from a large positive magnitude to negative values.
7328_C006.fm Page 228 Monday, October 16, 2006 10:51 AM

228 Biofluid Mechanics: The Human Circulation

Ku et al. (1985) proposed an oscillatory shear index, OSI, and attempted to

show a correlation of that with atheroma development As defined

 T
∫ τ w dt

 
OSI = 0.5  1 − T
0

 ∫ w 
0
τ dt


the OSI is greater than zero as long as there is some directional change during
the flow cycle. The maximum value of OSI is 0.5 in the case of pure oscillatory
flow without any net forward flow.
The spatial wall shear stress gradient (SWSSG) has been evaluated to
determine if local changes in forces acting on the endothelial cells could
induce gaps in the intercellular junction where mass transport would pro-
ceed in an uncontrolled manner. Since the wall shear stress tensor has two
components, one in the mean wall shear stress direction “m” and one normal
“n” to the same, the SWSSG is defined as

2 2
 ∂τ   ∂τ 
SWSSG =  m  +  n 
 ∂m   ∂n 

with its mean value being given as

∫ SWSSG dt
1
SWSSG =
T
0

In spite of all these theories, no one is able to completely explain the

initiation and growth of atherosclerotic plaque. For example, with injury to
the endothelium, plaque formation and development will also heavily
depend on the transport of LDL-cholesterol and its absorption into the sub-
endothelial region. Furthermore, some evidence has been presented in the
literature that regions of hypoxia contribute towards plaque formation.
Availability of oxygen is a limiting factor in the ability of cells to metabolize
lipids and some models have suggested that in regions of stagnation and
recirculation, the concentration of lipid is high (due to large residence time),
whereas oxygen concentration is low (due to reduced saturation of oxygen
in red blood cells in the region of stagnation and recirculation).

6.5 Wall Shear Stress and Its Effect on Endothelial Cells

While the heart provides the generation of blood flow, control of the dis-
tribution of flow to various organs is determined by changes in the resis-
tance of individual vessels. These changes are dynamic and can be
7328_C006.fm Page 229 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 229

controlled either systemically or locally. Systemic regulation can be accom-

plished by either neural or hormonal signals. Within a given vessel, how-
ever, independent responses can be made to local changes in flow
dynamics. In particular, changes in blood pressure and flow rate can be
“sensed” by the vessel through a signal transduction process mediated
through the endothelial cells. Since the endothelial cells are in direct contact
with the blood, they can readily detect changes in flow dynamics by mon-
itoring the forces acting on them, such as the normal force (i.e., pressure)
and the longitudinal/axial force (i.e., shear stress). Thus, they become the
key mechano-signal transducers of “information” between the blood and
the vessel wall.
To accomplish this function, the endothelial cells are equipped with a
variety of communication channels attached to the cell membrane, which
can sense force changes and then transmit corresponding signals to elicit
specific responses (Figure 6.17). Some of these sensors are actually embedded
in the membrane (e.g., integrins, G-protein and other receptors, specific ion
channels, etc.) while others act through structures, which are attached to the
membrane and extend through the cytoplasm (e.g., actin fibers).
When these sensors are activated, the cell can elicit a number of responses
that may involve its morphology, level of mass transport, stimulatory
effects, or proliferative state. For example, one of the earliest responses of
an endothelial cell to changes in the flow environment, especially to the
shear stress acting upon it, is morphologic and includes such things as a
change in shape, orientation, and the amount of its intracellular actin
content (Figure 6.18).
Other responses include either increased or decreased passage of charged
ions into the cell, changes in actin fiber orientation, and the release of various
agents. It is also possible for these membrane-sensed signals to trigger the
activity of the endothelial cell’s nucleus, resulting in the production of specific
proteins, which alter the vessel’s degree of thrombogenicity, constriction or
dilation, or ability to increase wall thickness. This latter effect is mediated
through growth factors, which generally act upon neighboring subendothe-
lial cells, primarily the smooth muscle cells and fibroblasts in the vessel media.
Growth factors have the ability to cause these cells to replicate, migrate, or,
occasionally, to phenotypically change (i.e., take on new functions).

6.6 Flow through Curved Arteries and Bifurcations

Unlike many manmade flow systems that consist of long, straight, and
uniformly sized tubes connected at the ends or at right angles to each other,
the human arterial vasculature is a complex network of tapered and curved
tubes connected at various angles and bifurcations with no distinct end
points (Figure 6.19). While these geometric structures are necessary for
7328_C006.fm Page 230 Monday, October 16, 2006 10:51 AM

230 Biofluid Mechanics: The Human Circulation

Shear Extracellular space

stress

Tyrosine G-protein pH channel

G-protein kinase receptors K+ Ca++
Integrin receptors receptors channel channel

PGl2 ?

PGG2
MAG + AA
PA

Go PLC
PKC Gq AC Gi GC Gi
rho DAG Grb2 Src ATP GTP
Sos cAMP cGMP K+ Ca++ Cytosol
PIP2 IP3
Ca++
sequestering ras Protein
compartment ?
Ca++ MEKK kinases
raf

MEK JNK p38MAPK/RK

MT
MF L- NOS
IF ERK 1/2 c-lun Hsp25
Hsp27 arginine L-citruline

p50 p50
B p65 c-fos Cytoskeleton NO
p85
changes
NF-kB

Cytoskeleton Transcription
factors Protein
changes +/– Coding region
DNA AUUUA
IEG Cis-elemen
Paxillin t
protein mRNA
Nucleus
rho
FAK Focal
src adhesions
Cell membrane

Extracellular matrix

FIGURE 6.17
Schematic representation of the multiple “shear–stress responsive” signaling pathways, their
crosstalk, and integration. (From Papadaki, M. and Eskin, S. G., (1997) Effects of fluid shear
stress on gene regulation of vascular cells, in Biotechnol. Prog. 13: 212. With permission.)

carrying blood from the heart to other organs, they impose major changes
on the patterns of flow through them, which, in turn, affect the dynamics
of the circulation and the forces acting on the inner lining of the vessels. As
discussed below, this latter effect is the subject of much research into the
possible causes of arterial disease. (See section 6.4: Hemodynamic Theories
of Atherosclerosis).
7328_C006.fm Page 231 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 231

(a) (b)

(c) (d)

FIGURE 6.18
Phase contrast micrographs of BAEC (bovine aortic endothelial cells) monolayers subjected to
various flow conditions for 24 hours. (From Helminger et al., (1991) Effects of pulsatile flow in
cultured vascular endothelial cell morphology, in J. Biomech. Eng. 113: 128. With permission.)

6.6.1 Curved Vessels

One of the first anatomic features encountered by blood on its exit from the
heart is the aortic arch (Figure 6.19). This curved shape enables the blood,
which is being ejected from the heart in a cephalic (i.e., toward the head)
direction, to reverse and move in a caudal (i.e., toward the feet) direction
toward the body. Dean (1927) analyzed fully developed viscous flow in
curved tubes with small curvature and, since then, numerous computational
and experimental studies on curved tube flow have been reported in the
literature. Briefly, as the flow enters a curvature site, the fluid is being forced
to change direction to follow the curve and, thus, a radial pressure gradient
develops between the outer and inner wall of curvature. The fluid is also
subjected to centrifugal force that is proportional to the square of the axial
velocity. In the case of inviscid flow, these two forces balance each other and,
7328_C006.fm Page 232 Monday, October 16, 2006 10:51 AM

232 Biofluid Mechanics: The Human Circulation

Facial

External carotid
Carotid bifurcation
Common carotid

Vertebral

Right subclavian Axillary

Brachiocephalic Left subclavian

Ascending aorta Aortic arch

Thoracic aorta
Intercostal

Thoracic aorta

Diaphragm

Coeliac Abdominal aorta

Renal
Superior mesenteric Lumbar
Inferior mesenteric

Iliac
Level of inguinal
Sacral ligament

Femoral

Saphenous
Tibial

FIGURE 6.19
Anatomy of major systemic arteries (canine). (From Caro, C. G., Pedley, T. J., Schroter, R. C., and
Seed, W. A. (1978) The Mechanics of the Circulation, Edward Arnold, London. With permission.)
7328_C006.fm Page 233 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 233

Outer
wall
Axial velocity
profile Secondary
motions

(a) (b)

FIGURE 6.20
Schematic of the flow development in a curved vessel with secondary flow.

hence, the axial velocity profile is skewed toward the inner wall of curvature
and no secondary flow develops. With viscous fluid, the two opposing forces
do not balance each other and a secondary flow develops resulting in two
vortices, called “Dean Vortices.” The axial velocity profile becomes skewed
towards the outer wall of curvature, as shown in Figure 6.20 where higher
wall shear stresses are then found.
As the curvature continues, however, those high velocity regions eventu-
ally impact with the outer wall and are deflected in a circumferential direc-
tion. The result is that blood with a high axial velocity now acquires a
circumferential component as well and is convected around the outer surface
of the aorta toward the inner curvature of the vessel. The velocity profile at
this point takes on a reverse image of the outwardly skewed profile seen
more proximally. The axial locations where these transitions occur vary
depending upon the exact geometry of the vessel and the flow rate (i.e., the
cardiac output). Superimposed on this is the fact that flow into the aortic
arch is a developing flow since the velocity profile at the aortic root past the
aortic valve is relatively flat. In the developing flow, the viscous effect is
confined to the boundary layer region initially and the fluid in the core region
acts as inviscid fluid. Hence, the velocity profile is initially skewed toward
the inner wall of the tube and slowly moves toward the outer wall in the
downstream region. Furthermore, the aortic arch is not only curved in
the lateral plane (i.e., left to right), but it also has a degree of curvature in
the anterio–posterior plane (i.e., front to back). Arterial branches from the
arch directed toward the head (e.g., the innominate, carotid, and vertebral
arteries) also affect these flow patterns as they siphon off about 25% of the
cardiac output and the cross section in the descending aorta is reduced by
about 40% from that in the ascending aorta.
The classic approach to analyzing steady flows in this type of geometry
is to apply the Bernoulli equation (see Chapter 1, section 1.6: The Bernoulli
7328_C006.fm Page 234 Monday, October 16, 2006 10:51 AM

234 Biofluid Mechanics: The Human Circulation

Equation) in a form that also allows for energy losses to occur. This
“modified” Bernoulli equation

V2
p+ρ + ρgz = H (1.52)
2

includes the term, H, called the “head loss” (in units of pressure), which
accounts for the fact that energy is dissipated through frictional and nonaxial
flow motion effects. The value of H is obtained by applying an experimental
2
coefficient, KL, to the term ρ V2 , where V is the mean velocity in the tube. For
a 180-degree bend, KL is approximately equal to 0.2.
Further complicating these geometric factors is the high degree of pulsatil-
ity of flow in the aorta ( α ≈ 20 ), which leads to rapid changes in inertia,
limited boundary layer development, and more stable (i.e., less turbulent)
flow. Analysis of purely oscillatory flow in a curved tube has shown that the
viscous effects are confined to the boundary layer region, thus producing four
Dean vortices in a cross section where the vortices in the boundary layer
region are opposite in direction to the vortices in the core region. During
systole, the flow is accelerated forward along the curved tube, and secondary
flow leads to skewing of the axial velocity profile. During early diastole, flow
moving along the inner wall of curvature reverses direction; hence, the wall
shear stress along the inner wall is reversed, whereas slight forward flow
persists along the outer wall of curvature. Thus, the axial velocity profile is
initially skewed toward the outer wall, the maximum axial velocity then
moves toward the inner wall, and subsequently moves toward the outer wall
again. As a result, the inner wall region is subjected to an oscillating shear
stress, which is forward in systole but reverse in diastole (Figure 6.21a/b).
The secondary flow consists of a single vortex in the form of a corkscrew
that persists below the diaphragm in the descending aorta and this secondary
flow affects the flow distribution in the renal arteries as well as in the iliac
bifurcation. The secondary flow also reverses direction during diastole.
Because of the importance of vessel curvature upon flow development, a
nondimensional parameter known as the Dean number, ND, has been
defined as

 RV 
ND = Re  
 RC 

and can be used for evaluating flow patterns in the aortic arch. This param-
eter is based on the more familiar Reynolds number, Re, which has been
modified by the inclusion of the relative size of the vessel, Rv, and its radius
of curvature, Rc. As can be seen, a high Dean number indicates significant
inertial effects and/or a small radius of curvature resulting in enhanced
secondary flow motions. A low Dean number, on the other hand, would
indicate relatively unimportant secondary flow effects. In the human aorta,
7328_C006.fm Page 235 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 235

50

Axial velocity (cm/s) 40

30

20

10

Inner 0 Outer
wall –1.0 0 +1.0 wall
Nondimensional radius
(a)

40
L/a 3.4
L/a 10.2
30
L/a 16.9

20
Axial velocity (cm/s)

10

0 Outer
0 +1.0 wall

–10

–20
Inner
wall –30

Nondimensional radius

(b)

FIGURE 6.21
Axial velocity profiles in (a) systole and (b) diastole at various cross sections in a curved tube.
(From Chandran, K.B., (1993) Flow dynamics in the human aorta, in J. Biomech. Eng. 115: 611.
With permission.)
7328_C006.fm Page 236 Monday, October 16, 2006 10:51 AM

236 Biofluid Mechanics: The Human Circulation

Outer wall Anterior Anterior

Posterior Outer Left
Anterior wall Right
Inner wall
Inner wall Posterior
Posterior
(a) (b) (c)

FIGURE 6.22
Forward and backward directed helices in the aorta. (From Frazin, L. et al., (1990) Functional
chiral asymmetry in descending thoracic aorta, in Circulation 82: 1992. With permission.)

the average Dean number is high (~ 1500) due to a large Re and small Rc.
The increased magnitude of the Reynolds number during systole (~ 8000)
also leads to a higher Dean number and, thus, stronger secondary motions.
The combination of strong secondary motions and rapid changes in flow
rate gives rise to reversals of the helical pattern in the aortic arch as flow
first accelerates (systole) and then decelerates (diastole) (Figure 6.22).
Other vascular sites where curvature effects can be important are in the
coronary bifurcation (whose geometry varies dynamically with the beating
heart) and the aorto-iliac bifurcation, which has a slight curvature out of the
anterio-posterior (AP) plane.

6.6.2 Bifurcations and Branches

While there are several major bifurcations in the arterial system, there is only
one that is nearly symmetrical — the Aortoiliac bifurcation (Figure 6.23).
This site is distinguished by branch vessels each having cross-sectional areas
of about 60% of that of the terminal aorta (i.e., their total branch area equals
1.2 times the aortic area) and angles of about 30 degrees from the body axis.
Branching of a vessel (Figure 6.24) can originate from a main artery (such
as the renal arterial branch from the descending aorta).
As blood enters the bifurcation site, the flow divides into the two daughter
vessels at the flow divider region (medial junction of the iliac arteries).
7328_C006.fm Page 237 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 237

f
so
d iu ure
t
Ra r va
cu
Velocity profile
Parent vessel

Angle of bifurcation
Flow divider

Boundary layer

Inlet velocity Flo

ws
profile ep
reg arat
ion ion

FIGURE 6.23
Schematic of the aorto-iliac bifurcation geometry and the flow development in the branch
vessels.

At the divider walls of bifurcations, the boundary layers are relatively thin
with maximum axial velocity outside the boundary layer. As the blood enters
the daughter vessel with a finite radius of curvature, the faster moving fluid
is found towards the flow divider due to secondary flow development. If
the corners of the outer wall are sharp, flow separation may be present along
the outer wall of the bifurcation (lateral iliac arteries).

High wall
shear stress

Flow divider
Low wall
shear stress High wall
shear stress

A′

Low wall
shear stress

FIGURE 6.24
Schematic of an arterial branching site.
7328_C006.fm Page 238 Monday, October 16, 2006 10:51 AM

238 Biofluid Mechanics: The Human Circulation

With an asymmetric bifurcation, the flow in the daughter vessels will not
be equal and flow separation may exist in the branch with reduced flow
(Q2 > Q1 ) as shown in Figure 6.25. A similar flow distribution would be
anticipated in a branch vessel, such as a T-junction, roughly corresponding
to renal arterial branches off the descending aorta. With an unequal flow
distribution, flow separation and low wall shear stresses are found along
the outer walls.

(a)

uAvg

(b)

FIGURE 6.25
Axial velocity profiles in the left anterior descending and left circumflex coronary artery during
(a) systole and (b) diastole. (From He, X. and Ku, D.J. (1996) Pulsatile flow in the human left
coronary artery bifurcation: average conditions. Biomech. Eng., 118: 74. With permission.)
7328_C006.fm Page 239 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 239

(a) (b)

FIGURE 6.26
Stream lines in a model of the carotid bifurcation. (From Berger, S.A. and Jou, L.A., (2000) Flows
in stenotic vessels, in Ann. Rev. Fluid Mech., 32: 247. With permission.)

Other important arterial bifurcations occur in the carotid (Figure 6.26) and
coronary (Figure 6.27) circulations where similar flow effects are seen.
The common carotid artery originates at the aortic arch and bifurcates into
the external and internal carotid arteries at the neck level. The external
carotid supplies blood to the facial area while the internal carotid provides
blood to the brain. A unique feature of the internal carotid artery is that it
has a bulge at the bifurcation region, referred to as the carotid sinus. Presence
of the sinus acts to further accentuate the tendency for flow separation along
the outer wall, resulting in a large separation zone, which has been widely
observed in healthy volunteers. Over time, however, these features have
been implicated in the initiation of atherosclerosis due to the resultant low
and reversing shear stresses along the outer vessel walls. (See section 6.4:
Hemodynamic Theories of Atherogenesis.)
Arterial branches are characterized by the fact that they have a large
discrepancy between the cross-sectional areas of the side branch and the
main artery (Figure 6.24). Typical ratios for the total area of the outlet
branches to the area of the inlet artery can range from 0.8 to 1.3. There is
also a wide variation of take-off angles for the branches with many of these
being large and some approaching 90 degress, such as the intercostal arteries
(branches off the thoracic aorta) and the renal arteries (branches off the
abdominal aorta). Due to the large asymmetry at a branch, flow in the branch
vessel is characterized by (1) velocity profiles that are highly skewed toward
the flow divider, (2) secondary flow motions, and (3) a tendency for flow
separation along the outer walls. Distal to the branch point, flow in the main
vessel (e.g., the abdominal aorta distal to the mesenteric artery) becomes
skewed toward the branch, contains secondary flow, and may demonstrate
separated flow along the wall opposite to the branch orifice (Figure 6.28).
As a result of these features, wall shear stresses are generally higher along
7328_C006.fm Page 240 Monday, October 16, 2006 10:51 AM

240 Biofluid Mechanics: The Human Circulation

X1 X2 X3 X4 X5

Flow

γ⋅w = 1060 1040 900 630 550 sec–1

Flow

γ⋅w = 1060 600 480 sec–1

500 370
Flow
Flow

20 cm/sec 20% 10%

FIGURE 6.27
Composite laser Doppler velocity profiles in flow model at five axial locations and at four times
during the flow pulsatile wave. Peak flow rates estimated from slope of velocity profile and
the corresponding wall shear rates are shown on second figure from top. (From Lutz et al.
(1983), Comparison of steady and pulsatile flow in a double branching arterial model, in J. Biomech.
9: 753–766. With permission.)

(a) (b)

FIGURE 6.28
Arteriograms of a left common femoral artery. (From Cho, K.J. and Williams, D.M. (1991) Current
Therapy in Vascular Surgery, 2nd ed., C.B. Ernst and J.C. Stanley, Eds., B.C. Decker, Mosby
Yearbook, Elsevier, Philadelphia. With permission.)
7328_C006.fm Page 241 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 241

the flow divider and lower along both the outer wall of the branch and the
wall opposing the branch. Again, these latter features coincide closely with
the observed sites of atherosclerosis development in patients.
It is possible to estimate the changes in energy distribution through a
branch section if we apply the “modified” Bernoulli equation to this geom-
etry. Here, the appropriate value of KL would depend on the degree of corner
roundness, the branch angle, etc., and would typically have a value of
approximately 1.0.

6.7 Flow through Arterial Stenoses and Aneurysms

The result of atherosclerosis is the development of an obstruction, or steno-
sis, within the vessel, which eventually either reduces or completely blocks
its blood flow. Initially, as the lesion grows, the arterial wall remodels and
becomes thinner in order to accommodate the lesion and maintain the lumen
cross-sectional area. However, after the limit for remodeling is reached, the
lesion starts to protrude into the lumen and occludes the cross-sectional area
available for flow of blood. With increasing occlusion and reduced cross-
sectional area becoming available for flow of blood, the blood flow dynamics
past the stenosis become complex (Figure 6.29). Until the lesion is at
advanced stages, there are no clinical symptoms and currently there are no
reliable methods to detect the lesions in the early stages when appropriate
pharmacological or mechanical interventions can be utilized to arrest or
reverse the disease process. In general, the atherosclerotic plaque often
progresses to a moderate or severe level (for example, 50 to 99% occlusion)
and then either a blood clot forms in the remaining vessel lumen, which
stops flow, or a piece of the plaque and/or thrombus breaks off and embo-
lizes (i.e., migrates) distally to a small arterial branch causing ischemia or
tissue death.
Stenoses are characterized by being very focal (i.e., they only extend over
a short distance) and generally eccentric (i.e., asymmetric) in shape.
From a fluid mechanics standpoint, stenoses are sites that can produce
relatively large changes in local velocity as governed by the Conservation
of Mass

V1 A1 = V2 A2

or

A 
V2 =  1  V1
 A2 
7328_C006.fm Page 242 Monday, October 16, 2006 10:51 AM

242 Biofluid Mechanics: The Human Circulation

V2

A2

V1

A1

FIGURE 6.29
Schematic of flow through a stenosis.

where v1,2 are the mean velocities and A1,2 are the areas of cross sections 1
and 2, respectively.
Since A1 is always greater than A2 (site of stenosis), V2 will be larger than
V1. This increased velocity is accompanied by formation of a jet through the
stenosis (Figure 6.29).
Due to the Bernoulli effect, the increased velocity present at the throat of
the stenosis also produces a reduced pressure. With increasing degree of
stenosis, the pressure at the throat can decrease significantly and may even
cause the vessel to collapse. Distal to the stenosis, vortices may be present
and the flow may become turbulent with more severe degrees of stenosis.
In these cases of advanced stenosis, autopsy results have shown that the
plaques actually rupture, producing thrombus deposition at this site, which
induces a sudden occlusion of the vessel. Research studies are also being
7328_C006.fm Page 243 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 243

performed to characterize unstable plaques in order to better predict their

rupture.
Upon exit into the nondiseased artery downstream, the high jet momen-
tum causes it to persist for a finite distance before expanding to fill the normal
vessel cross-sectional area. In this distal region, the flow is said to separate
from the vessel wall before eventually reattaching. This separated flow
region between the jet and the vessel wall contains slower moving and
recirculating (i.e., reversing) flow. Typical simulations of flow velocity and
wall shear stress distribution in 40 to 80% area reduction stenosis of axisym-
metric geometry are shown in Figure 6.30. Studies have shown that flow
separation and reattachment generally start with stenosis with 40% occlusion
and the region of separation increases with increasing flow rate. Formation
of multiple vortices is seen with higher area occlusion. As can be observed,
flow separation distal to the stenosis is not observed with the asymmetric
stenosis. Further growth in the lesion is probably required for flow separa-
tion to be induced with asymmetric stenosis. These simulations are all based
on steady flow analysis in rigid stenosis of regular geometry, where an actual
stenosis will have an irregular three-dimensional geometry. With further
growth in stenosis, flow separation, pressure fluctuations, and turbulent
stresses may induce rupture at the distal site resulting in thrombus growth
and occlusion of the vessel or emboli lodging in the distal flow field creating
cardiac arrest or neurological deficit.
In some cases, blood in the jet may accelerate to velocities where the
Reynolds number may be in the turbulent range. One consequence of this is
the production of heat and diagnostically useful sounds. In stenosed heart
valves, these sounds are called murmurs while in stenosed arteries they are
referred to as Bruits. Because of the observed complex flow patterns, there
is a measurable pressure drop across a stenosis due to energy losses associ-
ated with recirculating and nonaxial flow movement and the possible onset
of turbulence. This pressure drop may be approximated by considering the
stenosis as a sudden contraction/expansion combination similar to the
“modified” Bernoulli equation used earlier in the analysis of curves and
branches to obtain

 V2 
∆p = K L  ρ 
 2 

where KL for a rounded entrance would be approximately 0.04, and KL for a

rounded exit would be approximately 1.0.
Examples of pulsatile flow through a series of symmetric stenoses are
shown in Figure 6.31 to Figure 6.34.
The diagrams illustrate flow patterns at and downstream of symmetrical
stenoses of 20%, 40%, 60%, and 80% diameter reduction under flow conditions
simulating those in a carotid artery. As can be seen, a jet is formed through
the stenosis, which reaches a maximum length for the 40% and 60% cases.
7328_C006.fm Page 244 Monday, October 16, 2006 10:51 AM

244 Biofluid Mechanics: The Human Circulation

0.02

0.015
Y

0.01 Inlet velocity = 0.2 m/s

Reynold’s number = 483
40% area stenosis
Reattachment point: x = 0.0709 m
0.005

0
0.05 0.055 0.06 0.065 0.07
X
(a)

0.02 WSS (Pa)

7.41773
6.87834
6.33895
5.79956
5.26017
0.015 4.72078
4.18139
3.642
3.10261
Y

2.56322
0.01 2.02383
1.48444
Inlet velocity = 0.2 m/s 0.945055
Reynold’s number = 483 0.405666
40% area stenosis –0.133724
0.005

0
0.04 0.045 0.05 0.055 0.06 0.065
X
(b)

FIGURE 6.30
Flow simulation in varying degrees of symmetric stenoses displaying the stream lines and shear
stress distribution.
7328_C006.fm Page 245 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 245

0.02

0.015
Y

0.01
Inlet velocity = 0.2 m/s
Reynold’s number = 483
60% area stenosis
0.005 Reattachment point: x = 0.06868 m

0
0.05 0.055 0.06 0.065 0.07
X
(c)

0.02 WSS (Pa)

17.6431
16.3444
15.0456
13.7468
12.4481
0.015 11.1493
9.85052
8.55176
7.25299
Y

5.95422
0.01 4.65545
3.35669
Inlet velocity = 0.2 m/s 2.05792
Reynold’s number = 483 0.75915
60% area stenosis –0.539618
0.005

0
0.04 0.045 0.05 0.055 0.06 0.065
X
(d)

FIGURE 6.30 (CONTINUED).

7328_C006.fm Page 246 Monday, October 16, 2006 10:51 AM

246 Biofluid Mechanics: The Human Circulation

0.02

0.015
Y

0.01
Inlet velocity = 0.2 m/s
Reynold’s number = 483
80% area stenosis Reattachment point: x = 0.13267 m
0.005 (Beyond this image)

0
0.05 0.055 0.06 0.065 0.07
X
(e)

0.02 WSS (Pa)

59.9533
55.4764
50.9995
46.5227
42.0458
0.015 37.5689
33.092
28.6152
24.1383
Y

19.6614
0.01 15.1846
Inlet velocity = 0.2 m/s 10.7077
6.23081
Reynold’s number = 483 1.75394
80% area stenosis –2.72294
0.005

0
0.04 0.045 0.05 0.055 0.06 0.065
X
(f )

FIGURE 6.30 (CONTINUED).

7328_C006.fm Page 247 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 247

FIGURES 6.31
Flow patterns through symmetric stenoses of 20% diameter reduction. (From Rittgers, S.E. and
Shu, M.C.S. (1990) Doppler color-flow images from a stenosed arterial model: Interpretation of
flow patterns, in J. Vasc. Surg. 12: 511–22. With permission.)

With further stenosis (80% case), the jet shortens and becomes turbulent.
In all cases, a separation zone is seen outside the jet in which flow recircu-
lation occurs across a region proportional in size to the degree of stenosis.
It can further be seen that these patterns fluctuate during the flow cycle,

FIGURES 6.32
Flow patterns through symmetric stenoses of 40% diameter reduction. (From Rittgers, S.E. and
Shu, M.C.S. (1990) Doppler color-flow images from a stenosed arterial model: Interpretation of
flow patterns, in J. Vasc. Surg. 12 : 511–22. With permission.)
7328_C006.fm Page 248 Monday, October 16, 2006 10:51 AM

248 Biofluid Mechanics: The Human Circulation

FIGURES 6.33
Flow patterns through symmetric stenoses of 60% diameter reduction. (From Rittgers, S.E. and
Shu, M.C.S. (1990) Doppler color-flow images from a stenosed arterial model: Interpretation of
flow patterns, in J. Vasc. Surg. 12 : 511–22. With permission.)

FIGURES 6.34
Flow patterns through symmetric stenoses of 80% diameter reduction. (From Rittgers, S.E. and
Shu, M.C.S. (1990) Doppler color-flow images from a stenosed arterial model: Interpretation of
flow patterns, in J. Vasc. Surg. 12 : 511–22. With permission.)
7328_C006.fm Page 249 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 249

Wss
2.45699
2.29319
Asymmetric stenosis 1 Asymmetric stenosis 1 2.12939
1.96559
1.80179
0.015 0.015 1.63799
1.47419
1.31039
1.14659
0.982794
0.01 0.01 0.818995
0.655196
0.491397
Y

Y
0.327598
0.163799
0.005 0.005

0 0

0.045 0.05 0.055 0.06 0.065 0.07 0.045 0.05 0.055 0.06 0.065 0.07
Z Z
(a) (b)

Wss
20 2.45699
2.29319
Asymmetric stenosis 2 Asymmetric stenosis 2 2.12939
20 1.96559
1.80179
15 1.63799
1.47419
1.31039
15 1.14659
0.982794
0.818995
10 0.655196
Y

10 0.491397
Y

0.327598
0.163799
5 5

0 0

50 60 50 60 70
Z Z
(c) (d)

FIGURE 6.35
Flow simulation displaying the stream lines and shear stress distribution with varying degrees
of asymmetric stenoses.

with the jet being most pronounced at either peak systole or during the
deceleration phase. Flow separation zones are generally larger and more
stable during the late diastolic phase. The presence of a severe stenosis
(Figure 6.34) may cause a reduced flow rate through the artery and then a
lack of oxygen (ischemia) and nutrients to tissues downstream. Mild to
moderate stenoses, however, often do not produce a significant reduction in
blood flow because of the vascular bed dilatation that occurs downstream,
compensating for the increased resistance upstream.
Stenosis eccentricity produces a further elevation of the wall shear stresses,
especially along the crown, or cap, of the plaque as well as enlarging the
size of the separation zone (Figure 6.35 and Figure 6.36).
It is easy to see from these data how wall shear stresses might cause plaque
cap rupture in which the connective tissue covering the atherosclerotic
7328_C006.fm Page 250 Monday, October 16, 2006 10:51 AM

250 Biofluid Mechanics: The Human Circulation

Vmax

(a)

(b)

(c)

(d)

FIGURE 6.36
Flow simulation in a 95% eccentric arterial stenosis.

plaque is removed, exposing the highly thrombogenic plaque contents to

the blood stream and inducing rapid thrombosis.
Model results, such as those shown above, are very useful in identifying the
most relevant parameters to measure for diagnostic purposes and also for
determining the potential for clinical consequences (i.e,. thrombosis formation)
7328_C006.fm Page 251 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 251

by studying the high shear levels in the jet and the long residence times in
the recirculation zones. If the flow is turbulent, it may be severe enough to
cause hemolysis of red blood cells or activation of platelets. This latter effect
is an important consequence since, once activated, platelets may enter the
recirculation zones and reside long enough to attach to the vessel wall and
stimulate other platelets to become activated. If the resultant platelet aggre-
gation proceeds to a critical mass, a thrombus may form and grow to occlude
the entire vessel.
Another important clinical problem involves the abnormal dilation of
major arteries, called aneurysm formation. Abdominal aortic aneurysms
(AAAs) are by far the most common sites, while aneurysms of the descend-
ing thoracic aorta, transverse aortic arch, thoraco-abdominal aorta, and
popliteal artery are also seen, often in conjunction with AAAs (Figure 6.37).
It is thought that an aneurysm forms due to a weakening of the vessel
wall, possibly because of an excess of a degradation enzyme known as
elastase, which acts to break down elastin, a normal load-bearing fiber in
the artery wall. There are several biomechanical implications of this disease.
One is that the vessel will experience increased wall stresses, which, if
allowed to progress, will eventually rupture the vessel. The level of wall
stresses can be approximated by using LaPlace’s Equation for a symmetric,
thin-walled cylinder (Figure 6.3). Summing the forces acting on the upper
half of the vessel and setting them equal to zero gives

pR
σθ = (2.24)
t

where σ is the circumferential wall stress [N/m2], t is the wall thickness [m],
p is the internal pressure [N/m2], and R is the cylinder radius [m]. Thus, the
stress in the wall will increase directly with its diameter, assuming that the
pressure and wall thickness remain constant (in practice, the wall also thins
due to conservation of mass and continued enzymatic activity). This, in
turn, further increases the vessel diameter until it reaches a size and wall
stress beyond which rupture is imminent. From clinical experience, an
abdominal aorta of 5 cm diameter (compared to a normal diameter of ≤ 2 cm)
has a rupture risk of about 5% per year and is, therefore, used as a value for
initiating treatment (usually placement of a bypass graft). It is critical to
diagnose the condition by this time since patients with aneurysms that
rupture have a mortality rate of about 50%.
From a fluid mechanics viewpoint, this large expansion of the vessel pro-
duces unusual flow characteristics that include large vortices near the vessel
walls and a central stream whose dimensions depend on the size and eccen-
tricity of the AAAs. Vortices form in the bulge region of an aneurysm and
due to the Bernoulli effect, the pressure at the bulged cross section increases.
Due to stagnant blood in the bulge, thrombus deposition and growth can
also occur in the bulge region. Figure 6.38 shows examples of flow simulation
7328_C006.fm Page 252 Monday, October 16, 2006 10:51 AM

252 Biofluid Mechanics: The Human Circulation

Descending
thoracic aorta
Transverse arch
of aorta

Ascending
aorta

Abdominal
aorta

Thoracoabdominal
aorta

Descending
aorta

FIGURE 6.37
Patterns of aneurysmal vascular disease. (From Gotto, A.M. et al. (1997) Atherosclerosis, in
Surgical Management of Atherosclerotic Vascular Disease, DeBakey, M.E. and McCollum, M.E., III,
Eds., Pfizer Company, New York. With permission.)

in asymmetric aneurysms including the stream lines, wall shear stress, and
pressure distribution in the bulge region. The increase in transmural pressure
can induce additional forces on the wall and, hence, the bulge can grow.
Conservation of Mass requires that the average velocity through the aneu-
rysm must decrease and since blood has a tendency to clot when moving
7328_C006.fm Page 253 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 253

Wss
1.43277
Asymmetric aneurysm 1 Asymmetric aneurysm 1 1.33382
0.025 0.025 1.23488
1.13593
1.03698
0.938038
0.02 0.02 0.839091
0.740145
0.641198
0.542252
0.015 0.015 0.443305
0.344359
Y

Y
0.245412
0.146466
0.01 0.01 0.047519
–0.0514275
–0.150374
0.005 0.005

0 0

0.05 0.06 0.07 0.05 0.06 0.07

Z Z
(a) (b)

Asymmetric aneurysm 2 Asymmetric aneurysm 2 Wss

0.025 0.025 1.33507
1.23334
1.13161
1.02988
0.02 0.02 0.928156
0.826427
0.724698
0.622969
0.015 0.015 0.52124
0.419511
0.317782
Y

0.216053
0.01 0.01 0.114324
0.012595
–0.089134

0.005 0.005

0 0

0.04 0.05 0.06 0.07 0.04 0.05 0.06 0.07

Z Z
(c) (d)

FIGURE 6.38
Simulation of stream lines, shear stress, and pressure distribution in an asymmetric abdominal
aortic aneurysm.

slowly or when static, most aneurysms develop thrombus within the dilated
pockets. Some may then form a pseudo flow channel similar in size to that
of the nondiseased (normal) vessel. This can be quite dangerous because the
aorta may appear normal on radiographic arteriography when, in fact, the
vessel wall is highly stretched and weakened. In some cases, thrombus from
an AAA may embolize to distal sites, such as the foot, producing an ischemic
condition known as a “blue toe” syndrome. Surgical treatment of AAA
involves exposure of the aorta, removal of the diseased segments, and place-
ment of a graft (typically Dacron) and this also has an associated risk to the
patient. Since not all AAAs larger than 5 cm rupture and some smaller
aneurysms do, biomechanical studies are ongoing in order to study their
fluid dynamics and effects upon the wall to better predict wall stress distri-
butions in the AAA and the possible potential for rupture.
7328_C006.fm Page 254 Monday, October 16, 2006 10:51 AM

254 Biofluid Mechanics: The Human Circulation

6.8 Summary
In this chapter, some of the unsteady flow models to simulate blood flow in
arteries are presented. Even though the wave propagation through the elastic
tube is included in the models, more emphasis is put on the nature of blood
flow through the arteries. Description of the various hemodynamic theories
correlating flow dynamics in curved arterial segments, as well as in regions
of bifurcations and branches with the genesis and growth of atheroma in
these regions, has been included. The nature of unsteady flow dynamics in
arterial curvature and bifurcation sites, as well as in the region of stenosis
and aneurysm is also discussed.

6.9 Problems
6.1 Show that the Womersley parameter is analogous to the Reynolds num-
ber for unsteady flow.
6.2 For a typical human aorta, the diameter of the lumen is 30 mm and the
thickness of the wall is 4 mm. Assuming a blood density of 1.056 gm/cc and
a viscosity of 0.035 Poise, calculate the Womersley number if the heart rate
is 72 bpm. Diameters for the carotid and femoral arteries of a typical human
are about 0.8 cm and 0.5 cm, respectively. Compute the Womersley numbers
for each of these arterial segments for the same heart rate.
6.3 In an experiment on an exposed abdominal aorta of a dog, the pulse
wave speed is determined to be 1.5 m/s. The wall thickness of the artery
was measured to be 5% of the diameter. Estimate the Young’s modulus for
the aorta. In another segment of the same artery, the pulse wave speed was
double the previous value. What can you conclude from this?
6.4 List five assumptions made in deriving the Moens–Korteweg relation-
ship for wave propagation.
6.5 In deriving the relationship for the velocity profiles for oscillatory
motion in an elastic tube, Morgan and Kiely assumed that the pressure
gradient followed the relationship

p = Pe i ( kz−ωt ) where k = k1 + ik2

Under these conditions, show that k2 represents a damping constant.

6.6 In the derivations of velocity profiles for oscillatory flow of viscous
fluid through an elastic tube assuming axisymmetric flow (Figure 6.3), list
7328_C006.fm Page 255 Monday, October 16, 2006 10:51 AM

Unsteady Flow and Nonuniform Geometric Models 255

the boundary conditions used in obtaining the solution. Here, u and w are
the velocity components, while η and ξ are the tube displacements in the
radial and axial directions, respectively.
6.7 The pressure gradient and flow rate measured in the femoral artery of
a dog are shown in Figure 6.11. Resolve them into their first four harmonics
and then plot them. Also, plot the sum of the four harmonics and compare
this with the measured signals.
7328_C006.fm Page 256 Monday, October 16, 2006 10:51 AM
7328_C007.fm Page 257 Monday, September 18, 2006 2:06 PM

7
Native Heart Valves

CONTENTS
7.1 Introduction ................................................................................................257
7.2 Aortic and Pulmonic Valves ....................................................................259
7.2.1 Mechanical Properties ..................................................................261
7.2.2 Valve Dynamics .............................................................................263
7.3 Mitral and Tricuspid Valves.....................................................................267
7.3.1 Mechanical Properties ..................................................................269
7.3.2 Valve Dynamics .............................................................................270

7.1 Introduction
The heart has four valves whose main function is to control the direction of
blood flow through the heart, permitting forward flow and preventing back
flow. On the right side of the heart, the tricuspid and pulmonic valves
regulate the flow of blood that is returned from the body to the lungs for
oxygenation. The mitral and aortic valves control the flow of oxygenated
blood from the left side of the lungs back to the body. The aortic and
pulmonic valves allow blood to be pumped from the ventricles into arteries
on the left and right side of the heart, respectively. Similarly, the mitral and
tricuspid valves lie between the atria and ventricles of the left and right sides
of the heart, respectively. The aortic and pulmonic valves open during systole
when the ventricles are contracting and close during diastole when the
ventricles are filling with blood, which enters through the open mitral and
tricuspid valves. During isovolumic contraction and relaxation, all four
valves are closed (Figure 7.1a). In Figure 7.1b, the pressure and volume
information for the left heart is shown to illustrate the timings of a normal
cardiac cycle. The details of the cardiac events and the opening of the valves
are included in Chapter 3, section 3.3: Cardiac Valves.
When closed, the pulmonic and tricuspid valves must withstand a pressure
of approximately 30 mmHg. The closing pressures on the left side of the
heart, however, are much higher. Specifically, the aortic valve withstands

257
7328_C007.fm Page 258 Monday, September 18, 2006 2:06 PM

258 Biofluid Mechanics: The Human Circulation

Aorta

Left pulmonary
artery
Right pulmonary
artery
Pulmonary trunk
Superior vena Left atrium
cava Mitral valve

Pulmonary valve Aortic valve

Left ventricle
Right atrium Muscular
interventricular
septum
Tricuspid valve
Right ventricle

Schematic of the heart

(a)

120 Aorta
Pressure (mm Hg)

Left
ventricle

20 Left atrim

0 0.2 0.4 0.6 0.8

Time (sec)
(a)

100
Left
Volume (ml)

ventricle
60
Left atrim
20

0 0.2 0.4 0.6 0.8

Time (sec)
(b)

FIGURE 7.1
(a) A schematic of the four chambers of the heart with the valves, (b) typical pressure and flow
curves for the left heart.
7328_C007.fm Page 259 Monday, September 18, 2006 2:06 PM

Native Heart Valves 259

pressures of approximately 100 mmHg, while the mitral valve resists pres-
sures up to 150 mmHg. Since diseases of the valves on the left side of the
heart are more prevalent than those on the right side, most of this chapter
will focus on the aortic and mitral valves. Where pertinent, reference will be
made to the pulmonic and tricuspid valves, which tend to be impacted more
by congenital heart disease.

7.2 Aortic and Pulmonic Valves

The aortic valve is composed of three semilunar cusps, or leaflets, contained
within a connective tissue sleeve (Figure 7.2). The valve cusps are attached
to a fibrous ring embedded in the fibers of the ventricular septum and the
anterior leaflet of the mitral valve. Each of the leaflets is lined with endo-
thelial cells and has a dense collagenous core adjacent to the high-pressure
aortic side (Figure 7.3). The side adjacent to the aorta is termed the fibrosa

Sinuses
of valsalva

RCC

LCC

Aorta
left ventricle

NCC

FIGURE 7.2
Illustration of the aortic sinuses and valve in the closed position. The noncoronary cusp (NCC)
is in front. The left and right coronary cusps (LCC and RCC) are positioned as marked. The
aorta is above the closed valve in this orientation and the left ventricle is below the dashed
line. (From Yoganathan, A.P. et al. (2000) Heart valve dynamics, in The Biomedical Engineering
Handbook, 2nd ed., J.D. Bronzino (Ed.), CRC Press, Boca Raton, FL. With permission.)
7328_C007.fm Page 260 Monday, September 18, 2006 2:06 PM

260 Biofluid Mechanics: The Human Circulation

Cusp free
edge Aorta S

V
Fibrosa
Spongiosa

Ventricularis

FIGURE 7.3
Schematic and histologic section of an aortic valve leaflet showing the fibrosa, spongiosa, and
ventricularis layers.

and is the major fibrous layer within the belly of the leaflet. The layer
covering the ventricular side of the valve is called the ventricularis and is
composed of both collagen and elastin. The ventricularis is thinner than the
fibrosa and presents a very smooth surface to the flow of blood. The central
portion of the valve, called the spongiosa, contains variable loose connective
tissue, proteins, and glycosaminoglycans (GAGs) and is normally not vas-
cularized. The collagen fibers within the fibrosa and ventricularis are not
organized in the unstressed state, but when a stress is applied, they become
oriented primarily in the circumferential direction with a lower concentra-
tion of elastin and collagen in the radial direction.
The fibrous annular ring of the aortic valve separates the aorta from the
left ventricle. Superior to this ring is a structure called the sinus of Valsalva,
or the aortic sinus. The sinus is comprised of three bulges at the root of the
aorta, with each bulge aligned with the belly or central part of the specific
valve leaflet. Each valve cusp and corresponding sinus is named according
to its anatomical location within the aorta. Two of these sinuses give rise
to coronary arteries that branch off the aorta, providing blood flow to the
heart itself. The right coronary artery is based at the right or right anterior
sinus, the left coronary artery exits the left or left posterior sinus, and the
third sinus is called the noncoronary or right posterior sinus. Figure 7.2
shows the configuration of the normal aortic sinuses and valve in the closed
position.
Because the length of the aortic valve cusps is greater than the annular
radius, a small overlap of tissue from each leaflet protrudes and forms a
coaptation surface within the aorta when the valve is closed. This overlapped
tissue, called the lunula, may help to ensure that the valve is sealed. When
the valve is in the open position, the leaflets extend to the upper edge of the
sinuses of Valsalva. The anatomy of the pulmonic valve is similar to that of
the aortic valve, but the surrounding structure is slightly different. The main
7328_C007.fm Page 261 Monday, September 18, 2006 2:06 PM

Native Heart Valves 261

differences are that the sinuses are smaller in the pulmonary artery and the
pulmonic valve annulus is slightly larger than that of the aortic valve.
The dimensions of the aortic and pulmonic valves and their leaflets have
been measured in a number of ways. Before noninvasive measurement tech-
niques, such as echocardiography, became available, valve measurements
were recorded from autopsy specimens. An examination of 160 pathologic
specimens revealed the aortic valve diameter to be 23.2 ± 3.3 mm, whereas
the diameter of the pulmonic valve was measured at 24.3 ± 3.0 mm. However,
according to M-mode echocardiographic measurements, the aortic root
diameter at end systole was 35 ± 4.2 mm and 33.7 ± 4.4 mm at the end of
diastole. The differences in these measurements reflect the fact that the
autopsy measurements were not performed under physiologic pressure con-
ditions and that intrinsic differences exist in the measurement techniques.
On average, pulmonic leaflets are thinner than aortic leaflets: 0.49 mm vs.
0.67 mm, although the leaflets of the aortic valve show variable dimensions
depending on the respective leaflet. For example, the posterior leaflet tends
to be thicker, have a larger surface area, and weigh more than the right or
left leaflet, and the average width of the right aortic leaflet is greater than
that of the other two.

7.2.1 Mechanical Properties

Due to the location and critical function of the aortic valve, it is difficult to
obtain measurements of its mechanical properties in vivo; however, reports
are available from a small number of animal studies. This section will refer-
ence the in vivo data whenever possible and defer to the in vitro data when
necessary. Since little mathematical modeling of the aortic valve’s material
properties has been reported, it will be sufficient to describe the known
mechanical properties of the valve. Like most biological tissues, the aortic
valve is anisotropic, inhomogeneous, and viscoelastic. The collagen fibers
within each valve cusp are aligned along the circumferential direction. Vesely
and Noseworthy (1992) found that both the ventricularis and fibrosa were
stiffer in the circumferential direction than in the radial direction. However,
the ventricularis was more extensible radially than circumferentially, while
the fibrosa had uniform extensibility in both directions.
There are also elastin fibers in each cusp, at a lesser concentration than the
collagen, that are oriented orthogonal to the collagen. It is this fiber structure
that accounts for the anisotropic properties of the valve. The variation in
thickness and composition across the leaflets is responsible for their inho-
mogeneous material properties. Although the aortic valve leaflet, as a whole,
is asymmetric in its distensibility, the basal region tends to be relatively
isotropic, while the central region shows the greatest degree of anisotropy.
The role and morphology of elastin and the way elastin is coupled to collagen
remain points of investigation. Scott and Vesely (1996) have shown that the
elastin in the ventricularis consists of continuous amorphous sheets or compact
7328_C007.fm Page 262 Monday, September 18, 2006 2:06 PM

262 Biofluid Mechanics: The Human Circulation

meshes, while elastin in the fibrosa consists of complex arrays of large tubes
that extend circumferentially across the leaflet. These tubes may surround
the large circumferential collagen bundles in the fibrosa. Mechanical testing
of elastin structures from the fibrosa and ventricularis have separately shown
that the purpose of elastin in the aortic valve leaflet is to maintain a specific
collagen fiber configuration and to return the fibers to that state during cyclic
loading. The valve's viscoelastic properties are actually dominated by the
elastic component (over the range of in vitro testing) so that the viscous
effects, which are largely responsible for energy losses, are small.
In addition to the collagen and elastin, clusters of lipids have been
observed in the central spongiosa of porcine aortic valves. Vesely et al. (1994)
have shown that the lipids tend to be concentrated at the base of the valve
leaflets, while the coaptation regions and free edges of the leaflets tend to
be devoid of these lipids. In addition, the spatial distribution of the lipids
within the spongiosal layer of the aortic leaflets corresponded to areas in
which calcification is commonly observed on bioprosthetic valves, suggesting
that these lipid clusters may be potential nucleation sites for calcification. In
contrast, pulmonic leaflets showed a substantially lower incidence of lipids.
The aortic valve leaflets have also been shown to be slightly stiffer than
pulmonary valve leaflets although the extensibilities and relaxation rates of
the two tissues are similar.
Using fluoroscopy, in which the aortic valve leaflets were surgically tagged
with radio-opaque markers and imaged with high speed x-rays, the leaflets
have been shown to change length during the cardiac cycle. The cusps are
longer during diastole than systole in both the radial and circumferential
direction. The variation in length is greatest in the radial direction, approx-
imately 20%, while the strain in the circumferential direction is about 10%
of the normal systolic length. The difference in strain is due to the presence
of the compliant elastin fibers aligned in this radial direction. The length
change in both directions results in an increased valve surface area during
diastole. During systole, the shortening of the valve leaflets helps to reduce
obstruction of the aorta during the systolic ejection of blood. It should be
noted that this change in area is by no means an active mechanism within
the aortic valve; the valve simply reacts to the stresses it encounters in a
passive manner.
In addition to this change in surface area, the aortic valve leaflets also
undergo bending in the circumferential direction during the cardiac cycle. In
diastole when the valve is closed, each leaflet is convex toward the ventricular
side. During systole when the valve is open, the curvature changes and each
leaflet is concave toward the ventricle. The bending is localized on the valve
cusp near the wall of the aorta. This location is often thicker than the rest of
the leaflet. The total diastolic stress in an aortic valve leaflet has been esti-
mated at 2.5 × 106 dynes/cm2 for a strain of 15%. The stress in the circumfer-
ential direction was found to be the primary load-bearing element in the
aortic valve. Due to the collagen fibers oriented circumferentially, the valve
is relatively stiff in this direction. The strain that does occur circumferentially
7328_C007.fm Page 263 Monday, September 18, 2006 2:06 PM

Native Heart Valves 263

is primarily due to scissoring of the fibrous matrix and straightening of

collagen fibers that are kinked or crimped in the presence of no external stress.
However, in the radial direction, because elastin is the primary element, the
valve can undergo a great deal of strain, ranging from 20 to 60% in tissue
specimens. In the closed position, the radial stress levels are actually small
compared to those in the circumferential direction. This demonstrates the
importance of the lunula in ensuring that the valve seals tightly to prevent
leakage. Because of their anatomical location, the lunula cause these high
circumferential stress levels by enabling the aortic pressure to pull each leaflet
in the circumferential direction toward the other leaflets.
The composition, properties, and dimensions of the aortic valve change
with age and in the presence of certain diseases. The valve leaflets become
thicker, the lunula become fenestrated (or mesh-like), and, in later stages of
disease, the central portion of the valve may become calcified. This thicken-
ing of the valve typically occurs on the ventricular side of the valve, in the
region where the tips of the leaflets come together. Another site of calcifica-
tion and fibrosis is the point of maximum cusp flexion and it is thought to
be a response to fatigue in the normal valve tissue.

7.2.2 Valve Dynamics

The aortic valve opens during systole when the ventricle is contracting and
then closes during diastole as the ventricle relaxes and fills from the atrium.
Systole lasts about one-third of the cardiac cycle and begins when the aortic
valve opens, which typically takes only 20 to 30 msec. Blood rapidly accel-
erates through the valve and reaches a peak velocity after the leaflets have
opened to their full extent and start to close again. Peak velocity is reached
during the first third of systole and the flow begins to decelerate rapidly
after the peak is reached, albeit not as fast as its initial acceleration. The
adverse pressure gradient that is developed affects the low momentum fluid
near the wall of the aorta more than that at the center; this causes reverse
flow in the sinus region. During systole, the pressure difference required to
drive the blood through the aortic valve is on the order of a few millimeters
of mercury; however, the diastolic pressure difference reaches 80 mmHg in
normal individuals. The valve closes near the end of the deceleration phase
of systole with very little reverse flow (< 2 ml/beat) through the valve.
During the cardiac cycle, the heart undergoes translation and rotation due
to its own contraction pattern. As a result, the base of the aortic valve varies
in size and also translates, mainly along the axis of the aorta. Using marker
fluoroscopy to study the base of the aortic valve in dogs, Thubrikar et al. (1981)
found that the base perimeter is at its largest at end diastole and decreases
in size during systole; it then reaches a minimum at the end of systole and
increases again during diastole. The range of this perimeter variation during
the cardiac cycle was 22% for an aortic pressure variation of 120/80 mmHg.
The valve annulus also undergoes translation, primarily parallel to the
7328_C007.fm Page 264 Monday, September 18, 2006 2:06 PM

264 Biofluid Mechanics: The Human Circulation

aortic axis. The aortic annulus moves downward toward the ventricle during
systole and then recoils back toward the aorta as the ventricle fills during
diastole. The annulus also experiences a slight side-to-side translation with
its magnitude approximately one-half the displacement along the aortic axis.
During systole, vortices develop in all three sinuses behind the leaflets of
the aortic valve. Leonardo da Vinci first described the function of these vor-
tices in 1513, and they have been researched extensively in this century
primarily through the use of in vitro models. It has been hypothesized that
the vortices help to close the aortic valve so that blood is prevented from
returning to the ventricle during the closing process. These vortices create a
transverse pressure difference that pushes the leaflets toward the center of
the aorta and each other at the end of systole, thus minimizing any possible
closing volume. However, as shown in vitro by Reul and Talukdar (1981), the
axial pressure difference alone is enough to close the valve. Without the vortices
in the sinuses, the valve still closes, but its closure is not as quick as when
the vortices are present. The adverse axial pressure difference within the aorta
causes the low inertia flow within the developing boundary layer along the
aortic wall to be the first to decelerate and reverse direction. This action forces
the belly of the leaflets away from the aortic wall and toward the closed
position. When this force is coupled with the vortices that push the leaflet
tips toward the closed position, a very efficient and fast closure is obtained.
Closing volumes have been estimated to be less than 5% of the forward flow.
The parameters that describe the normal blood flow through the aortic
valve are the velocity profile, time course of the blood velocity or flow, and
magnitude of the peak velocity. These are determined in part by the pressure
difference between the ventricle and aorta and by the geometry of the aortic
valve complex. As seen in Figure 7.4, the velocity profile at the level of the
aortic valve annulus is relatively flat. However there is usually a slight skew
toward the septal wall (less than 10% of the center-line velocity), which is
caused by the orientation of the aortic valve relative to the long axis of the
left ventricle. This skew in the velocity profile has been shown by many
experimental techniques, including hot film anemometry, Doppler ultra-
sound, and magnetic resonance imaging (MRI). In healthy individuals, blood
flows through the aortic valve at the beginning of systole and then rapidly
accelerates to its peak value of 1.35 ± 0.35 m/s; for children this value is
slightly higher at 1.5 ± 0.3 m/s. At the end of systole, there is a very short
period of reverse flow that can be measured with Doppler ultrasound. This
reverse flow is probably either a small closing volume or the velocity of the
valve leaflets as they move toward their closed position. In the presence of
aortic stenosis, peak velocities through the valve could be as high as 5 m/s
and large magnitudes of turbulent shear stresses that could induce hemolysis.
The flow patterns just downstream of the aortic valve are of particular
interest because of their complexity and relation to arterial disease. Highly
skewed velocity profiles and corresponding helical flow patterns have been
observed in the human aortic arch using magnetic resonance phase velocity
mapping.
7328_C007.fm Page 265 Monday, September 18, 2006 2:06 PM

Native Heart Valves 265

50 cm/mm

L P L P L

12 m
m
A R A R A

P L P L P L

R A R A R A

P L P L P L

R A R A R A

P P L
P L L

R A R A
R A

FIGURE 7.4
Velocity profiles measured 2 cm downstream of the aortic valve with hot film anemometry in
dogs. The marker on the aortic flow curve shows the timing of the measurements during the
cardiac cycle. (From Paulsen, P. K. and Hansenkam, J. M. (1983). Three-dimensional visualiza-
tion of velocity profiles in the ascending aorta in dogs, measured with a hot-film anemometer,
in J. Biomech. 16: 201–210. With permission.)

Flow through the pulmonic valve behaves similarly to that of the aortic
valve, but the magnitude of the velocity is not as great. Typical peak velocities
for healthy adults are 0.75 ± 0.15 m/s; again these values are slightly higher
for children at 0.9 ± 0.2 m/s. As seen in Figure 7.5, a rotation of the peak
velocity can be observed in the pulmonary artery velocity profile. During
acceleration, the peak velocity is observed inferiorly with the peak rotating
counterclockwise throughout the remainder of the ejection phase. The mean
7328_C007.fm Page 266 Monday, September 18, 2006 2:06 PM

266 Biofluid Mechanics: The Human Circulation

m/s
1.0
0.8
39 ms 70 ms 101 ms 0.6

Trans_neg
0.4
Left 0.2
Inf. 0.0
–0.2
Sup. –0.4
Right
(a) (b) (c)

m/s
132 ms 163 ms 256 ms 1.0
0.8
0.6

Trans_neg
0.4
0.2
0.0
–0.2
–0.4

(d) (e) (f )

m/s
1.0
0.8
349 ms 442 ms 721 ms 0.6
Trans_neg

0.4
0.2
0.0
–0.2
–0.4

(g) (h) (i)

FIGURE 7.5
Velocity profiles downstream of the human pulmonary valve obtained with magnetic resonance
phase velocity mapping. The timing of the measurements is shown by the marker on the flow
curve. (From Sloth, E. et al. (1994). Three-dimensional visualization of velocity profiles in the
human main pulmonary artery with magnetic resonance phase-velocity mapping, in Am. Heart
J. 128: 1130–1138. With permission.)
7328_C007.fm Page 267 Monday, September 18, 2006 2:06 PM

Native Heart Valves 267

spatial profile is relatively flat, however, although there is a region of reverse

flow that occurs in late systole, which may be representative of flow separation.
Typically, there is only a slight skew to the profile. The peak velocity is gen-
erally within 20% of the spatial mean throughout the cardiac cycle. Secondary
flow patterns can also be observed in the pulmonary artery and its bifurcation.
In vitro laser Doppler anemometry experiments have shown that these flow
patterns are dependent on the valve geometry/degree of stenosis and, thus,
can be used to evaluate function and fitness of the heart valve.

7.3 Mitral and Tricuspid Valves

The mitral (Figure 7.6) and tricuspid valves are similar in structure with both
valves composed of four primary elements: the valve annulus, the valve leaf-
lets, the papillary muscles, and the chordae tendinae. The base of the mitral
leaflets forms the mitral annulus, which attaches to the atrial and ventricular
walls and aortic root. At the free edge of the leaflets, the chordae tendinae insert
at multiple locations and extend to the tips of the papillary muscles. This
arrangement provides continuity between the valve and ventricular wall to
enhance valvular function. The valvular apparatus, or complex, requires an
intricate interplay between all components throughout the cardiac cycle.
The mitral annulus is an elliptical ring composed of dense collagenous tissue
surrounded by muscle. It goes through dynamic changes during the cardiac
cycle by not only changing in size, but also by moving three-dimensionally.

Annulus
Anterior leaflet

Posterior leaflet

Chordae
tendineae
Posteromedial
papillary muscle
Anterolateral
papillary muscle

FIGURE 7.6
Schematic of the mitral valve showing the valve leaflets, papillary muscles, and chordae ten-
dinae. (From Yoganathan, A.P. et al. (2000) Heart valve dynamics, in The Biomedical Engineering
Handbook, 2nd ed., J.D. Bronzino (Ed.), CRC Press, Boca Raton, FL. With permission.)
7328_C007.fm Page 268 Monday, September 18, 2006 2:06 PM

268 Biofluid Mechanics: The Human Circulation

Anterior leaflet Posterior leaflet

Posteromedial Anterolateral
commisure commisure
Medial Lateral
scallop Central scallop
scallop

Chordae Chordae
tendineae tendineae

Posteromedial Anterolateral
papillary muscle papillary muscle

FIGURE 7.7
Diagram of the mitral valve as a continuous piece of tissue. The posterior and anterior leaflets
are indicated, as are the scallops, chordae tendinae, and papillary muscles. (From Yoganathan,
A.P. et al. (2000) Heart valve dynamics, in The Biomedical Engineering Handbook, 2nd ed., J.D.
Bronzino (Ed.), CRC Press, Boca Raton, FL. With permission.)

The circumference of the mitral annulus ranges from 8 to 12 cm during

diastole. Recent studies involving the measurement of annular shape have
also shown that the mitral annulus is not planar, but instead has a three-
dimensional form. The annulus actually forms a saddle, or skislope, shape.
This three-dimensional shape must be taken into account when noninvasively
evaluating valvular function.
The mitral valve is a bileaflet valve comprised of an anterior and posterior
leaflet. The leaflet tissue is primarily collagen-reinforced endothelium, but
also contains striated muscle cells, nonmyelinated nerve fibers, and blood
vessels. The anterior and posterior leaflets of the valve are actually one
continuous piece of tissue, as shown in Figure 7.7. The free edge of this
tissue shows several indentations, two of which are regularly placed, called
the commisures. The commisures separate the tissue into the anterior and
posterior leaflets. The location of the commisures can be identified by the
fan-like distribution of chordae tendinae and the relative positioning of the
papillary muscles. The combined surface area of both leaflets is approxi-
mately twice the area of the mitral orifice. This extra surface area permits
a large line of coaptation and ample coverage of the mitral orifice during
normal function and provides compensation in cases of disease. The pos-
terior leaflet encircles roughly two-thirds of the mitral annulus and is essen-
tially an extension of the mural endocardium from the free walls of the left
atrium. The anterior leaflet portion of the annulus is a line of connection
for the leaflet, the wall of the ascending aorta, the aortic valve, and the atrial
septum. The anterior leaflet is slightly larger than the posterior leaflet and
is roughly semilunar in shape as opposed to the quadrangular-shaped
posterior leaflet. The normal width and height of the anterior leaflet are
approximately 3.3 cm and 2.3 cm, respectively. The height of the posterior
7328_C007.fm Page 269 Monday, September 18, 2006 2:06 PM

Native Heart Valves 269

leaflet is 1.3 cm, while the commisure height is less than 1.0 cm. The pos-
terior leaflet typically has indentations, called scallops, that divide the
leaflet into three regions — the medial, central, and lateral scallops.
The mitral leaflet tissue can be divided into both rough and clear zones.
The rough zone is the thicker part of the leaflet and is defined from the free
edge of the valve to the valve’s line of closure. The term “rough” is used to
denote the texture of the leaflet due to the insertion of the chordae tendinae
in this area. The clear zone is thinner and translucent and extends from the
line of closure to the annulus in the anterior leaflet and to the basal zone in
the posterior leaflet. Unlike the mitral valve, the tricuspid valve has three
leaflets — an anterior leaflet, a posterior leaflet with a variable number of
scallops, and a septal leaflet. The tricuspid valve is larger and structurally
more complicated than the mitral valve and the separation of the valve tissue
into distinct leaflets is less pronounced than with the mitral valve. The
surface of the leaflets is similar to that of the mitral valve; however, the basal
zone is present in all of the leaflets.
Chordae tendinae from both leaflets attach to each of the papillary muscles
(anterior and posterior). The chordae tendinae consist of an inner core of col-
lagen surrounded by loosely meshed elastin and collagen fibers with an outer
layer of endothelial cells. In the mitral complex structure, there are marginal
and basal chordae that insert into the mitral leaflets. From each papillary
muscle, several chordae originate and branch into the marginal and basal
chordae. The thinner marginal chordae insert into the leaflets’ free edge at
multiple insertion points, while the thicker basal chordae insert into the leaflets
at a higher level toward the annulus. The marginal chordae function is to keep
the leaflets stationary while the basal chordae seem to act more as supports.
The left side of the heart has two papillary muscles, called anterolateral
and posteromedial, that attach to the ventricular free wall and tether the
mitral valve in place via the chordae tendinae. This tethering prevents the
mitral valve from prolapsing into the left atrium during ventricular ejection.
On the right side of the heart, the tricuspid valve has three papillary muscles.
The largest one, the anterior papillary muscle, attaches to the valve at the
commissure between the anterior and posterior leaflets. The posterior pap-
illary muscle is located between the posterior and septal leaflets. The smallest
papillary muscle, called the septal muscle, is sometimes not even present.
Improper tethering of the leaflets will result in valve prolapse during ven-
tricular contraction, permitting the valve leaflets to extend into the atrium.
This incomplete apposition of the valve leaflets can cause regurgitation,
which is leaking of the blood being ejected back into the atrium.

7.3.1 Mechanical Properties

Studies on the mechanical behavior of the mitral leaflet tissue have been
conducted to determine the key connective tissue components that influence
the valve function. Histological studies have shown that the tissue is com-
posed of three layers, which can be identified by differences in cellularity
7328_C007.fm Page 270 Tuesday, September 19, 2006 12:39 PM

270 Biofluid Mechanics: The Human Circulation

and collagen density. Analysis of the leaflets under tension indicated that
the anterior leaflet would be more capable of supporting larger tensile loads
than the posterior leaflet. The differences between the mechanical properties
of the two leaflets may require different material selection for repair or
replacement of the individual leaflets.
Studies have also been done on the strength of the chordae tendinae. The
tension of chordae tendinae in dogs was monitored throughout the cardiac cycle
by Salisbury et al. (1963). They found that the tension only paralleled the left
ventricular pressure tracings during isovolumic contraction, indicating slack-
ness at other times in the cycle. Investigation of the tensile properties of the
chordae tendinae at different strain rates by Lim and Boughner (1975) found
that the chordae had a nonlinear stress–strain relationship. They found that the
size of the chordae had a more significant effect on the development of the
tension than did the strain rate. The smaller chordae with cross-sectional areas
of 0.001 to 0.006 cm2 had a modulus of 2 × 109 dynes/cm2, while larger chordae
of cross-sectional areas of 0.006 to 0.03 cm2 had a modulus of 1 × 109 dynes/cm2.
Ghista and Rao (1972) performed a theoretical study of the stresses sus-
tained by the mitral. This study determined that the stress level could reach
as high as 2.2 × 106 dynes/cm2 just prior to the opening of the aortic valve,
with the left ventricular pressure rising to 150 mmHg. A mathematical model
has also been created for the mechanics of the mitral valve. It incorporates
the relationship between chordae tendinae tension, left ventricular pressure,
and mitral valve geometry. This study examined the force balance on a closed
valve, and determined that the chordae tendinae force was always more than
half the force exerted on the mitral valve orifice by the transmitral pressure
gradient. During the past 10 years, computational models have been devel-
oped of mitral valve mechanics, with the most advanced modeling being
three-dimensional finite element models (FEM) of the complete mitral appa-
ratus. Kunzelman et al. (1993) has developed a model of the mitral complex
that includes the mitral leaflets, chordae tendinae, contracting annulus, and
contracting papillary muscles. From these studies, the maximum principal
stresses found at peak loading (120 mmHg) were 5.7 × 106 dynes/cm2 in the
annular region, while the stresses in the anterior leaflet ranged from 2 x ×106
to 4 × 106 dynes/cm2. This model has also been used to evaluate mitral valve
disease, repair in chordal rupture, and valvular annuloplasty.

7.3.2 Valve Dynamics

The valve leaflets, chordae tendinae and papillary muscles all participate to
ensure normal functioning of the mitral valve. During isovolumic relaxation,
the pressure in the left atrium exceeds that of the left ventricle, and the mitral
valve cusps open. Blood flows through the open valve from the left atrium
to the left ventricle during diastole. The velocity profiles at both the annulus
and the mitral valve tips have been shown to be skewed and, therefore, are
not flat, as is commonly assumed. This skewing of the inflow profile is shown
in Figure 7.8.
7328_C007.fm Page 271 Monday, September 18, 2006 2:06 PM

Native Heart Valves 271

Apex of the heart

Aortic valve L
A Posterior mitral
leaflet
P R
Anterior
mitral leaflet
Base of the heart

Velocity (cm/s)
Velocity (cm/s)

20.0 Vel 20.0 Vel

oc 0.0
o ci
0.0 ity ty (
(cm c m/
/s) s)
0.0 500.0 0.0 500.0
60 60
Time (ms) Time (ms)
40 40
20 20
0 0
–20 –20
270 ms –40 470 ms –40
A L A L
(a) P R (b) P R
Velocity (cm/s)
Velocity (cm/s)

20.0 Vel 20.0 Vel

0.0
oc ity 0.0 oc ity
(cm (cm
/s) /s)
0.0 500.0 0.0 500.0
Time (ms) 60 60
Time (ms)
40 40
20 20
0 0
–20 –20
560 ms –40 590 ms –40
A L A L
(c) P R (d) P R
Velocity (cm/s)

20.0
Vel Vel
0.0 oc ity oc ity
(cm (cm
/s) /s)
0.0 500.0
Time (ms) 60 60
40 40
20 20
0 0
–20 –20
670 ms –40 Mean –40
A L A L
(e) P R (f ) P R

FIGURE 7.8
Two-dimensional transmitral velocity profiles recorded at the level of the mitral annulus in a
pig — (a) systole, (b) peak E-wave, (c) deceleration phase of early diastole, (d) mid-diastolic
period (diastasis), (e) peak A-wave, and (f) time-averaged diastolic cross-sectional mitral
velocity profile. (From Kim et al. (1994). Two-dimensional mitral flow velocity profiles in pig
models using epicardial doppler echocardiography, in JACC 24: 532–545. With permission.)
7328_C007.fm Page 272 Monday, September 18, 2006 2:06 PM

272 Biofluid Mechanics: The Human Circulation

The initial filling is enhanced by the active relaxation of the ventricle, main-
taining a positive transmitral pressure. The mitral velocity flow curve shows
a peak in the flow curve, called the E-wave, which occurs during the early
filling phase. Normal peak E-wave velocities in healthy individuals range from
50 to 80 cm/s. Following active ventricular relaxation, the fluid begins to
decelerate and the mitral valve undergoes partial closure. Then, the atrium
contracts and the blood accelerates through the valve again to a secondary
peak, termed the A-wave. The atrium contraction plays an important role in
additional filling of the ventricle during late diastole. In healthy individuals,
velocities during the A-wave are typically lower than those of the E-wave,
with a normal E-wave/A-wave velocity ratio ranging from 1.5 to 1.7. Thus,
normal diastolic filling of the left ventricle shows two distinct peaks in the
flow curve with no flow leaking back through the valve during systole.
The tricuspid flow profile is similar to that of the mitral valve although the
velocities in the tricuspid valve are lower because it has a larger valve orifice.
In addition, the timing of the valve opening is slightly different. Since the peak
pressure in the right ventricle is less than that of the left ventricle, the time
for right ventricular pressure to fall below the right atrial pressure is less than
the corresponding time period for the left side of the heart. This leads to a
shorter right ventricular isovolumic relaxation and, thus, an earlier tricuspid
opening. Tricuspid closure occurs after the mitral valve closes, since the
activation of the left ventricle precedes that of the right ventricle.
A primary focus in explaining the fluid mechanics of mitral valve function
has been understanding the closing mechanism of the valve. Bellhouse and
Bellhouse (1968) first suggested that the vortices generated by ventricular filling
were important for the partial closure of the mitral valve following early diastole.
Their in vitro experiments suggested that without the strong outflow tract vor-
tices, the valve would remain open at the onset of ventricular contraction, thus
resulting in a significant amount of mitral regurgitation before complete closure.
Later in vitro experiments by Reul et al. (1981) in a left ventricle model made
from silicone suggested that an adverse pressure differential in mid-diastole
could explain both the flow deceleration and the partial valve closure, even in
the absence of a ventricular vortex. Thus, the studies by Reul suggest that the
vortices may provide additional closing effects at the initial stage; however, the
pressure forces are the dominant effect in valve closure.
A more unified theory of valve closure put forth by Yellin et al. (1981)
includes the importance of chordal tension, flow deceleration and ventricular
vortices, with chordal tension being a necessary condition for the other two.
The animal studies by Yellin indicated that competent valve closure could
occur even in the absence of vortices and flow deceleration. Recent studies
using magnetic resonance imaging to visualize the three-dimensional flow
field in the left ventricle showed that in normal individuals a large anterior
vortex is present at initial partial closure of the valve, as well as following
atrial contraction. Studies conducted in our laboratory using magnetic
resonance imaging of healthy individuals clearly show vortices in the left
ventricle, which may be an indication of normal diastolic function.
7328_C007.fm Page 273 Monday, September 18, 2006 2:06 PM

Native Heart Valves 273

Another area of interest has been the motion of the mitral valve complex.
The heart itself moves throughout the cardiac cycle and, similarly, the mitral
apparatus also moves and changes shape. Recent studies have been conducted
that examined the three-dimensional dynamics of the mitral annulus during
the cardiac cycle. These studies have shown that, during systole, the annular
circumference decreases from the diastolic value due to the contraction of the
ventricle and this reduction in area ranges from 10 to 25%. This result agrees
with an animal study by Tsakiris et al. (1971) that looked at the difference in
the size, shape, and position of the mitral annulus at different stages in the
cardiac cycle. They noted an eccentric narrowing of the annulus during both
atrial and ventricular contractions that reduced the mitral valve area by 10 to
36% from its peak diastolic area. This reduction in the annular area during
systole is significant, resulting in a smaller orifice area for the larger leaflet
area to cover. The annulus not only changes size, but it also translates during
the cardiac cycle. The movement of the annulus toward the atrium has been
suggested to play a role in ventricular filling, possibly increasing the efficiency
of blood flow into the ventricle. During ventricular contraction, there is a
shortening of the left ventricular chamber along its longitudinal axis, and also
the mitral and tricuspid annuli move toward the apex.
The movement of the papillary muscles is also important in maintaining
proper mitral valve function. The papillary muscles play an important role
in keeping the mitral valve in position during ventricular contraction.
Abnormal strain on the papillary muscles could cause the chordae to rupture,
resulting in mitral regurgitation. It is necessary for the papillary muscles to
contract and shorten during systole to prevent mitral prolapse; therefore, the
distance between the apex of the heart to the mitral apparatus is important.
The distance from the papillary muscle tips to the annulus was measured
in normal individuals during systole and was found to remain constant. In
patients with mitral valve prolapse, however, this distance decreased, cor-
responding to a superior displacement of the papillary muscle toward the
annulus.
The normal function of the mitral valve requires a balanced interplay
between all of the components of the mitral apparatus as well as the inter-
action of the atrium and ventricle. The in vitro engineering studies of mitral
valve function have provided some insight into its mechanical properties
and dynamics. Further fundamental and detailed studies are needed to aid
surgeons in the repair of diseased mitral valves and in understanding the
changes in function due to mitral valve pathologies. In addition, these stud-
ies are crucial for improving the design of prosthetic valves that more closely
replicate native valve function. (See Chapter 8, Prosthetic Heart Valve Fluid
Dynamics.)
7328_C007.fm Page 274 Monday, September 18, 2006 2:06 PM
7328_S003.fm Page 275 Friday, July 28, 2006 4:36 PM

Part III

Cardiovascular Implants
and Biomechanical
Measurements
7328_S003.fm Page 276 Friday, July 28, 2006 4:36 PM
7328_C008.fm Page 277 Tuesday, October 10, 2006 11:04 AM

8
Prosthetic Heart Valve Fluid Dynamics

CONTENTS
8.1 Introduction ................................................................................................278
8.2 A Brief History of Heart Valve Prostheses ............................................279
8.2.1 Mechanical Valves.........................................................................279
8.2.2 Tissue Valves ..................................................................................284
8.2.3 Summary of Mechanical Versus Tissue Valves ........................288
8.2.4 Current Types of Prostheses........................................................288
8.3 Hemodynamic Assessment of Prosthetic Heart Valves ......................289
8.3.1 Pressure Drop ................................................................................290
8.3.2 Effective Orifice Area (EOA) .......................................................292
8.3.3 Regurgitation..................................................................................294
8.3.3.1 Turbulent Jet Theory ......................................................295
8.3.3.2 Proximal Flow Convergence.........................................295
8.3.4 Flow Patterns and Shear Stresses ...............................................296
8.3.4.1 Caged Ball Valve .............................................................298
8.3.4.2 Tilting Disc Valve............................................................299
8.3.4.3 Bileaflet Valve ..................................................................299
8.3.4.4 Porcine Valve ...................................................................302
8.3.4.5 Pericardial Valve .............................................................303
8.3.5 Leakage Flow through Heart Valve Prostheses .......................304
8.3.6 Cavitation and HITS.....................................................................306
8.4 In Vitro Studies of Coagulation Potential and Blood Damage ..........308
8.4.1 Implications for Thrombus Deposition .....................................308
8.5 Durability of Prosthetic Heart Valves ....................................................309
8.5.1 Wear.................................................................................................310
8.5.2 Fatigue.............................................................................................310
8.5.3 Mineralization................................................................................ 311
8.6 Current Trends in Valve Design..............................................................312
8.7 Conclusions.................................................................................................312
8.8 Problem .......................................................................................................313

277
7328_C008.fm Page 278 Tuesday, October 10, 2006 11:04 AM

278 Biofluid Mechanics: The Human Circulation

8.1 Introduction
Heart valve disease is one of the main afflictions of the cardiovascular system
and can be caused by rheumatic fever, ischemic heart disease, bacterial or
fungal infection, connective tissue disorders, trauma, and malignant carci-
noid. In advanced form, it leads to various disabilities and, ultimately, to
death. The valves that are most commonly affected are the mitral, aortic, and
tricuspid valves. Malfunction of the valves affects their hemodynamic (i.e.,
fluid dynamic) performance in two primary ways:

1. Stenosis — narrowing of the valve, which results in a greater resis-

tance to blood flow and, therefore, a greater pressure drop across
the valve.
2. Incompetence — failure of the valve to close completely, allowing
blood to flow in the reverse direction (also called “regurgitation”)
when the valve should be shut.

Both of these conditions reduce the efficiency of the heart and place addi-
tional stresses and strains upon it.
The decision to perform corrective surgery on the natural valve or to
replace it with a prosthetic valve is often made on the basis of an evaluation
of the functional impairment of the natural valve. A classification for such
an evaluation as proposed by the New York Heart Association is utilized
and surgery is usually limited to patients belonging to Classes III and IV.
The first clinical use of a cardiac valvular prosthesis took place in 1952
when Dr. Charles Hufnagel implanted an artificial caged-ball valve to treat
aortic insufficiency. The Plexiglas cage contained a ball occluder, and was
inserted into the descending aorta without the need for cardiopulmonary
bypass. It did not cure the underlying disease, but it did relieve regurgitation
from the lower two-thirds of the body.
The first implant of a replacement valve in the actual anatomic position
took place in 1960 along with the advent of cardiopulmonary bypass. Since
that time, the achievements in valve design and the success of artificial heart
valves as replacements have been remarkable with more than 50 different
cardiac valves having been introduced over the past 35 years. Unfortunately,
after many years of experience and success, not all problems associated
with heart valve prostheses have been eliminated. The most serious current
problems and complications are:

• Thrombosis and thromboembolism

• Anticoagulant-related hemorrhage
• Tissue overgrowth
• Infection
7328_C008.fm Page 279 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 279

• Paravalvular leaks due to healing defects

• Valve failure due to material fatigue or chemical change

While new valve designs continue to be developed, it is important to

understand their history in order to be able to improve their future effec-
tiveness.

8.2 A Brief History of Heart Valve Prostheses

This section on replacement valves highlights only a relatively small number
of the many designs that have been tried, but includes those that are either
the most commonly used today or those that have made notable contribu-
tions to the advancement of replacement heart valves.

8.2.1 Mechanical Valves

The use of the caged-ball valve in the descending aorta became obsolete with
the development in 1960 of what today is referred to as the Starr–Edwards
ball-and-cage valve. Similar in concept to the original Hufnagel valve, it
was designed to be inserted in place of the excised diseased natural valve.
This form of intracardiac valve replacement was used in the mitral position,
and for aortic and multiple valve replacements. Since 1962, the Starr–Edwards
valve has undergone many modifications in order to improve its perfor-
mance in terms of reduced hemolysis and thromboembolic complications.
However, the changes have focused on materials and techniques of construc-
tion and have not altered the overall concept of the valve design in any way
(Figure 8.1a).
Other manufacturers have produced variations of the ball and cage valve,
notably the Smeloff–Cutter valve and the Magovern prosthesis. In the
case of the former, the ball is slightly smaller than the orifice. A subcage
on the proximal side of the valve retains the ball in the closed position
with its equator in the plane of the sewing ring. A small clearance around
the ball ensures easy passage of the ball into the orifice. This clearance also
gives rise to a mild regurgitation, which is felt, but not proven, to be
beneficial in preventing thrombus formation. The Magovern valve, no
longer in use, consisted of a standard ball-and-cage format, which incor-
porated two rows of interlocking mechanical teeth around the orifice ring.
These teeth were used for inserting the valve and were activated by remov-
ing a special valve holder once the valve had been correctly located in the
prepared tissue annulus. Dislocation of this valve due to imperfect place-
ment in a calcified annulus was observed and, thus, the Magovern valve
is no longer viable.
7328_C008.fm Page 280 Tuesday, October 10, 2006 11:04 AM

280 Biofluid Mechanics: The Human Circulation

(a)

(b)

(c)
FIGURE 8.1
Photographs of various mechanical heart valves: (a) Starr–Edwards caged-ball valve, (b) caged-
disc heart valve, (c) Björk–Shiley tilting disc valve, (d) Medtronic Hall tilting disc valve, (e) St.
Jude Medical bileaflet valve, and (f) CarboMedics bileaflet valve. (From Yoganathan, A.P. (2000)
Cardiac valve prostheses, in The Biomedical Engineering Handbook, Bronzino, J.D. (Ed.), CRC
Press, Boca Raton, FL. With permission.)
7328_C008.fm Page 281 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 281

(d)

(e)

(f)

FIGURE 8.1 (CONTINUED).

7328_C008.fm Page 282 Tuesday, October 10, 2006 11:04 AM

282 Biofluid Mechanics: The Human Circulation

Due to the high profile design characteristics of the ball valves, low
profile caged-disc valves were developed in the mid-1960s, especially for
the mitral position. Examples of the caged-disc designs are the Kay–Shiley
and Beall prostheses, which were introduced in 1965 and 1967, respectively
(Figure 8.1b). These valves were used exclusively in the atrioventricular
position, but due to their inferior hemodynamic characteristics, caged-disc
valves are rarely used today.
It is interesting to note that even after 35 years of valve development; the
ball-and-cage format remains the valve of choice for some surgeons. How-
ever, it is no longer the most popular mechanical valve, having been super-
seded, to a large extent, by tilting-disc and bileaflet valve designs. These
designs overcome two major drawbacks of the ball valve — namely, high
profile configurations and excessive occluder-induced turbulence in the flow
through and distal to the valve.
The most significant developments in mechanical valve design
occurred in 1969 and 1970 with the introduction of the Björk–Shiley and
Lillehei–Kaster tilting-disc valves (Figure 8.1c). Both prostheses involve
the concept of a ‘‘free’’ floating disc, which, in the open position, tilts to
an angle that depends on the design of the disc-retaining struts. In the
original Björk–Shiley valve, the angle of the tilt was 60 degrees for the aortic
and 50 degrees for the mitral model. The Lillehei–Kaster valve has a
greater angle of tilt of 80 degrees, but, in the closed position, is preinclined
to the valve orifice plane by an angle of 18 degrees. In both cases the
closed valve configuration permits the occluder to fit into the circumfer-
ence of the inflow ring with virtually no overlap, thus reducing mechan-
ical damage to erythrocytes. A small amount of regurgitation backflow
induces a ‘‘washing out’’ effect of debris and platelets and theoretically
reduces the incidence of thromboemboli.
The main advantage of the tilting-disc valve is that in the open position
it acts like an aerofoil in the blood flow where induced flow disturbances
are substantially less than those obtained with a ball occluder. Although the
original Björk–Shiley valve employed a Delrin occluder, all present-day
tilting-disc valves use pyrolytic carbon for these components. It should also
be noted that the ‘‘free’’ floating disc can rotate during normal function, thus
preventing excessive contact wear from the retaining components on one
particular part of the disc. Various improvements to this form of mechanical
valve design have been developed, but have tended to concentrate on alter-
ations either to the disc geometry, as in the Björk-–Shiley convexo-concave
design, or to the disc retaining system, as with the Medtronic Hall and
Omniscience valve designs (Figure 8.1d).
The Medtronic Hall prosthesis was introduced in 1977 and is characterized
by a central, disc-control strut, with a mitral opening angle of 70 degrees
and an aortic opening angle of 75 degrees. An aperture in the flat, pyrolytic,
carbon-coated disc affixes it to the central guide strut. This strut not only
retains the disc, but also controls its opening angle and allows it to move
downstream 1.5 to 2.0 mm; this movement is termed disc ‘‘translation’’ and
7328_C008.fm Page 283 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 283

improves flow velocity between the orifice ring and the rim of the disc. The
ring and strut combination is machined from a single piece of titanium for
durability. All projections into the orifice (pivot points, guide struts, and disc
stop) are open-ended, streamlined, and are placed in the region of highest
velocity to prevent the retention of thrombi by valve components. The sew-
ing ring is of knitted Teflon®. The housing is rotatable within the sewing
ring for optimal orientation of the valve within the tissue annulus.
Most recent mechanical valve designs are based on the bileaflet, all-
pyrolytic carbon valve designed by St. Jude Medical, Inc. and introduced in
1978 (Figure 8.1e). This design incorporates two semicircular, hinged, pyro-
lytic carbon occluders (leaflets), which, in the open position, are intended to
provide minimal disturbance to flow. The leaflets pivot within grooves made
in the valve orifice housing. In the fully open position, the flat leaflets are
designed to open to an angle of 85 degrees.
The CarboMedics bileaflet prosthesis gained FDA approval in 1993
(Figure 8.1f). It is also made of Pyrolite® which is known for its durability
and thrombo-resistance. The valve has a recessed pivot design and is rotat-
able within the sewing ring. The two leaflets are semicircular, radiopaque,
and open to an angle of 78 degrees. A titanium-stiffening ring is used to
lessen the risk of leaflet dislodgment or impingement. The CarboMedics Top
Hat — an aortic version of this prosthesis — was designed for implantation
in the supraannular position. This allows for an increase in the size of the
implanted valve, thus improving forward flow hemodynamics of the valve.
Improving bulk flow hemodynamics of the bileaflet valve type has been
a recent design goal for St. Jude Medical as well. The St. Jude Medical HP
valve, introduced in 1993, features a decrease in thickness from the sewing
cuff design of the original St. Jude Medical valve. This modification allows
the placement of a proportionally larger housing within the cuff for a given
tissue annulus diameter, thus increasing the orifice area available for flow.
The St. Jude Medical Regent valve offers a further improvement in bulk
flow hemodynamics over the original St. Jude Medical valve. This improve-
ment is created by structurally reinforcing the housing of the St. Jude Medical
HP, allowing a decrease in its thickness, which further increases the internal
orifice area available for flow.
The Sorin Bicarbon valve, marketed by Sorin Biomedica of Italy, is popular
outside the U.S. The curvature of the leaflets of the Bicarbon valve reduces
pressure loss caused by the small central orifice of the original bileaflet valve
design. The pivots of this valve are made up of two spherical surfaces of
different radii of curvature. This design may reduce wear because the leaflet
projection rolls across the housing, instead of sliding against it, when the
valve opens and closes.
The ATS Open Pivot valve also features a change in the pivot design.
Traditional bileaflet valve pivots are made up of an extension of the leaflet
fitting into a recess in the housing where thrombus then has a tendency to
form. The Open Pivot design inverts this traditional pivot mechanism, expos-
ing the pivot to bulk forward flow. Exposure to high velocity flow may then
7328_C008.fm Page 284 Tuesday, October 10, 2006 11:04 AM

284 Biofluid Mechanics: The Human Circulation

remove the protein and cell deposition on the pivot surface that leads to
thrombus formation.
The On-X valve, marketed by the Medical Carbon Research Institute, is
the most recent design introduction to the U.S. Some of the innovative
features of this valve are a length-to-diameter ratio close to that of native
heart valves, a smoothed pivot recess that allows the leaflets to open at an
angle of 90 degrees relative to the valve housing, and a two-point landing
mechanism during valve closure. The first two features result in improved
bulk flow hemodynamics, while the latter may result in a more even distri-
bution of stresses during valve closure, a quiet closure event, and a reduced
cavitation potential.
The majority of valve replacements today utilize mechanical valves and
the most popular design is the bileaflet, with approximately 80% of the
mechanical valves implanted today being bileaflet prostheses.

8.2.2 Tissue Valves

Two major disadvantages with the use of mechanical valves are (1) the need
for life-long anticoagulation therapy and (2) the accompanying problems of
bleeding. Furthermore, the hemodynamic function of even the best designed
valve differs significantly from that of the natural healthy heart valve. An
obvious step forward in the development of heart valve substitutes was the
use of naturally occurring heart valves. This lead to the approach of using
antibiotic or cryo-treated human aortic valves (called homografts) removed
from cadavers for implantation in place of a patient’s own diseased valve.
The first of these homograft procedures was performed by Ross in 1962,
and the overall results so far have been satisfactory. This is, perhaps, not
surprising since the homograft replacement valve is optimum both from the
point of view of structure and function. In the open position, these valves
provide unobstructed central orifice flow and also have the ability to respond
to deformations induced by the surrounding anatomical structure. As a
result, such substitutes are less damaging to the blood when compared with
the rigid mechanical valve. The main problem with these cadaveric allografts
is that they are no longer living tissue and, therefore, lack the unique quality
of cellular regeneration typical of normal living systems, which makes them
more vulnerable to long-term damage. Furthermore, they are available in
very limited quantities.
An alternative approach is to transplant the patient’s own pulmonary
valve into the aortic position. This operation was first carried out by Ross
as well in 1967, and his study of 176 patients followed up over 13 years
showed that such transplants continued to be viable in their new position
with no apparent degeneration. This transplantation technique, however, is
limited in that it can only be applied to one site.
The next stage in development of tissue valve substitutes was the use of
autologous fascia lata (a wide layer of membrane that encases the thigh
7328_C008.fm Page 285 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 285

muscles) either as free or frame-mounted leaflets. The former approach for

aortic valve replacement was reported by Senning in 1967 and details of a
frame-mounted technique were published by Ionescu and Ross in 1969. The
approach combined the more natural leaflet format with a readily available
living autologous tissue. Although early results seemed encouraging, even
Senning expressed doubt about the value of this approach in 1971, and by
1978, fascia lata was no longer used in either of the above or any other form
of valve replacement. The failure of this technique was due to the inadequate
strength of this tissue when subjected to long-term cyclic stressing within
the heart.
Alternative forms of tissue leaflet valves were being developed in parallel
with the work on fascia lata valves. In these designs, however, more empha-
sis was placed on optimum performance characteristics than on the use of
living tissue. In all of these cases, the configuration involved a three-leaflet
format that was maintained by the use of a suitably designed mounting
frame. It was realized that naturally occurring animal tissues would be
rejected by the host if used in an untreated form. Consequently, a method
of chemical treatment had to be found that prevented this antigenic response,
but which did not degrade the mechanical strength of the tissue.
Formaldehyde has been used by histologists for many years to arrest
autolysis and to ‘‘fix’’ tissue in the state in which it was removed. It had been
used to preserve biological tissues in cardiac surgery but, unfortunately, was
found to produce shrinkage and also increase the ‘‘stiffness’’ of the resulting
material. For these reasons, formaldehyde was not considered ideal as a
method of tissue treatment. Glutaraldehyde is another histological fixative
that has been used especially for preserving fine detail for electron micros-
copy. It is also used as a tanning agent by the leather industry. In addition
to arresting autolysis, glutaraldehyde produces a more flexible material due
to increased collagen cross-linking. Glutaraldehyde has the additional ability
of reducing the antigenicity of xenograft tissue to a level at which it can be
implanted into the heart without significant immunological reaction.
Clinical experience with different tissue valve designs has increasingly
indicated time-dependent (i.e., 5 to 7 years) structural changes, such as
calcification and leaflet wear, leading to valve failure and subsequent replace-
ment. Valve leaflet calcification is more prevalent in children and young
adults and, therefore, tissue valves are rarely used in those cases at the
present time. While such problems have not been eliminated by fixation with
glutaraldehyde, new technologies are promising. The U.S. Food and Drug
Administration (FDA) currently approves alpha amino oleic acid and an
ethanol treatment for glutaraldehyde-fixed tissue for fixing tissue valves.
Alpha amino oleic acid reduces calcification by shielding negatively charged
aldehyde groups that could attract positively charged calcium ions with
amino groups on adjacent molecules. Treatment with ethanol extracts neg-
atively charged lipids from aldehyde-fixed tissue. The antimineralization
qualities of these fixatives have been shown in vivo, but long-term results of
their clinical performance has yet to be seen.
7328_C008.fm Page 286 Tuesday, October 10, 2006 11:04 AM

286 Biofluid Mechanics: The Human Circulation

In 1969, Kaiser et al. described a valve substitute using an explanted glut-

araldehyde-treated porcine aortic valve, which was mounted on a rigid support
frame. Following modification in which the frame was replaced by one having
a rigid base ring with flexible posts, this valve became commercially available
as the Hancock Porcine Xenograft in 1970 (Figure 8.2a). It remains one of the
three most popular valve substitutes of this type. One of the other two
valves is the Carpentier–Edwards Bioprosthesis (using a totally flexible
support frame), introduced commercially by Edwards Laboratories in 1976.

(a)

(b)

FIGURE 8.2
Photographs of biological leaflet valves: (a) a Hancock porcine bioprosthetic valve and (b) a
Carpentier–Edwards pericardial valve prosthesis. (From Yoganathan, A.P. (1995) Cardiac valve
prostheses, in The Biomedical Engineering Handbook, Bronzino, J.D. (Ed.), CRC Press, Boca Raton,
FL. With permission.)
7328_C008.fm Page 287 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 287

In 1977, production began of the Hancock Modified Orifice (MO) valve,

which is a refinement of the Hancock Standard valve. The Hancock MO
represents a composite nature wherein the right coronary leaflet containing
the muscle shelf is replaced by a noncoronary leaflet of the correct size from
another porcine valve. The valve is then mounted into a Dacron-covered
polypropylene stent. The Hancock II and Carpentier–Edwards supra-annular
porcine bioprostheses are second-generation bioprosthetic valve designs,
which were introduced in the early 1980s. The porcine tissue used in these
valves is initially fixed at 1.5 mmHg and then again at high pressure, which
is a fixation method designed to ensure good tissue geometry. Both valves
are treated with antimineralization solutions and have obtained FDA
approval.
In porcine prostheses, the use of the intact biologically formed valve makes
it unnecessary to manufacture individual valve cusps. While this has the
obvious advantage of reduced complexity of construction, it does require a
facility for harvesting an adequate quantity of valves so that an appropriate
range of valve sizes with suitable quality can be made available. This latter
problem was not an issue in the production of the three-leaflet calf pericar-
dium valve developed by Ionescu et al., since this valve involved the mold-
ing of fresh tissue to a tricuspid configuration around a support frame. As
the tissue is held in this position, it is treated with a glutaraldehyde solution.
The valve, marketed in 1976 as the Ionescu–Shiley Pericardial Xenograft,
was discontinued in the mid-1980s due to structural failure problems.
Early clinical results obtained with tissue valves indicated their superiority
to mechanical valves with respect to a lower incidence of thromboembolic
complications. For this reason the use of tissue valves increased significantly
during the late 1970s.
The Carpentier–Edwards pericardial valve consists of three pieces of
pericardium mounted completely within an Elgiloy® wire stent to reduce
potential abrasion between the Dacron-covered frame and the leaflets. The
pericardium is retained inside the stent by a Mylar® button rather than by
holding sutures (Figure 8.2b).
The Medtronic Mosaic valve, approved by the FDA in 2000, is the suc-
cessor to the Hancock II porcine bioprosthesis. This valve features a compli-
ant sewing ring that seals tightly in a valve annulus, a thin stent design, and
a zero pressure fixation technique. The first two of these features improve
bulk flow hemodynamics of the valve, while the latter prevents tissue from
being fixed in a prestressed state and may improve its longevity.
The most recent major change in heart valve design is the introduction of
the stentless bioprosthesis by Dr. Tirone David. Dr. David hypothesized that
fixing prosthesis leaflets to the aortic root would distribute mechanical
stresses into the aortic tissue instead of focusing them on the leaflet tissue
near the stents of traditional bioprostheses. In 1994, Dr. Yoganathan and
collaborators stated that the absence of the stent was also thought to improve
hemodynamics, as there is less obstruction to flow. Dr. David collaborated
with St. Jude Medical to produce the Toronto SPV aortic valve, which was
7328_C008.fm Page 288 Tuesday, October 10, 2006 11:04 AM

288 Biofluid Mechanics: The Human Circulation

approved by the FDA in 1997. Hemodynamic tests of this valve showed very
little resting pressure gradient and clinical results of resistance to wear are
very promising. Two other stentless aortic valve designs have since been
brought to the market: the Medtronic Freestyle and Edwards Prima valves.
The Prima valve allows full root or subcoronary implantation and removes
the coronary arteries from the root, while the Freestyle offers implantation
by a variety of surgical techniques and offers ligated coronary arteries.
Stentless bioprostheses are currently only approved for aortic valve
replacements in the U.S. Stentless mitral valves have been under develop-
ment for a number of years, as clinical trials with these valves were first
reported in 1992. Clinical and in vitro studies of stentless mitral valves have
shown that they have superior hemodynamics, but the questionable dura-
bility of these valves and complexity of the implantation technique have
delayed their approval for the U.S. market.

8.2.3 Summary of Mechanical Versus Tissue Valves

The clear advantage of mechanical valves is their long-term durability. Current
mechanical valves are manufactured from a variety of materials, such as pyro-
lytic carbon and titanium. Although structural failure of mechanical valves is
rare, it is usually catastrophic when it occurs. One major disadvantage of the
use of mechanical valves is the need for continuous, life-long anticoagulation
therapy to minimize the risk of thrombosis and thromboembolic complica-
tions. Unfortunately, the anticoagulation therapy may lead to bleeding prob-
lems; therefore, careful control of anticoagulation medication is essential for
the patient’s well-being and quality of life. Another concern is the hemody-
namic performance of the prosthesis. The function of even the best mechanical
valve designs differs significantly from that of normal heart valves.
The major advantage of tissue valves compared to mechanical valves is
that tissue valves have a lower incidence of thromboembolic complications.
Therefore, most patients receiving tissue valves are not required to take
anticoagulants long-term. The major disadvantages to tissue valves are mate-
rial fatigue and/or wear of valve leaflets, and calcification of valve leaflets,
especially in children and young adults. Valve deterioration, however, usu-
ally takes place slowly with these valves and, thus, patients can be monitored
by echocardiography and other noninvasive techniques.

8.2.4 Current Types of Prostheses

At present, over 180,000 prosthetic valves are implanted each year through-
out the world. The five most commonly used basic types of prostheses are:

• Caged ball
• Tilting disc
• Bileaflet
• Stented bioprosthesis
7328_C008.fm Page 289 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 289

Stentless bioprosthesis valve manufacturers continue to develop new

designs of mechanical and tissue valves. While the ideal heart valve prosthesis
does not yet exist and may never be realized, the characteristics of the ‘‘perfect’’
prostheses should be noted. Specifically, the ideal heart valve should:

• Be fully sterile at the time of implantation and be nontoxic.

• Be surgically convenient to insert at or near the normal location of
the heart.
• Conform to the heart structure rather than the heart structure con-
forming to the valve (i.e., the size and shape of the prosthesis should
not interfere with cardiac function).
• Show a minimum resistance to flow so as to prevent a significant
pressure drop across the valve.
• Only have a minimal reverse flow necessary for valve closure so as
to keep incompetence of the valve at a low level.
• Demonstrate resistance to mechanical and structural wear.
• Be long-lasting (25 years) and maintain its normal functional per-
formance (i.e., it must not deteriorate over time).
• Cause minimal trauma to blood elements and the endothelial tissue
of the cardiovascular structure surrounding the valve.
• Show a low probability for thromboembolic complications without
the use of anticoagulants.
• Be sufficiently quiet so as not to disturb the patient.
• Be radiographically visible.
• Have an acceptable cost.

In terms of specific issues related to heart valve design, the basic engineer-
ing considerations are:

• Flow dynamics
• Durability (structural mechanics and materials)
• Biological response to the prosthetic implant

8.3 Hemodynamic Assessment of Prosthetic Heart Valves

Many of the basic principles of fluid mechanics that we have discussed in
earlier chapters can be applied to the design and evaluation of prosthetic
heart valves. Considerations, which are of particular importance to this
analysis, are (1) transvalvular pressure drop, (2) effective valve orifice area,
(3) regurgitant flow rate, and (4) local flow patterns.
7328_C008.fm Page 290 Tuesday, October 10, 2006 11:04 AM

290 Biofluid Mechanics: The Human Circulation

8.3.1 Pressure Drop

Pressure drop and pressure recovery through prosthetic valves can be sig-
nificant factors affecting the pressure within the left ventricle. A larger pres-
sure drop across a prosthetic valve demands that a larger systolic pressure
be generated in the left ventricle to drive flow through the circulation. Since
it has been shown that the level left ventricular (LV) pressure produced is
the primary determinant of myocardial oxygen consumption, it is imperative
that the LV pressure be minimized when dealing with prosthetic valves. This
continues to be an area of active research as there have been numerous
publications detailing the effects of recovery on pressure drop measurements
for both natural and prosthetic heart valves. The pressure drop ( ∆p) across
a prosthetic valve is related to the energy losses caused by the valves’ pres-
ence. Pressure drops across natural valves can be measured with invasively
performed catheter techniques. However, it is both difficult and dangerous
to pass a catheter through a prosthetic valve. Fortunately, the Bernoulli
equation can provide a noninvasive estimate of the pressure drop across
prosthetic, and also, natural valves. Neglecting the viscous effects and apply-
ing Equation 1.53, from Chapter 1, section 1.5: The Bernoulli Equation, under
mean conditions, we obtain

1
p1 − p2 = ρ(V22 − V12 ) (1.53)
2

where Position 1 is proximal to the valve and Position 2 is at the orifice. If

the velocity upstream of the valve is much smaller than at the valve, V1 can
be neglected and Equation 1.55 is obtained,

p1 − p2 = 4V22 (1.55)

where the constant has units of (mmHg)/(m2/s2). Thus, if the velocity (m/s)
is measured with continuous-wave Doppler ultrasound velocimeter (see
Chapter 10, section 10.6: Ultrasound Doppler Velocimetry), the pressure drop
(mmHg) can be obtained noninvasively using Equation 1.55.
When measuring pressure distal to prosthetic valves, it is important to
note whether or not the recovered pressure is measured. This is especially
important when using continuous-wave Doppler ultrasound to evaluate
prosthetic heart valves. Because of its large sample volume size, continu-
ous-wave Doppler ultrasound measures the highest flow velocity, which
occurs at the vena contracta. Consequently, the Doppler-derived pressure
drops are always based on the pressure at the vena contracta, which is at a
site proximal to where pressure recovery occurs. Thus, one is likely to
overestimate the transvalvular pressure drop with this method because
pressure recovery is not considered. However, even though Doppler ultra-
sound may overestimate transvalvular pressure drops, they are extremely
7328_C008.fm Page 291 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 291

useful in patient diagnosis. A recent study has shown that Doppler-derived

pressure drops measured across small diameter prosthetic mechanical
valves (e.g., Medtronic-Hall and St. Jude) correlate just as strongly with
fluid mechanical energy losses as the recovered pressure drops measured
by catheterization.
Because of the larger separation region inherent in flow over bluff bodies,
configurations, such as the caged-disc and caged-ball, have notably large
pressure gradients. Tilting disc and bileaflet valve designs present a more
streamlined configuration to the flow and, while separation regions may
exist in these designs, the pressure gradients are typically smaller than for
the bluff shapes. Stented bioprostheses have relatively acceptable pressure
gradients for larger diameter valves because they more closely mimic
natural valve geometry and motion than mechanical valves, but the smaller
sizes (< 23 mm) generally have higher pressure gradients than their
mechanical valve counterparts, as shown in Figure 8.3. Stentless biopros-
theses, in contrast, have comparable or lower pressure drops than mechan-
ical prostheses. The clinical importance of pressure gradients in predicting
long-term performance is not clear. The fact that these gradients are a
manifestation of energy losses resulting from viscous-related phenomena
makes it intuitive that minimizing pressure gradients across an artificial
valve is highly desirable in order to reduce the workload of the pump
(i.e., left ventricle).

30
SJM regent
Medtronic hall
Pressure difference (mm Hg)

Hancock II
CE pericardial
20
Medtronic freestyle

10

0
200 300 400 200 300 400 200 300 400
Flow rate (cm3/s) Flow rate (cm3/s) Flow rate (cm3/s)
390 µm
(a) (b) (c)

FIGURE 8.3
Comparison of transvalvular pressure drop as a function of root mean square flow rate across
various mechanical and biological valve prostheses with tissue annulus diameter of (a) 19/20 mm,
(b) 25 mm, and (c) 27 mm. (Note that the pressure drop decreases with larger tissue annulus
diameters due to decrease in resistance of flow across the valve.)
7328_C008.fm Page 292 Tuesday, October 10, 2006 11:04 AM

292 Biofluid Mechanics: The Human Circulation

8.3.2 Effective Orifice Area (EOA)

Cardiologists have used a variety of methods to determine the forward flow
cross-sectional area of a valve. Because of acoustic anomalies caused by pros-
thetic valves and the limited spatial resolution of echocardiography, it is often
impossible to measure the valve area directly. However, applying a control
volume analysis to a prosthetic valve in the aortic position allows for estima-
tion of the mean area over which fluid flows, as shown in Equation 8.1.

∫
V1 (t)A1
A2 = dt (8.1)
V2 (t)
T

A2 is the area of the vena contracta or the effective valve area and A1 is the
area at the upstream face of the control volume, measured by echocardio-
graphy. The integral is taken with respect to time over the systolic flow
period, V1 is the velocity upstream of the aortic valve, and V2 is the velocity
at the vena contracta. Inherent to Equation 8.1 is the assumption that the
spatial velocity profile is uniform over the entire orifice area, but because
the velocity profile is nonuniform in mechanical prostheses, it is difficult to
apply this equation accurately in that case.
By extending the control volume from the left ventricular outflow tract to
encompass the entire left ventricle, it is possible to use this technique to
determine the mean mitral valve area. The time-velocity integral of both the
mitral and the left ventricular outflow tract positions must be calculated over
the entire cardiac cycle. It is only necessary to integrate the velocity at the
mitral valve during diastole and the velocity of the aortic valve during systole
because these are the only times in which there is flow through the valves.
If regurgitation is not present in either valve, Equation 8.1 can be applied and
the effective area of the mitral valve can be obtained. When aortic regurgita-
tion is present, the velocity and area at the right ventricular outflow tract can
be used instead for calculation of effective mitral orifice area.
Another method of determining valve area can be derived using the
Bernoulli equation and Conservation of Mass. A contraction coefficient (Cd)
can be defined as the area of the vena contracta divided by the area of the
valve orifice (Ao). The continuity equation for flow through an orifice, includ-
ing the contraction coefficient appears in Equation 8.2, where Q is the flow
rate and v is the velocity at the vena contracta

Q = A0VCd (8.2)

Solving for V2 in Equation 1.54, substituting into Equation 8.2 and rearrang-
ing, yields an equation that can be used to determine the effective orifice
area, EOA, of a cardiac valve

Q rms
EOA(cm 2 ) = (8.3)
51.6 ∆p
7328_C008.fm Page 293 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 293

TABLE 8.1
Test Data for 27 mm Tissue Annulus Diameter-Size Valves with
Measurements Obtained Using an In Vitro Pulse Duplicator with a Heart
Rate of 70 bpm and a Cardiac Output of 5.0 l/min
Peak Turb.
SS+
Reg. Vol. (dynes/
Valve Type EOA*(cm2) PI (cm3/beat) cm2) ∆P mmHg
Caged ball 1.75 0.30 5.5 1850 13.6
Tilting disc# 3.49 0.61 9.4 1800 3.4
Bileaflet# 3.92 0.68 9.15 1940 2.7

Porcine (stented)# 2.30 0.40 <2 2980 7.9

Pericardial (stented)# 3.70 0.64 <3 1000 3.1
Stentless BHV 3.75 0.65 <4 n/a 3.0
* Effective orifice area, EOA, computed by the application of Gorlin’s Equation
(Equation 8.3).
+ Turbulent stresses were measured at variable distances from the valve seat.

# Values reported are mean values from several valve models of the same type.

Source: Values compiled from Yoganathan, A. (2000) Cardiac valve prostheses,

in Biomedical Engineering Handbook, Bronzino, J.D. (Ed.), CRC Press and IEEE
Press, Boca Raton, FL, pp. 127.1–127.23. With permission.

where Qrms is the root mean square systolic/diastolic flow rate (cm3/s),
and ∆p is the mean systolic/diastolic pressure drop (mmHg). The nondi-
mensional constant, Cd, was determined by using the density of blood ( ρ =
1050 kg/m3) and converting the units to those convenient for biomedical
use. The effective area will be in cm2 if the root mean square systolic or
diastolic flow rate (Qrms) is in cm3/s and the mean systolic or diastolic
pressure drop ( ∆p) is measured in mmHg. Similar relationships were
derived earlier in Chapter 5 (Equation 5.4 and Equation 5.5) through the
application of Bernoulli’s equation where the magnitude of the constants
employed depends upon the discharge coefficient for a particular valve.
Table 8.1 lists EOAs obtained in vitro, for mechanical and tissue valve
designs in clinical use today. These results illustrate the fact that the newer
mechanical valve designs have better pressure gradient characteristics than
porcine bioprostheses in current clinical use.
Another method that has been developed to estimate the orifice area
involves measuring the pressure half-time of an orifice. The pressure half-time
( p t ) is the time required for left ventricular pressure to fall to one-half of
2
its peak value. An empirical relation has been established between the area
of the mitral valve ( AreaMV ) and p t as
2

220
AreaMV = pt (8.4)
2
7328_C008.fm Page 294 Tuesday, October 10, 2006 11:04 AM

294 Biofluid Mechanics: The Human Circulation

Equation 8.4 is based on observations that changes in the mitral pressure

drop with time are constant for a particular orifice area. The constant in the
equation, 220 (cm2 ⋅ s), is empirically determined. Unfortunately, this equa-
tion has been shown to be dependent on a number of factors other than the
valve area, including the severity of aortic regurgitation and ventricular wall
properties. Furthermore, this equation is ineffective for prosthetic heart
valves, thus limiting its application.

8.3.3 Regurgitation
Because mechanical prosthetic valves are fairly rigid, they are unable to form
tight seals between the occluder and supporting ring when closed. Conse-
quently, regurgitant jets are present in the resultant gaps in the prosthetic
valves under normal conditions, although the amount of regurgitation is
usually small. Normal regurgitant flow is characterized by having a closing
volume during valve closure and leakage after closure. These parameters
are illustrated in Figure 8.4. Regurgitant flow is often characterized by the
regurgitant volume. The regurgitant volume is the total volume of fluid
through the valve per beat due to the retrograde flow and is equal to the
sum of the closing volume and the leakage volume. The closing volume is
the volume of fluid flowing retrograde through the valve during valve clo-
sure. Any fluid volume accumulation after valve closure is due to leakage
and is referred to as the leakage volume.
Closing regurgitation is related to the valve shape and closing dynamics,
and the percentage of stroke volume that succumbs to this effect is in the
range of 2.0 to 7.5% for mechanical valves and is typically less for tissue
valves, ranging from 0.1 to 1.5%. Leakage, which depends upon how well
the orifices are ‘‘sealed’’ upon closure, has a reported incidence of 0 to 10%
in mechanical valves and 0.2 to 3% in bioprosthetic valves. The overall

Forward flow

Closing volume Leakage volume

FIGURE 8.4
A schematic illustrating the closing and leakage volume across a valve prosthesis. (From
Yoganathan, A.P. (1995) Cardiac valve prostheses, in The Biomedical Engineering Handbook,
Bronzino, J.D. (Ed.), CRC Press, Boca Raton, FL. With permission.)
7328_C008.fm Page 295 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 295

tendency is for regurgitation to be less for the bioprosthetic heart valves than
for mechanical valve designs (see Table 8.1). To distinguish normal from
abnormal valve function, it is important to differentiate between normal
regurgitant volume and additional regurgitation due to disease. Fluid
mechanical analysis of this problem has provided two different techniques —
Turbulent Jet Theory and Proximal Flow Convergence — that at least
partially fulfill this requirement.

8.3.3.1 Turbulent Jet Theory

Using free turbulent jet theory, relations were derived for the regurgitant
flow rate based solely on Doppler ultrasound measurements. Turbulent jets
have a number of unique characteristics. Upon entering a chamber, they
spread radially, pulling or entraining stationary fluid with them. Initially
though, the jet has a core of fluid that is not affected by the stationary fluid.
This potential core has the same velocity as the jet at the orifice and persists
for a few orifice diameters downstream. Once the core vanishes, the jet
reaches a state that is amenable to theoretical analysis. However, it is impor-
tant to recognize the assumptions inherent in this theoretical analysis. One
of these is that unsteady flow effects are neglected, and the analysis assumes
that the jet enters a stationary fluid and does not impinge on solid bound-
aries. In the heart, regurgitant jets are unsteady, often impinge on the walls
of the heart, and always encounter incoming forward flow. Additional work
has been performed that treats jets in confined and impinging geometries;
these geometries more closely resemble the chambers of the left heart. A
dimensional analysis model of valvular regurgitation has been reported
based on the centerline velocity decay of the jet and the width and length
of the receiving chamber and this equation has been shown to be accurate
for a number of in vitro geometries and conditions.

8.3.3.2 Proximal Flow Convergence

The principle of Conservation of Mass was also applied in the region prox-
imal to the valvular orifice to measure regurgitant volume. When fluid enters
a regurgitant orifice, it must accelerate to reach a peak velocity at the throat
of the orifice. If the orifice is circular, this acceleration region should be
axisymmetric about the center of the orifice. Thus, upstream of the regur-
gitant orifice, a series of concentric isovelocity contours can be defined within
the flow field, which is hemispherical in shape. This physical argument is
the basis for the Proximal Flow Convergence Method of quantifying regur-
gitant flow rate. If a control volume is constructed to coincide with a hemi-
spherical contour and the regurgitant orifice, the same amount of fluid that
enters the volume from an isovelocity contour will exit the control volume
through the regurgitant orifice. The control volume statement is represented
mathematically in Equation 8.5

Q0 = ( 4 πr 2 )Vr (8.5)
7328_C008.fm Page 296 Tuesday, October 10, 2006 11:04 AM

296 Biofluid Mechanics: The Human Circulation

Q0 is the flow rate at the regurgitant orifice and the term within parentheses
is the surface area of the hemispherical surface. Vr is the velocity measured
with the color Doppler echocardiography and r is the radial distance at which
the Velocity is measured.
Because of its simplicity and ease of application, this technique has
received a great deal of attention, especially for mitral regurgitation. The
effects of regurgitant orifice motion, orifice geometry variation, and ultra-
sound machine settings have all been addressed with both in vitro and in
vivo investigations. In addition, its application to prosthetic valve regurgita-
tion has also been considered. Recent studies have found that isovelocity
contours from regurgitant orifices can be described better by a hemi-elliptical
shape than a hemispherical shape, yielding more accurate estimations of
regurgitant flow rate in vitro. Unfortunately, rigorous in vivo validation of
the proximal flow convergence technique is difficult.

8.3.4 Flow Patterns and Shear Stresses

The potential of nonphysiological blood flow patterns to promote detri-
mental changes in the cardiovascular system has long been recognized.
The German pathologist R.L.K. Virchow, in 1856, recognized that altered
blood flow was one of the three general causes of thrombosis. Nonphys-
iological blood flow can initiate thrombus formation by imposing forces
on blood elements or by changing their frequency of contact. Forces
imposed by nonphysiological flow on blood cells can damage their struc-
ture or change their conformation, acting as a signal to promote coagula-
tion. Blood flow patterns dictate how often cells and proteins contact the
vessel walls and each other. A localized increase in this amount of contact
can promote thrombus formation.
Therefore, recirculation zones and stagnation points are flow patterns that
can be detrimental to the healthy circulation. Recirculation zones are vortical
regions in which activated platelets and coagulation-inducing proteins can
concentrate. Since these flow patterns often occur just distal to regions of
high shear, they receive more of the coagulating agents than other areas of
the flow field. All platelets within a fluid path line leading to a stagnation
point are directed toward a vessel wall and, thus, the stagnation point is a
likely area for platelet adhesion, also being dependent on the magnitude of
flow normal to the wall. Platelet aggregation in recirculation zone structures
and adhesion to a vessel wall has been linked with platelet collision fre-
quency, which is influenced by the speed of the flow.
Peak arterial viscous shear stresses within the normal circulation are
between 50 and 200 dynes/cm2. In valvular flow fields, the shear stresses
platelets experience can be an order of magnitude higher than normal phys-
iological values, increasing the likelihood of cell damage. Several studies
have been carried out to determine the effects of viscous stresses on blood
cells. Results from these studies have shown that (1) the threshold for cell
7328_C008.fm Page 297 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 297

damage by viscous stress occurs at a lower magnitude in the presence of

foreign surfaces, (2) cell damage or activation depends on the time in which
the cells are exposed to a viscous stress as well as the stress magnitude, and
(3) platelets are far more sensitive to a viscous stress stimulus than red cells.
These studies suggest that the threshold for procoagulant platelet activity
initiated by viscous shear stresses during flow across heart valve prostheses
lies between 500 and 5000 dynes/cm2. The effects of the presence of a pros-
thetic valve surface on this threshold, however, are currently unknown.
Turbulence rarely occurs in the circulation of a healthy human, but when
it does, it increases the likelihood of cell damage. Information about the
effects of turbulent shear stresses on blood cells is not as abundant as infor-
mation about the effects of viscous shear stresses. The turbulent shear stress
threshold for hemolysis during a 1 ms exposure time has recently been
measured at 8000 dynes/cm2, but sublethal damage to red cells has been
observed at turbulent shear stress levels of 500 dynes/cm2. An earlier study
showed a linear correlation between the turbulent intensity within an arterial
shunt and the weight of thrombus formed in this shunt, demonstrating that
platelets are affected by turbulent shear stresses. However, thus far, no
attempts have been made to characterize a turbulent shear stress threshold
for procoagulant platelet activity, despite the fact that this threshold is likely
lower than the threshold for red cell damage. The turbulent shear stress
threshold for procoagulant platelet activity is currently thought to be similar
to the threshold for platelet activity induced by viscous stresses. These two
forms of stresses could act differently, however. Turbulent stresses, unlike
viscous stresses, act on a length scale much larger than the molecular scale.
If the length scale over which a turbulent stress acts is much larger than the
diameter of a blood cell, the cell may not experience the turbulent stress.
Thrombosis and embolism, tissue overgrowth, hemolysis, energy loss, and
damage to endothelium adjacent to the heart valve are directly related to
the velocity and turbulence fields created by various valve designs, and have
been addressed in detail during the past 3 decades by investigators studying
cardiovascular fluid mechanics.
In order to illustrate the abnormal flow fields and elevated levels of
turbulent shear stresses created by prosthetic valves, in vitro forward flow
measurements conducted on valve designs in current clinical use in the
U.S. (i.e., FDA approved) are presented for typical valves in each of five
categories — caged ball, tilting disc, bileaflet, porcine, and pericardial valves.
These measurements were obtained on size 27 mm aortic valves in a pulse
duplicator that simulated physiological pulsatile flow using the laser Doppler
anemometry technique described in Chapter 10, section 10.7. The details of
the pulse duplicators used for such measurements in various laboratories
can be found in publications describing such measurements and comparing
various mechanical and biological heart valve prostheses. Figures of the in
vitro flow field studies are presented as schematic diagrams, and represent
velocity and turbulence profiles obtained at peak systole, with a cardiac
output of 5.0 l/min and a heart rate of 70 bpm. All downstream distances
7328_C008.fm Page 298 Tuesday, October 10, 2006 11:04 AM

298 Biofluid Mechanics: The Human Circulation

are measured from the valve sewing ring. Table 8.1 lists peak turbulent shear
stress levels measured downstream of the common 27-mm valve prostheses.

8.3.4.1 Caged Ball Valve

Flow emerging from the valve forms a circumferential jet that separates from
the ball, hits the wall of the flow chamber, and then flows along the wall. It
has very high velocities in the annular region. The maximum velocity, mea-
sured 12 mm downstream of the valve, is 220 cm/s at peak systole. The peak
systolic velocity, measured 30 mm downstream of the valve, is 180 cm/s, as
shown in Figure 8.5. High velocity gradients are observed at the edges of
the jet and the maximum velocity gradient (1700 cm s−1/cm) is detected in
the annular region adjacent to the surface of the ball during peak systole. A
large velocity defect is also seen in the central part of the flow chamber as
a wake develops distal to the ball. A region of low velocity, reverse flow is
observed at peak systole and during the deceleration phase immediately
distal to the apex of the cage, which has a diameter of about 8 mm. A
maximum reverse velocity of –25 cm/s occurs at peak systole 30 mm down-
stream of the valve. The intensity of the reverse flow during the deceleration
phase is not as high as that observed at peak systole and no reverse flow is
observed during the acceleration phase. The velocity in the central part of
the flow channel, however, is low.
High turbulent shear stresses are present at the edges of the jet. The
maximum turbulent shear stress measured is 1850 dynes/cm2, which occurs

200.0 2000.0

100.0 1000.0

0.0 0.0

(a) (b)

FIGURE 8.5
(a) Velocity and (b) turbulent shear stress profiles 30 mm downstream of a caged ball valve at
peak systole (from Yoganathan, A.P. Cardiac Valve Prostheses, in The Biomedical Engineering
Handbook, Bronzino, J.D. (Ed.), 1995. With permission from CRC Press, Boca Raton, Florida).
7328_C008.fm Page 299 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 299

at the location of the highest velocity gradient. The intensity of turbulence

during peak systole does not decay very rapidly downstream of the valve
and elevated turbulent shear stresses occur during most of systole. Turbulent
shear stresses as high as 3500 dynes/cm2 are estimated in the annular region
between the flow channel wall and the ball.

8.3.4.2 Tilting Disc Valve

High velocity, jet-like flows are observed from both the major and minor
orifice outflow regions. The orientation of the jets with respect to the axial
direction changes as the valve opens and closes. The major orifice jet is larger
than the minor orifice jet and has a slightly higher velocity. The peak veloc-
ities measured 7 mm downstream of the valve are 210 cm/s and 200 cm/s
in the major and the minor orifice regions, respectively (Figure 8.6). A region
of reverse flow is observed adjacent to the wall in the minor orifice region
at peak systole, which extends 2 mm from the wall with a maximum reverse
velocity of –25 cm/s. The size of this region increases to 8 mm from the wall
during the deceleration phase. A small region of flow separation is observed
adjacent to the wall in the major orifice region. In the minor orifice region,
a profound velocity defect is observed between 7 and 11 mm distal to the
minor orifice strut. Furthermore, the region adjacent to the wall immediately
downstream of the minor orifice is stagnant during the acceleration and
deceleration phases and has very low velocities (<15 cm/s) during peak
systole.
In the major orifice region, high turbulent shear stresses are confined to
narrow regions at the edges of the major orifice jet. The peak turbulent
shear stresses measured at peak systole are 1200 and 1500 dynes/cm2, at 7
and 13 mm downstream of the valve, respectively. During the acceleration
and deceleration phases, the turbulent shear stresses are relatively low. High
turbulent shear stresses are more dispersed in the minor orifice than those
in the major orifice region. The turbulent shear stress profiles across the
major and minor orifices 15 mm downstream of the valve show a maximum
turbulent shear stress of 1450 dynes/cm2 at the lower edge of the minor
orifice jet.

8.3.4.3 Bileaflet Valve

The bileaflet valve has two semicircular leaflets, which divide the area avail-
able for forward flow into three regions — two lateral orifices and a central
orifice. The major part of the forward flow emerges from the two lateral
orifices. The measurements along the centerline plane 8 mm downstream of
the valve show a maximum velocity of 220 cm/s and 200 cm/s for the lateral
and central orifice jets, respectively, at peak systole. The velocities of the jets
remain about the same as the flow travels from 8 to 13 mm downstream
(Figure 8.7). The velocity profiles show two defects that correspond to the
location of the two leaflets. The velocity measurements indicate that the flow
7328_C008.fm Page 300 Tuesday, October 10, 2006 11:04 AM

300 Biofluid Mechanics: The Human Circulation

200.0
200.0 200.0

100.0 100.0
100.0

0.0
0.0
0.0

(a) (b)

2000.0
2000.0 2000.0

1000.0 1000.0
1000.0

0.0
0.0 0.0

(c) (d)

FIGURE 8.6
(a) Velocity profile 15 mm downstream, (b) velocity profiles 13 mm downstream, (c) turbulent
shear stress profile 15 mm downstream, and (d) turbulent shear stress profiles 13 mm down-
stream of a tilting disc valve at peak systole (from Yoganathan, A.P. Cardiac Valve Prostheses,
in The Biomedical Engineering Handbook, Bronzino, J.D. (Ed.), 1995. With permission from CRC
Press, Boca Raton, Florida).

is more evenly distributed across the flow chamber during the deceleration
phase than during the acceleration phase. Regions of flow separation are
observed around the jets adjacent to the flow channel wall as the flow
separates from the orifice ring. The measurements across the central orifice
show that the maximum velocity in the central orifice is 220 cm/s. Small
regions of low-velocity reverse flow are observed adjacent to the pivot/hinge
mechanism of the valve. More flow emerges from the central orifice during
the deceleration phase than during the acceleration phase.
High turbulent shear stresses occur at locations of high velocity gradients
and at locations immediately distal to the valve leaflets. The flow along the
7328_C008.fm Page 301 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 301

200.0

200.0

100.0

100.0

0.0

0.0

(a) (b)

2000.0

2000.0

1000.0

1000.0

0.0

0.0

(c) (d)

FIGURE 8.7
Velocity profile 13 mm downstream of a bileaflet valve; (a) across the leaflets, (b) across the
central orifice and turbulent shear stress profiles 13 mm downstream of a bileaflet valve, (c)
across the leaflets, and (d) across the central orifice at peak systole (from Yoganathan, A.P.
Cardiac Valve Prostheses, in The Biomedical Engineering Handbook, Bronzino, J.D. (Ed.), 1995.
With permission from CRC Press, Boca Raton, Florida).

centerline plane becomes more disturbed as the flow travels from 8 to 13 mm

downstream of the valve. The peak turbulent shear stresses measured along
the centerline plane at peak systole are 1150 and 1500 dynes/cm2 at 8 and
13 mm downstream of the valve, respectively. The profiles across the central
orifice show that the flow is very disturbed in this region. The maximum
7328_C008.fm Page 302 Tuesday, October 10, 2006 11:04 AM

302 Biofluid Mechanics: The Human Circulation

400.0 5000.0

200.0 2500.0

0.0 0.0

(a) (b)

FIGURE 8.8
(a) Velocity profile and (b) turbulent shear stress profiles 15 mm downstream of a porcine
bioprosthesis at peak systole (from Yoganathan, A.P. Cardiac Valve Prostheses, in The Biomedical
Engineering Handbook, Bronzino, J.D. (Ed.), 1995. With permission from CRC Press, Boca Raton,
Florida).

turbulent shear stress measured in the central orifice (1700 dynes/cm2)

occurs at peak systole. Since these high turbulent shear stresses across the
central orifice are measured 11 mm downstream, it is probable that even
higher turbulent shear stresses occur closer to the valve.

8.3.4.4 Porcine Valve

The velocity profiles for an atypical porcine valve taken 10 mm downstream
of the valve, along the centerline plane, show that the peak velocity of the
jet-like flow emerging from the valve is as high as 220 cm/s at peak systole
(Figure 8.8). The peak velocities measured during the acceleration and decel-
eration phases are about the same — 175 and 170 cm/s, respectively. How-
ever, the flow is much more evenly distributed during the acceleration phase
than during the deceleration phase. No regions of flow separation are
observed throughout the systolic period in this plane of measurement. How-
ever, the annular region between the outflow surfaces of the leaflets and the
flow chamber wall is relatively stagnant throughout systole. The velocity of
the jet increases to about 370 cm/s at peak systole as the flow travels from
10 to 15 mm downstream of the valve. This indicates that the flow tends to
accelerate toward the center of the flow channel. High turbulent shear
stresses occur at the edge of the jet. The maximum turbulent shear stress is
2750 dynes/cm2 and is measured 10 mm downstream of the valve along the
centerline plane at peak systole. The turbulent shear stresses at the edge of
7328_C008.fm Page 303 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 303

the jet increase as the flow travels from 10 to 15 mm downstream of the

valve. The mean and maximum turbulent shear stress measured at peak
systole increase to 2000 and 4500 dynes/cm2, respectively.

8.3.4.5 Pericardial Valve

The velocity profiles obtained distal to a typical pericardial valve along the
centerline plane 17 mm downstream of the valve at peak systole show a
maximum velocity of 180 cm/s. The maximum velocities measured during
the acceleration and deceleration phases are 120 and 80 cm/s, respectively.
A region of flow separation, which extends about 6 mm from the wall, is
observed at peak systole and during the deceleration phase. This region is
relatively stagnant during the acceleration phase. The maximum velocity of
the jet at peak systole does not change as the flow field travels from 17 to
33 mm downstream of the valve (Figure 8.9). However, the size of the region
of flow separation decreases and extends only 1 mm from the wall.
Turbulent shear stress measurements taken along the centerline plane 17 mm
downstream of the valve show that, during the deceleration phase, elevated
turbulent shear stresses are spread out over a wide region (with a maximum
value of 1200 dynes/cm2). At peak systole, the high turbulent shear stresses
are confined to a narrow region, with a maximum value of 850 dynes/cm2.
The intensity of turbulence at peak systole increases as the flow travels from
17 to 33 mm downstream of the valve.

200.0 2000.0

100.0 1000.0

0.0 0.0

(a) (b)

FIGURE 8.9
a) Velocity profile and (b) turbulent shear stress profiles 17 mm downstream of a pericardial
tissue valve prosthesis at peak systole (from Yoganathan, A.P. Cardiac Valve Prostheses, in The
Biomedical Engineering Handbook, Bronzino, J.D. (Ed.) 1995. With permission from CRC Press,
Boca Raton, Florida).
7328_C008.fm Page 304 Tuesday, October 10, 2006 11:04 AM

304 Biofluid Mechanics: The Human Circulation

8.3.5 Leakage Flow through Heart Valve Prostheses

Retrograde flow in mechanical prostheses is composed of the squeeze flow,
which occurs during valve closure, and the leakage flow, which occurs after
valve closure. The former, which can be observed in bileaflet and tilting disc
prosthetic designs, is a result of the rising fluid pressure across the valve
and the closing occluder. These conditions “squeeze” blood through the
interstitial gap between the closing occluder and the valve housing. The
latter flow is comprised of the flow through the hinge region, which is found
in the bileaflet designs only, and the leakage flow that can be observed
through the perimeter of the closed occluder. Hinge flow is rapidly con-
stricted from a large chamber to the small area of the hinge region, and is
driven by the elevated transvalvular pressure after valve closure. These
retrograde flow phenomena are characterized by elevated fluid velocities
(2 to 3 m/s) and high turbulent shear stresses (150 to 800 N/m2) through
gaps an average of 0.5 mm in width.
“Squeeze” flow is closely related to cavitation. Numerical and in vitro
experiments have shown that the pattern of “squeeze” flow is related to the
moving boundaries, i.e., the approaching leaflets and the valve housing, and
to the velocity of the closing leaflets. Typical closing velocities at the occluder
tips range from 1 to 3 m/s. With a gap width of approximately 50 µm, the
peak squeeze-flow velocity has been estimated as high as 30 to 40 m/s with
corresponding negative pressures of approximately 15 MPa. With a gap
width of 48 µm and a high occluder tip velocity of 4 m/s, a peak squeeze
flow of approximately 30 m/s has been reported. It has also been demon-
strated that a peak squeeze flow velocity as high as 150 m/s can exist with
gap width of only 0.3 µm. On the other hand, a peak squeeze flow of less
than 10 m/s has been measured through a gap width of approximately
100 µm. The broad range observed in the measurement of the gap width can
be attributed to the difficulties of determining the exact location of the
moving occluder with respect to the valve housing, different valve designs,
and their manufacturing tolerances.
It has only recently been recognized that hinge flow is potentially dam-
aging to the blood. Subsequent studies have shown that hinge geometry
design is capable of influencing retrograde flow characteristics. With the
myriad bileaflet valve designs currently available, it is important to compare
the merits of various hinge designs to assess their potential for thrombogen-
esis (Figure 8.10 and Figure 8.11). A certain degree of regurgitant flow is
usually incorporated into the design of a bileaflet mechanical heart valve
(MHV) to ensure the washout of critical areas of the valve, such as hinges
and regions between the leaflet edges and housing. This washout is intended
to prevent flow stasis that may be created in the vicinity of these regions
and, therefore, minimize, if not eliminate, the formation of thrombosis. A
properly conducted hinge flow study should compare the flow in both the
leakage and forward flow phases to assess stresses on the blood cells and
the degree of washout in that particular valve design.
7328_C008.fm Page 305 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 305

Hinge

Leaflet

Thumbnail

Housing

FIGURE 8.10
Schematic of the hinge for one bileaflet valve model. (From Leo et al. (2002) Microflow fields
in the hinge region of the CarboMedics bileaflet mechanical heart valve design, in J. Thorac.
Cardiovasc. Surg., 124: 561–574. With permission.)

Leakage can also exit the narrow spaces between the occluder and housing
just upstream of mechanical valves in the form of high speed jets. The pattern
of these leakage jets is particular to the valve design. Figure 8.12 shows the
results of a laser Doppler velocimetry experiment designed to measure
velocity and turbulence within half the leakage jets of a bileaflet valve in
vitro under aortic flow conditions. The lengths and directions of the arrows
in this figure are representative of the jet velocity magnitude and direction.
The valve diagram in the upper right hand corner of this figure shows the
relative positions of the leakage jets with respect to the valve. Two of these
jets exit the valve from pivots, while the remaining jet appears to exit from
the intersection of the two leaflets and the valve housing. The largest velocity
magnitude measured in these jets was 0.6 m/s, and the largest turbulent
shear stress magnitude measured was 28.9 N/m2.
A glance at Figure 8.12 shows that the jets exiting from the valve are
asymmetric, though the valve itself is symmetric. This is likely due to the

Leaflets

St Jude CarboMedics Medtronic Parallel (MP) Leaflet

(SJM) (CMI)

FIGURE 8.11
A comparison of hinge geometry for two commercially available bileaflet valve designs. (From
Leo et al. (2002) Microflow fields in the hinge region of the CarboMedics bileaflet mechanical
heart valve design, in J. Thorac. Cardiovasc. Surg., 124: 561–574. With permission.)
7328_C008.fm Page 306 Tuesday, October 10, 2006 11:04 AM

306 Biofluid Mechanics: The Human Circulation

1
2

1 2

3 Velocity
magnitude (m/s)
0.6

0.0

FIGURE 8.12
Laser Doppler measurement of a leakage jet of a bileaflet valve. (From Travis, B.R. et al. (2002)
An analysis of turbulent shear stress in leakage flow through bileaflet mechanical prostheses,
in J. Biomech Eng., 2002, 124(2):155–165.)

small differences in leakage gap widths created by manufacturing tolerances.

These tolerances may have a significant effect on the character of the leakage
flow. Velocities as high as 3.7 m/s and Reynolds shear stresses as high as
1000 N/m2 have been reported within mechanical valve leakage jets. If
Reynolds shear stresses of these magnitudes occur, the size of the smallest
turbulent length scale is about 7 µ m. Since this is nearly the same size as
the blood cells, it is possible that the turbulence would dissipate its energy
directly on them.

8.3.6 Cavitation and HITS

The sudden stop and rebound of a mechanical valve occluder creates high
magnitude negative pressure transients. If the magnitudes of these transients
are strong enough, local pressures in the vicinity of the valve may drop
below the vapor pressure of blood, which may result in the formation of
small vapor bubbles. The implosions of such cavitation bubbles generate
small, high velocity jets and large oscillations in pressure. Severe cavitation
phenomena in industry have been shown to damage stainless steel turbine
blades. In patients with mechanical valves, cavitation may cause damage to
the blood and possibly to the valve structure itself. Cavitation has been
suggested as a possible reason for the incidents of leaflet fracture in patients
7328_C008.fm Page 307 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 307

with Edwards–Duromedics bileaflet valves. Bubbles created by cavitation

may act as nucleation sites for the formation of gaseous emboli. A form of
gaseous emboli, dubbed HITS (high intensity transient signals) due to its
nature as detected by Doppler ultrasound, has been observed in examina-
tions of the middle cerebral and carotid arteries of patients with mechanical
prostheses.
Many in vitro studies have directly visualized the formation and collapse
of cavitation bubbles near heart valve prostheses, and have used the number
and density of observed bubbles to estimate cavitation intensity. The high
frequency pressure oscillations, which occur immediately after valve closure,
have been linked to cavitation and have been suggested as a more accurate
means of quantifying its intensity. This would enable the quantification of
cavitation in vivo because fluctuations of pressure can be measured via a
catheter with a high frequency response. Such high frequency pressure fluc-
tuations have been observed in both sheep and patients with mechanical
heart valves. In an effort to develop pressure fluctuations as a means of
quantifying cavitation, recent work has successfully separated fluctuations
resulting from mechanical resonance of the valve structure from other, higher
frequency components representative of cavitation.
Most cavitation studies have been carried out on valves in the mitral
position, which has the harshest closure conditions of the four valve posi-
tions. During valve closure, the leaflets of a resting human mitral valve are
driven shut by a loading condition that is estimated to lie between 750 and
2000 mm Hg/s. This causes a prosthetic valve occluder to move from the
fully open position to the closed position in a time varying between 20 and
30 ms.
Cavitation potential has been linked to the severity of the closure condi-
tions and to valve design. An early study found that pitting (a sign of wear
indicative of cavitation) was noted on valves implanted in sheep forced to
exercise, but not on valves implanted in resting sheep. Both in vitro and in
vivo studies have found stronger signs of cavitation, as the loading condition
during closure is increased. Increased cavitation potential has been hypoth-
esized for valve designs that feature an occluder rim, which overlaps the
annular rim of the valve housing. A dimensional analysis of factors affecting
cavitation in tilting disc valves has linked cavitation potential with several
aspects of occluder design and material.
The detection of HITS in the arteries of patients with mechanical prostheses
has prompted studies into the nature of these emboli. While vaporous cav-
itation bubbles implode within a few milliseconds, microbubbles created
from gasses dissolved in the blood may remain in the circulation much
longer. The inhalation of pure oxygen significantly reduces the number of
HITS in the circulation of patients with mechanical prostheses, suggesting
that HITS are gaseous in nature, and that these emboli are composed pri-
marily of nitrogen and/or carbon dioxide. An in vitro study performed using
porcine blood has shown that gaseous emboli are likely to occur when the
partial pressure of carbon dioxide exceeds 100 mmHg.
7328_C008.fm Page 308 Tuesday, October 10, 2006 11:04 AM

308 Biofluid Mechanics: The Human Circulation

8.4 In Vitro Studies of Coagulation Potential

and Blood Damage
Results from clinical studies on the blood of patients with mechanical heart
valves have shown elevated levels of hemolysis, p-selectin, platelet factor 4,
and other biochemical markers of blood damage and platelet procoagulant
activity. Some of these markers have been used in vitro to quantify the effects
of controlled flow conditions across a prosthetic valve upon changes in the
condition of the blood.
Initial in vitro studies used plasma free hemoglobin as a marker of blood
damage. Some of these studies have linked hemolysis of porcine blood to
severity of leakage and closure flow conditions. A recent study has found
that much more flow energy is dissipated across platelets than erythrocytes
in humans, and suggests that markers of platelet procoagulant activity are
more sensitive to flow stimuli than plasma-free hemoglobin. Leakage gap
width has been linked to an increase in the expression of plasma platelet
factor 4 activity and Annexin V binding to platelet surfaces. The relationship
between plasma lactate dehydrogenase activity, turbulent shear stress, and
exposure of platelets to such a stress has been mathematically modeled and
used to compare the potentials of several valve designs to stimulate platelets
during both forward flow and leakage flow.
The clotting of milk has been suggested as an analog to the formation of
thrombus, as the biochemical mechanisms for milk clot and thrombus for-
mation are similar, and because milk clots and thrombus tend to occur under
similar flow conditions. A few studies have attempted to develop in vitro
systems that initiate milk clotting as a useful tool for predicting locations
where thrombus is likely to occur on heart valve prostheses. Though results
from these experiments are encouraging, milk coagulation has not yet devel-
oped into a tool that is considered predictive of thrombus formation.

8.4.1 Implications for Thrombus Deposition

In the vicinity of mechanical aortic heart valves where peak turbulent shear
stresses can easily exceed 150 N/m2, mean turbulent shear stresses are fre-
quently in the range of 20 to 60 N/m2 (Table 8.1), and cavitation is possible,
platelet activation and aggregation can readily occur. Data indicating that
shear-induced platelet damage is cumulative are particularly relevant to
heart valves. During an individual excursion through the replacement valve,
the combination of shear magnitude and exposure time may not induce
platelet aggregation. However, as a result of multiple journeys though the
artificial valve, shear-induced damage may accumulate to a degree sufficient
to promote thrombosis and subsequent embolization.
All of the aortic and mitral valve designs (mechanical and tissue) studied
created mean turbulent shear stress in excess of 20 N/m2 during the major
7328_C008.fm Page 309 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 309

portion of systole and diastole, a level that could lead to damage to blood
elements. In the case of mechanical prostheses, the chances for blood cell
damage are increased due to the presence of foreign surfaces. Furthermore,
the regions of flow stagnation and/or flow separation that occur adjacent to
the superstructures of these valve designs could promote the deposition of
damaged blood elements, leading to thrombus formation on the prosthesis.
The clinical performance of the Medtronic Parallel valve is an example of
the effects of these flow patterns on likelihood for thromboembolic compli-
cations. This valve showed superior forward flow hemodynamics in an in
vivo porcine model, suggesting a good clinical performance. However, the
Parallel valve performed very poorly in clinical trials; approximately 20% of
patients in these trials developed thrombosis. Patient and material factors
were statistically eliminated from the potential reasons for the poor perfor-
mance. Subsequent in vitro analysis of flow through the valve pivot revealed
regions of highly disturbed vortical flow within this area during the leakage
phase. Thrombi from explant valves were localized within the pivots to these
disturbed flow regions.

8.5 Durability of Prosthetic Heart Valves

The performance of prosthetic valves is in several ways related to structural
mechanics. The design configuration affects the load distribution and
dynamics of the valve components, which, in conjunction with the material
properties, determine durability — notably, wear and fatigue life. Valve
configuration, in concert with the flow engendered by the geometry, also
dictates the extent of low wear (e.g., flow separation) and high shear (e.g.,
gap leakage) regions. The hinges of bileaflet and tilting disc valves are
particularly vulnerable because their design can produce sites of stagnant
flow, which may cause localized thrombosis and restrict occluder motion.
As discussed earlier, the rigid circular orifice ring is an unnatural configu-
ration for a heart valve since the elliptically shaped natural valve annulus
changes in size and shape during the cardiac cycle.
The choice of valve materials is closely related to structural factors, since
the fatigue and wear performance of a valve depends not only on its con-
figuration and loading, but on the material properties as well. Additionally,
the issue of biocompatibility is crucial to prosthetic valve design where
biocompatibility depends not only upon the material itself, but also on its
in vivo environment. In the design of heart valves, there are engineering
design tradeoffs, for example, materials that exhibit good biocompatibility
may have inferior durability and vice versa. For many patients, the
implanted prosthetic valve needs to last well over a decade and the need
for valve durability may be even greater in the case of young people.
Mechanical durability depends on the material properties and the loading
7328_C008.fm Page 310 Tuesday, October 10, 2006 11:04 AM

310 Biofluid Mechanics: The Human Circulation

cycle with examples of degradation including fatigue cracks, abrasive wear,

and biochemical attack on the material.

8.5.1 Wear
Abrasive wear of valve parts has been, and continues to be, a serious issue
in the design of mechanical prosthetic valves. Various parts of these valves
come in contact repeatedly with each other for hundreds of millions of cycles
over the lifetime of the device. A breakthrough occurred with the introduc-
tion of pyrolytic carbon (PYC) as a valve material because it has relatively
good blood compatibility characteristics and wear performance. It has been
shown that while wear of PYC upon PYC and PYC upon metals is relatively
low, wear of PYC by metals is considerably greater. One example of this is
a PYC disc mounted on a metallic orifice/hinge combination. The most
durable wear couple is PYC–PYC; therefore, PYC-coated components are
very attractive. The first valve to employ a PYC–PYC couple was the St. Jude
Medical valve, which has fixed pivots for the leaflets. Tests indicate that it
would take 200 years to wear halfway through the PYC coating on a leaflet
pivot. By creating designs that allow wear surfaces to be distributed rather
than focal (e.g., the Omnicarbon valve, which has a PYC-coated disc that is
free to rotate), it is possible to reduce wear even further.
Despite these advances, wear on prosthetic heart valve surfaces can still
be problematic, as can be seen by the mechanical failures of the Edwards–
Duromedics valve. The Edwards–Duromedics valve design was introduced
in 1982, but withdrawn from the market after one of the leaflets in a number
of these valves fractured and embolized within the vascular system of
patients. Scanning electron microscopy examinations of the fractured leaflets
revealed regions of pitting on the leaflet and housing surfaces where the
leaflet contacts the housing during closure and within the pivot area as well
as several regions of concentrated micro pores in the carbon surfaces near
where the fractures occurred. A subsequent study showed that these micro
pores existed only in the Edwards–Duromedics design, and hypothesized
that they were created as a result of the shaping of carbon components with
a mandrel, a process that was unique to the Edwards–Duromedics valve.
This study linked pitting (indicative of cavitation damage) to the presence
of these micro pores. Other studies have disputed the connection between
micro pores and pitting damage. The fact that the cause of leaflet emboliza-
tion in these valves remains unknown means that wear remains a potential
problem in the design of mechanical prostheses.

8.5.2 Fatigue
Metals are prone to fatigue failure. Their polycrystalline nature contains
structural characteristics that may produce dislocations under mechanical
loading. These dislocations can migrate when subjected to repeated loading
cycles and can accumulate at intercrystalline boundaries with the end result
7328_C008.fm Page 311 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 311

being tiny cracks. These tiny cracks are sites of stress concentration in which
the fissures may worsen until fracture occurs. The Haynes 25 Stellite
Björk–Shiley valve, which used a chromium–cobalt alloy, experienced the
most severe fatigue problem for a mechanical valve. While previous inves-
tigations had suggested that fatigue was not a problem for PYC, recent data
contradicts this and suggests that cyclic fatigue-crack growth occurs in
graphite/pyrolytic carbon composite materials. This work suggests a fatigue
threshold that is as low as 50% of the fracture toughness and views cyclic
fatigue as an essential consideration in the design and life prediction of heart
valves constructed form PYC. The FDA now requires detailed characteriza-
tion of PYC materials used in different valve designs.

8.5.3 Mineralization
The major cause of both porcine aortic and pericardial bioprosthetic valve
failure is calcification, a process which stiffens leaflets and frequently causes
cuspal tears. Calcific deposits occur most commonly at the commissures and
basal attachments and are most extensively deep in (intrinsic to) the cusps
in the spongiosa layer. Ultrastructurally, calcific deposits are associated with
cuspal connective tissue cells and collagen. Degenerative cuspal calcific
deposits are composed of calcium phosphates that are chemically and struc-
turally related to physiologic bone mineral (hydroxyapatite). The flexing
bladders in cardiac assist devices, flexing polymeric heart valves, and vas-
cular grafts have also been found to be vulnerable to calcific deposits. In
such cases, calcification is usually related to inflammatory cells adjacent to
the blood-contacting surface, rather than to the implanted material itself.
The mechanisms of calcification and the methods of preventing calcifica-
tion are an active area of current research. The most common methods of
studying calcification involve valve tissue implanted either subcutaneously
in 3-week-old weanling rats or valves implanted as mitral replacements in
young sheep or calves. Results of both types of studies show that biopros-
thetic tissue calcifies in a fashion similar to clinical implants, but at a greatly
accelerated rate. The subcutaneous implantation mode is a well accepted,
technically convenient, economical, and quantifiable model for investigating
mineralization issues. It is also very useful for determining the potential of
new antimineralization treatments.
Host, implant, and biomechanical factors impact the calcification of tissue
valves. Patients who are young or have renal failure are vulnerable to valve
mineralization, but immunological factors seem to be unimportant. Pretreat-
ment of valve tissue with an aldehyde cross-linking agent has been found
to cause calcification in rat subcutaneous implants; non-preserved cusps, on
the other hand, do not mineralize. Calcification of bioprosthetic valves is
greatest at the cuspal commissures and bases where leaflet flexion is the
greatest and deformations are maximal. Most data suggest that the basic
mechanisms of tissue valve mineralization result from aldehyde pretreat-
ment, which changes the tissue microstructure.
7328_C008.fm Page 312 Tuesday, October 10, 2006 11:04 AM

312 Biofluid Mechanics: The Human Circulation

In both clinical and experimental bioprosthetic tissue, the earliest mineral

deposits have been observed to be localized in transplanted connective tissue
cells with the collagen fibers being involved later. Mineralization of the con-
nective tissue cells of bioprosthetic tissue is thought to result from glutaral-
dehyde-induced cellular ‘‘revitalization’’ and the resulting disruption of
cellular calcium regulation. Normal animal cells have a low intracellular free
calcium concentration (approximately 10−7 M), while extracellular free cal-
cium is much higher (approximately 10−3 M), yielding a 10,000-fold gradient
across the plasma membrane. In healthy cells, cellular calcium is maintained
at low concentration by energy-requiring metabolic mechanisms. In addition,
organellar and plasma membranes and cell nuclei — the observed sites of
early nucleation of bioprosthetic tissue mineralization — contain consider-
able phosphorus, mainly in the form of phospholipids. In cells modified by
aldehyde cross-linking, passive calcium entry occurs unimpeded, but the
mechanisms for calcium removal are dysfunctional. This calcium influx reacts
with the preexisting phosphorous and contributes to the mineralization.

8.6 Current Trends in Valve Design

Initially, improvement of hemodynamic performance of bioprosthetic valves
was the primary focus of development. The clinical introduction of the
bovine pericardial valve solved the hemodynamic problem with such valves
exhibiting hemodynamics equal to or better than some mechanical prosthe-
ses. Since long-term durability data have now become available, however,
clinical durability of bioprosthetic valves is the major impediment to their
use. The long-term durability of porcine and bovine bioprostheses can be
improved through innovative stent designs that minimize stress concentra-
tions, and through improved fixation processing techniques that yield more
pliable tissue.
If the above-mentioned design challenges are met so that bioprostheses
can be produced that are durable and are at the same thromboresistant (and,
thus, anticoagulant therapy would not be required), there will most likely
be another swing back toward increased use of bioprostheses.

8.7 Conclusions
Direct comparison of the ‘‘total’’ performance of artificial heart valves is
difficult, if not impossible. The precise definition of criteria used to bench-
mark valve performance varies from study to study. To study long-term
performance, large numbers of patients and lengthy observation periods are
required. During these periods, there may be an evolution in valve materials
7328_C008.fm Page 313 Tuesday, October 10, 2006 11:04 AM

Prosthetic Heart Valve Fluid Dynamics 313

or design, or in the medical treatment of patients with prosthetic heart valves.

The age of the patient at implant and the underlying valvular heart diseases
are extremely important factors in valve choice and longevity as well. A
valve design suited for the aortic position may be inappropriate for the mitral
position. Consequently, it is not possible to categorize a particular valve as
the best. All valves currently in use, mechanical and bioprosthetic, produce
relatively large turbulent stresses (that can cause lethal and/or sublethal
damage to red cells and platelets), and also greater pressure gradients and
regurgitant volumes than normal heart valves.
Therefore, there are three promising directions for further advances in
heart valves (and, therefore, three challenges for engineers designing new
heart valves):

• Improved thromboresistance with new and better artificial materials.

• Improved durability of new tissue valves, through the use of non-
stented tissue valves, new anticalcification treatments, and better
fixation treatments.
• Improved hemodynamic characteristics, especially reduction or
elimination of low shear-stress regions near valve and vessel sur-
faces and of high turbulent shear stresses along the edges of jets
produced by valve outflow and/or leakage of flow.

While the current status of artificial valves leaves room for further
improvement, the superior prognosis for the patient with a replacement
heart valve is dramatic and convincing.

8.8 Problem
8.1 A 27-mm bileaflet valve was tested in a pulse duplicator at a mean
systolic flow rate of 15 lpm. The corresponding mean pressure drop across
the valve was measured to be 5.2 mmHg. Assuming a discharge coefficient
of 0.61 for the bileaflet valve, compute the effective valve orifice area and
the performance index for the valve.
7328_C008.fm Page 314 Tuesday, October 10, 2006 11:04 AM
7328_C009.fm Page 315 Monday, October 16, 2006 11:05 AM

9
Vascular Therapeutic Techniques

CONTENTS
9.1 Vascular Graft Implants............................................................................315
9.2 Arteriovenous Fistulas..............................................................................317
9.3 Types of Vascular Graft Materials Used ................................................319
9.4 Clinical Experience with Vascular Grafts ..............................................322
9.5 Biomechanics and Anastomotic Intimal Hyperplasia .........................324
9.6 Angioplasty, Stent, and Endoluminal Graft Implants.........................333
9.7 Biomechanics of Stent Implants ..............................................................338
9.8 Problems......................................................................................................342

9.1 Vascular Graft Implants

Earlier in Chapter 3 (section 3.9: Atherosclerosis), we discussed how athero-
sclerosis of the major systemic arteries (e.g., coronary, carotid, renal, etc.) is
the greatest cause of death in the Western world, leading to conditions such
as heart attacks, strokes, and disabilities of the legs. Later, in Chapter 6
(section 6.7: Flow through Arterial Stenoses and Aneurysms), we discussed
the topic of arterial stenoses and the nature of fluid dynamic changes, which
are associated with them. From a clinical standpoint, the presence of a stenosis
often leads to either reduced blood flow to the corresponding distal organs
or to the production of emboli (usually thrombi), which are convected down-
stream to smaller arterial branches where occlusion occurs. The most critical
sites for these events are the coronary (Figure 9.1), carotid, renal, and lower
extremity (e.g., iliac, femoral, etc.) arteries.
The primary means of treating these stenosed arteries is to provide an
alternative path, or a “bypass,” for the blood in order to circumvent the
obstruction and reach the dependent end organ. Arterial bypasses can be
readily created by implanting a vascular graft around the diseased area
(Figure 9.2) since arterial stenoses are focal in nature and, thus, leave neigh-
boring segments of the artery relatively unaffected.

315
7328_C009.fm Page 316 Monday, October 16, 2006 11:05 AM

316 Biofluid Mechanics: The Human Circulation

LMCA LAD

CIRC

OM

FIGURE 9.1
Angiogram of the left coronary system of a patent with severe coronary artery disease. (From
Gotto, A.M. Jr. et al. (1977) Atherosclerosis, Upjohn, Pfizer, Inc., New York. With permission.)

FIGURE 9.2
Coronary arteriograms, including 7-year follow-up of patient who underwent coronary artery
bypass to left anterior descending coronary artery. (From Gotto, A.M. Jr. et al. (1977) Atheroscle-
rosis, Upjohn, Pfizer, Inc., New York. With permission.)
7328_C009.fm Page 317 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 317

In this surgical procedure, the affected vessel is exposed and the bypass
graft is placed alongside it and sewn to arterial sites well upstream and
downstream of the stenosis. The junctions between the artery and the graft
are called anastomoses and are generally constructed in an end-to-side
fashion. The primary exception to this is with grafts used for repair of an
abdominal aortic aneurysm where the anastomoses are placed in an end-to-
end fashion. Once the graft is in place, blood flow is then reinstituted around
the diseased area, often with some residual flow still being carried through
the native artery. A variety of surgical procedures are used to perform cor-
onary artery bypass grafting with internal mammary arteries, ranging from
the traditional open-chest procedure to newer approaches using either a
thoracotomy (OPCAB) or only a small incision in the ribcage (MIDCAB),
both of which avoid the need for simultaneous heart–lung bypass support
during the procedure.

9.2 Arteriovenous Fistulas

Another clinical purpose for using vascular grafts is to simply provide a
direct access to the blood stream. This is important, for example, when
performing hemodialysis on a patient with renal failure or for administering
chemotherapy to cancer patients. With this procedure, a bypass is created
between an artery and a vein (rather than between two points on the same
or connected arteries), thus allowing for blood to be withdrawn and then
returned to the circulation or for therapeutic agents to be delivered directly
into the bloodstream. There are two main types of arteriovenous fistulas
(AVFs). One is a direct anastomosis between an artery and its adjacent vein
in a side-to-side fashion (called a native AVF), while the other uses a syn-
thetic graft implant (usually ePTFE [expanded polytetrafluoroethylene]
material), which is anastomosed between the artery and vein in an end-to-
side fashion (Figure 9.3). The most common locations for native AVFs and
synthetic graft fistulas are between the radial artery and vein of the forearm
and between the brachial artery and vein of the upper arm.
Since a fistula directly joins blood at arterial pressures (80 to 100 mmHg)
with blood at venous pressures (10 to 20 mmHg), fistula flow rates are
normally very high (often 10 to 20% of total cardiac output, depending upon
the anatomic site) and the flow is typically turbulent (Figure 9.4).
This elevated flow rate (often 800 to 1000 ml/min) is useful in the process
of hemodialysis, for example, since a large amount of blood is readily acces-
sible and the detoxification process can be accomplished more quickly (~3 to
4 hours). Despite these high flow rates, usually patients do not suffer from
major complications other than occasionally experiencing a “steal syndrome”
in which enough blood is diverted from distal vessels to cause coolness and
numbness of the tissue. These fistulas, however, do have a fairly high failure
7328_C009.fm Page 318 Monday, October 16, 2006 11:05 AM

318 Biofluid Mechanics: The Human Circulation

FIGURE 9.3
A brachiobasilic bridge (loop) graft. (From Ernst, C.B. and Stanley, J.C. (1991) Current Therapy
in Vascular Surgery, 2nd ed, B.C. Decker, Philadelphia. With permission.)

FIGURE 9.4
Flow pattern as seen in the high volume flow graft. (From Villemarette, P.T. et al. (1989) Use of
color flow Doppler to evaluate vascular access graft function, in J. Vasc. Tech. 13: 164–170. With
permission.)
7328_C009.fm Page 319 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 319

rate with an average longevity of only 1 to 2 years. Obviously, this event

further increases the patient’s overall morbidity. Most failures are immedi-
ately due to thrombosis, especially with graft fistulas, which then require
surgical revision or replacement of the graft. The thrombosis itself is often
secondary to the development of intimal hyperplasia at the anastomoses,
particularly at the venous anastomosis.

9.3 Types of Vascular Graft Materials Used

A variety of synthetic and biological materials are used for the construction
of bypass grafts. The preferred choice is to use one of the patient’s own
vessels since this provides the best possible degree of biocompatibility. In
practice, either an intrathoracic artery (i.e., the internal mammary artery) or
a vein from the leg (i.e., the saphenous vein) is most commonly harvested
for this purpose. The internal mammary artery (IMA) is used exclusively
for bypasses of the coronary artery and is nearly ideal because of its similarity
in size, availability (i.e., it is not a critically required vessel), and its natural
arterial structure. It is limited, however, by the fact that it is short and that
only two such vessels (left and right IMAs) exist in the body. The saphenous
vein is also used for coronary artery bypass procedures, but it is also com-
monly used to construct bypasses of the lower extremity (e.g., femoral/
popliteal artery bypasses) as well. The saphenous vein graft (SVG) is a good,
autologous conduit material with low thrombogenicity, but it is also limited
in its quantity (especially when multiple bypass grafts and/or procedures
are required) and sometimes by its quality in older patients. It also suffers
from a process termed ”arterialization,” in which the vein, by thickening its
walls, adapts to the abrupt increase in pressure, which it experiences upon
implantation in an arterial environment.
For these reasons, other synthetic materials have also been used as bypass
grafts. Chief among these are polyethylene terephthalate (PET) or Dacron®,
polytetrafluoroethylene (PTFE) or Teflon®, and polyurethane. Dacron is a
strong fabric with long life and is the normal material used in large diameter,
abdominal aortic aneurysm (AAA) repairs (Figure 9.5).
For smaller caliber grafts, such as those in the leg, PTFE is the main graft
material used due to its relatively inert surface properties (Figure 9.6). A
disadvantage of this material, however, is that it is very stiff (Young’s mod-
ulus, E > 107 N/m2). Therefore, a more compliant, expanded version of this
material, ePTFE, has been adopted as the current surgical standard. Poly-
urethane has the advantages of being relatively easy to manufacture and
having good mechanical properties, although it does not have thrombore-
sistant properties, which are comparable to those of PTFE.
An attempt has been made to improve the antithrombogenic properties
of synthetic materials and to exploit their advantage of greater availability
7328_C009.fm Page 320 Monday, October 16, 2006 11:05 AM

320 Biofluid Mechanics: The Human Circulation

(a) (b)

Baylor College of Medicine 1941

(c) (d) (e)

FIGURE 9.5
Technique for resection and replacement of infrarenal aneurysm of abdominal aorta. (From
Ernst, C.B. and Stanley, J.C. (1991) Current Therapy in Vascular Surgery, 2nd ed, B.C. Decker,
Philadelphia. With permission.)

by incorporating a natural biological surface on the material. This proce-

dure, known as endothelial cell seeding, involves the placing of endothe-
lial cells derived from the same patient onto the surface of the graft (usually
ePTFE material) prior to surgery in order that the graft will present a
natural appearing surface to the blood. The technique uses endothelial cells
from either a harvested vein or from microvascular tissue, which are then
“seeded” onto the graft surface using either gravitational, adhesive, or
electrostatic methods and allowed to replicate until they reach confluence
(i.e., full coverage) (Figure 9.7). Once the surface is stabilized, the graft is
implanted and the endothelial cells not only prevent a foreign body
7328_C009.fm Page 321 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 321

(a) (b)

FIGURE 9.6
Arteriogram in a patient with patent femoroposterior tibial ePTFE graft 10 years after placement
of the graft: (a) proximal graft, (b) distal anastamosis. (From Ernst, C.B. and Stanley, J.C. (1991)
Current Therapy in Vascular Surgery, 2nd ed, B.C. Decker, Philadelphia. With permission.)

FIGURE 9.7
Scanning electron micrograph illustrating results of the prime electrostatic endothelial cell
transplantation onto ePTFE (GORE-TEX®). (From Fields, C. et al. (2001) The persistence of
electrostatically seeded endothelial cells lining a small diameter expanded polytetrafluoroeth-
ylene vascular graft, in J. Biomaterials Applications 16:157-173. With permission.)
7328_C009.fm Page 322 Monday, October 16, 2006 11:05 AM

322 Biofluid Mechanics: The Human Circulation

response, but also may actively reduce the potential for thrombosis and
intimal hyperplasia by the production of agents, such as prostaglandins
and nitric oxide.

9.4 Clinical Experience with Vascular Grafts

In most cases, a bypass graft’s ability to remain open, called its patency, is
excellent immediately after surgery (>90%), but continually decreases with
time (~50% at 5 years) (Figure 9.8). The reasons for these failures are related
to either early (<30 days) thrombus formation or late (>30 days) intimal
hyperplasia formation. Graft thrombosis occurs due to the trauma associ-
ated with the implantation surgery and also as part of the body’s response
to the implantation of a foreign material. It is also more likely to occur in
regions of slow or static flow where the tendency for clotting is enhanced.
One result of this is that failure rates for grafts extending below the knee,
for example, are particularly high due to the low flow rates carried in these
smaller caliber (<6 mm I.D.) vessels. It is for this reason as well that, currently,

100
440
90
375
80
291 Secondary patency
201
70 120
81
57 27
60 21
% Patency

11 6 3 2
50 51%
Primary patency
40 44%

30

20

10

0
0 6 12 18 24 30 36 42 48 54 60 66 72
Months after operation

FIGURE 9.8
Cumulative life table patency rates for 440 PTFE femoropopliteal bypasses. The lower line
illustrates primary patency rates, while the upper line shows secondary (achieved after throm-
bectomy) patency rates. (From Sawyer, P. N. (1987) Modern Vascular Grafts, McGraw-Hill, New
York. With permission.)
7328_C009.fm Page 323 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 323

Flow
POS flow Flow
Proximal or donor end

Graft heel
Bypass graft

Hood
Artery
floor
Suture line

Graft toe

Dos flow
Distal or receptor end

Flow

FIGURE 9.9
Sites of intimal hyperplasia formation in a bypass graft distal anastomosis (POS — proximal
outflow segment, DOS — distal outflow segment). (From Li, X.-M. (1998) Evaluation of hemo-
dynamic factors at the distal end-to-side anastomosis of a bypass graft with different POS:DOS
ratios, Ph.D. Dissertation Dept. of Biomedical Engineering, The University of Akron, OH.)

no grafts made of synthetic materials are used for coronary artery (3 to 4

mm ID) bypass procedures.
The development of intimal hyperplasia (IH) is a natural response of the
vessel to specific conditions, especially those present at bypass graft anasto-
moses. As shown in Figure 9.9, IH occurs primarily along the suture line
regions between the graft and host artery (known as the anastomotic “heel”
and “toe” regions) and also along the “floor” of the host artery directly across
from the graft outlet.
As its name implies, IH consists of tissue in the intimal layer of the wall,
which experiences excessive growth. The principal cells involved in this
process are the same as those responsible for the development of atheroscle-
rosis (see Chapter 3, section 3.10: Atherosclerosis). Specifically, smooth mus-
cle cells and fibroblasts proliferate within the vessel media and adventitia,
respectively, and then migrate into the intimal layer as part of the vessel’s
healing response. While this process involves only normally present vascular
tissue and not other components, such as cholesterol, calcium, and necrotic
tissue, it can still culminate in the same outcome vessel occlusion due to a
partial obstruction combined with eventual thrombosis of the graft.
The responses of the artery wall appear to coincide with changes in its
local biomechanical environment, such as pressure and wall shear stress
7328_C009.fm Page 324 Monday, October 16, 2006 11:05 AM

324 Biofluid Mechanics: The Human Circulation

(see Chapter 6, section 6.5: Wall Shear Stress and its Effect on Endothelial
Cells). Since these variables are particularly affected by the geometric
configuration of the anastomosis, extensive work has been performed to
examine the effect of the choice of graft caliber and the anastomotic angle,
in addition to the flow rate, upon eventual patency of the graft.

9.5 Biomechanics and Anastomotic Intimal Hyperplasia

While implantation of a bypass graft is usually successful in restoring
needed blood flow to distal tissues, it does so by creation of nonanatomic
junctions between vessels which, in turn, lead to complex, 3-D flow pat-
terns. These flow patterns are mainly confined to either the proximal or
distal anastomoses, where sharp angles and abrupt changes in vessel diam-
eters are present. Although the flow at these sites is typically laminar in
character, complex features are present, such as skewed velocity profiles,
separation zones, and reattachment points, as well as strong secondary
motions (Figure 9.10).
Important consequences of this include the presence of wide variations
(i.e., high, low, and reversing) in local wall shear stresses and extended

(a)

(b)

FIGURE 9.10
Three-dimensional sketch of the path line of the high inertia fluid seen under steady flow
conditions at Re = 950 and 1300, as well as for unsteady flow at around peak flow: (a) side
view, showing development of the double helix; (b) top view. (From Ojha, M. et al. (1990)
Influence of angle on wall shear stress distribution for an end-to-side anastomosis model, in
J. Vasc. Surg. 12: 747–53. With permission.)
7328_C009.fm Page 325 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 325

residence times for blood cells and other components to interact with the
vessel wall. A compounding factor related to this is the stiffness, or lack of
compliance (= Volume/Pressure), of the graft material since changes in
vessel diameter under pulsatile flow conditions will further alter the local
geometry. A mismatch in compliance between the graft and native artery
could also lead to significant changes in local fluid and solid wall stresses,
which might then elicit various tissue responses. We have mentioned earlier,
for example, the relatively high stiffness of ePTFE material, which when
interposed with more elastic native vessels may produce elevated wall
stresses and impedance mismatches. This problem may persist, however,
even when using more compliant biological materials, such as veins, since
they may also become very stiff when exposed to arterial pressures. It is not
a surprising observation, then, that intimal hyperplasia commonly develops
at the toe and heel regions as well as along the artery floor of the anastomosis
(Figure 9.11).
Because these sites coincide closely with areas of altered biomechanical
variables, a number of studies have been performed to look at the effect of
geometric and material parameters upon the local fluid dynamics and sub-
sequent vascular responses. Specific variables that have been investigated
include: (1) material stiffness differences between the two vessels, (2) the
angle of the anastomosis, (3) the diameter mismatch between the graft and
the host artery, and (4) the flow distribution through the artery (i.e., the

FIGURE 9.11
Intimal hyperplasia has developed at the anastomosis of this Dacron aortofemoral graft with
the profunda femoris artery. At the time of patch angioplasty, a smooth, glistening, white plaque
was found. (From Gupta, S.K. (1993) Vascular Graft Monitoring, R.G. Landes Co., Austin, TX.)
7328_C009.fm Page 326 Monday, October 16, 2006 11:05 AM

326 Biofluid Mechanics: The Human Circulation

Compliance vs % Patency

R = 0.873
100

75
% Patency

50

Arterial autografts
25 Vein autografts
Dacron
P.T.F.E.

0
1 2 3 4 5 6
Compliance (%/mm Hg × 10–2)

FIGURE 9.12
Linear regression analysis of compliance vs. graft patency. (From Abbott W.M. and Bouchier-
Hayes D.J. The Role of Mechanical Properties in Graft Design in (1978) Graft Materials in Vascular
Surgery, Dardik, H., Ed. Year Book Medical Publishers, Chicago. With permission.)

proximal vs. distal outflows), particularly at the distal anastomosis. From

these studies, better insights have been gained as to the potential advantages
and disadvantages of using various geometric and material configurations.
Specifically, graft-to-artery compliance differences have been shown to cor-
relate with subsequent loss of viability of the graft (Figure 9.12). Thus, a
graft, whether biological or synthetic, would be predicted to have higher
failure rates as the stiffness of the graft material increases (i.e., graft compli-
ance decreases).
Regarding changes in the graft-to-artery anastomotic angle, results from
models with anastomoses of 30, 45, and 60 degrees showed an increased
skewing of the velocity profile and an increased flow separation in the
outflow artery segment over that angle range (Figure 9.13).
Computation of the wall shear rates [WSRs] along the vessel wall relate
these findings in more detail and show the wide range of forward and reverse
shear effects present (Figure 9.14), especially at the larger angle. This suggests
that small angles would be preferable — a conclusion that is supported by
the fact that grafts anastomosed in an end-to-end fashion generally have
higher patency rates than those anastomosed in an end-to-side fashion.
Clearly, these fluid mechanical changes appear to correspond quite closely
with earlier observed sites of IH (see Figure 9.9) at the heel and toe and
along the artery floor of the anastomosis. In the above figures, it is also
evident that the graft flow rate (which is proportional to Re in these models)
7328_C009.fm Page 327 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 327

FIGURE 9.13
Flow patterns in anastamoses of 30, 45, and 60 degrees at Re = 205 (Note: Images were obtained
using H2 bubble flow visualization.) (From Keynton R.S. et al. (1991) The effect of graft caliber
upon wall shear within in vivo distal vascular anastomosis, in J. Biomech. Eng. 113: 458–463.
With permission.)

plays an important role in determining WSR values with larger regions of

the artery being exposed to low WSRs at the lower Re. This observation is
consistent with clinical findings that, in general, patency rates of bypass
grafts are better under high flow (i.e., larger, more proximal vessels) rather
than low flow (i.e., smaller, more distal vessels) conditions.
In many bypass procedures, there is also a discrepancy between the caliber
of the graft and the host artery. This occurs because (1) there are limited
choices of sizes available with biological grafts, (2) SVGs must be reversed
in order to avoid flow obstruction by their intact valves (thus creating a size
mismatch at both the proximal and distal anastomoses), and (3) synthetic
grafts are normally uniform in diameter while human arteries and veins are
tapered. From animal experiments, it was shown that increasing the graft/
artery diameter ratio from 1.0 to 1.5 (at the same 30 degree anastomotic
angle) produces a greater magnitude and extent of flow reversal exposure
in the anastomotic region (Figure 9.15).
Finally, another factor, which is thought to significantly affect the local anas-
tomotic flow patterns, is the relative distribution of blood flow to the proximal
outflow segments (POS) and distal outflow segments (DOS) of the artery. This
POS:DOS flow ratio depends primarily on the degree of obstruction in the
bypassed artery (i.e., whether it is occluded or partially stenosed) and the
resistance of the distal bed (i.e., the degree of vasodilatation/vasoconstriction).
Presence of proximal arterial outflow tends to reduce flow separation zone at
the heel and artery floor, but increases that effect at the toe region and, par-
ticularly, in the hood region of the graft. It also reduces the wall shear stresses
along the entire distal outflow segment of the artery (Figure 9.16).
In clinical practice, it is quite common for many of these factors to occur
in combination. For example, a Dacron material might be chosen as a bypass
7328_C009.fm Page 328 Monday, October 16, 2006 11:05 AM

328 Biofluid Mechanics: The Human Circulation

Normalized axial shear rate vs. Axial position

Re = 100
30° Inner wall

Outer wall

Re = 205
30° Inner wall

Outer wall

Normalized axial shear rate vs. Axial position

Re = 100
45° Inner wall

Outer wall

Re = 205
45° Inner wall

Outer wall

Normalized axial shear rate vs. Axial position

Re = 100
60° Inner wall

Outer wall

Re = 205
60° Inner wall

Outer wall

FIGURE 9.14
Normalized axial shear rate in anastomoses of 30, 45, and 60 degrees at Re = 100 and 205. (Note:
Wall shear rates were derived from laser Doppler anemometry velocity measurements.) (From
Keynton R.S. et al. (1991) The effect of graft caliber upon wall shear within in vivo distal vascular
anastomosis, in J. Biomech. Eng. 113: 458–463. With permission.)
7328_C009.fm Page 329 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 329

60

Mean velocity (cm/s)

Radial position Toe region
0.35 mm
0.70 mm
1.05 mm 40

20
Graft inlet

0 4 8 12 (mm)
Artery outlet
–12 –8 –4 0 4 8 12 (mm)

60
Mean velocity (cm/s)

40

20

Artery floor
–20

60
Toe region
Mean velocity (cm/s)

Radial position
0.35 mm
0.70 mm 40
1.05 mm

20
Graft inlet
0
0 4 8 12 (mm)
Artery outlet
–12 –8 –4 0 4 8 12 (mm)

60
Mean velocity (cm/s)

40

20

Artery floor
–20

FIGURE 9.15
Typical plots of mean axial velocities at three radial positions along the artery toe and floor
regions for diameter ratio (DR) = 1.0 (top) and DR = 1.5 (bottom). (Note: Velocity measurements
made using pulse Doppler ultrasound.) (From Keynton, R.S. et al. (1999) The effect of graft
caliber upon wall shear within in vivo distal vascular anastomosis, in J. Biomech. Eng. 121: 79–88.
With permission.)
7328_C009.fm Page 330 Monday, October 16, 2006 11:05 AM

330 Biofluid Mechanics: The Human Circulation

Flow split: 50:50 Flow split: 75:25

50 cm/s

t = 40 ms t = 40 ms

t = 150 ms t = 150 ms

t = 210 ms t = 210 ms

S
R

t = 300 ms t = 300 ms

t = 625 ms t = 625 ms

FIGURE 9.16
Vector plots in the ETS anastomosis for flow splits (ratio of distal to proximal arterial outflow)
of 50:50 and 75:25 at various times during the flow cycle. Flow separation at S and reattachment
at R occur on the graft hood during the systolic deceleration phase (t = 210 and 300 ms) with
a flow split of 50:50. (From How T.V. et al. (2000) Interposition vein cuff anastomosis alters wall
shear stress distribution in the recipient artery, in J. Vasc. Surg. 31: 1008–17. With permission.)

graft for replacement of an abdominal aortic aneurysm and anastomosed

distally to the iliac artery in either an end-to-end or end-to-side fashion.
Mean wall stresses in the anastomosis are not only elevated due to the
compliance mismatch between the artery and graft, but also by the choice
of anastomotic geometry (Figure 9.17 and Figure 9.18).
While the above results provide trends, which may be useful in surgical
treatment, they don’t provide a single geometric, hemodynamic, or material
factor that is responsible for the development of intimal hyperplasia
observed in practice. Consequently, many researchers have attempted to
identify a possible common intermediate variable, which may be directly
related to IH development. This parameter might vary in response to
changes in the anatomic and physiologic conditions investigated above and,
in turn, produce a predictable effect upon the vessel wall. If such a parameter
7328_C009.fm Page 331 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 331

Y Symmetry plane
X

Z
Mean stress
Host 32
Sutures
artery 28
24
20
Dacron 16
graft 12
8
4
Symmetry plane
0

FIGURE 9.17
Mean stress distribution at the end-to-end Dacron graft–artery anastamosis projected onto a
three-dimensional image of the geometry. (From Ballyk P.D. et al. (1998) Compliance mismatch
may promote graft-artery intimal hyperplasia by altering suture-line stresses, in J. Biomech. 31:
229–37. With permission.)

were identified as having direct influence on the cell–tissue response, it

might then be controlled in order to produce the desired clinical outcome
(i.e., reduced IH). Several key candidates that have been suggested are mean
and oscillatory wall shear and its gradients (spatial and temporal), wall
stress, wall strain, etc. Given our growing knowledge of the vascular wall
and its ability to monitor local dynamic conditions and to actively respond
to them, data have been obtained that compare the eventual (12 week)

Z
Y X
all
ew Mean stress
Sid
Sutures 32
28
Toe 24
Host 20
artery 16
12
Heel Dacron
graft 8
4
Plane of 0
Symmetry

FIGURE 9.18
Mean stress distribution at the end-to-side Dacron graft–artery anastomosis projected onto a
three-dimensional image of the geometry. (From Ballyk P.D. et al. (1998) Compliance mismatch
may promote graft-artery intimal hyperplasia by altering suture-line stresses, in J. Biomech. 31:
229–237. With permission.)
7328_C009.fm Page 332 Monday, October 16, 2006 11:05 AM

332 Biofluid Mechanics: The Human Circulation

1200

Intimal hyperplasia (µm) 1000

800

600

400

200

–200
–500 –250 0 250 500 750 1000 1250 1500
Mean wall shear rate (1/s)
(a)

1200

1000
Intimal hyperplasia (µm)

800

600

400

200

–200
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Oscillatory shear index
(b)

FIGURE 9.19
Nonlinear regression of (top) intimal hyperplasia (IH) vs. mean wall shear rate (WSR) (r = – 0.483)
and (bottom) IH vs. oscillatory shear index (OSI) (r = 0.600). (From Keynton R.S. et al.
(2001) Intimal hyperplasia and wall shear in arterial bypass graft distal anastomoses, in
J. Biomech. Eng. 123: 464–73. With permission.)

development of IH with the level of wall shear rate (WSR) at the time of
implantation (Figure 9.19). These data show a modest inverse relationship
(r = – 0.483) between IH and WSR, where IH is greatest at low and reversing
WSRs and reaches a near-zero asymptote around 1000 s−1 — a level compa-
rable to ~ 2 times the normal mean WSR seen in major arteries. Furthermore,
there is a direct correlation (r = 0.6) of IH with the OSI (oscillatory shear
7328_C009.fm Page 333 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 333

index), a parameter sensitive to the degree of pulsatility in the flow (see

Chapter 6, section 6.4.2: Time Varying Wall Shear Stress, Oscillatory Shear
Index, and Wall Shear Stress Gradients).
These data suggest that, while wall shear is not the only factor responsible
for IH, it does have an influence on the subsequent changes in the vessel
wall following bypass grafting and is at least one important factor that
should be controlled through modulation of other physiologic (e.g., flow
rate) and anatomic (e.g., anastomotic angle, graft caliber) variables.
In practice, it is always desirable to try to “translate” these investigational
findings into practical recommendations for use by the surgeon in order to
improve the success of an arterial bypass procedure. Certainly, reduction of
flow through a vessel has been shown to be a potent stimulant of subsequent
changes in the vessel wall. Langille and O’Donnell (1986), for example, have
shown that reduction of blood flow in rabbit carotid arteries causes a rapid
and significant decrease in the artery diameter and involves regulation of
vascular cell migration and mitosis and apoptosis rates, control of matrix
synthesis and degradation, and regulation of matrix reorganization. These
changes are mediated by the endothelial cells, which sense mechanical forces
through shear-sensitive ion channels, and the shear strain rate acting on the
cell. Thus, it is common to experience better success in bypassing larger
arteries with higher flow rates rather than smaller arteries with low flow
rates and to also attempt to place the distal anastomosis proximal rather
than distal to a branch point so that more total flow will be carried in the
graft. Furthermore, the use of near-zero graft-to-artery anastomotic angles
(i.e., end-to-end) has produced superior clinical results compared to those
placed at greater angles, and so, a low-angle hooded anastomosis is generally
created. Although the presence of a POS from an anastomosis (due to the
presence of a partial rather than a complete proximal arterial obstruction)
has been shown to dramatically affect the local flow patterns and, especially
the wall shear stress, its possible effect on the development of anastomotic
intimal hyperplasia has not been as clearly confirmed. Thus, the degree of
proximal stenosis (e.g., by tying off the stenosed artery) is not actively con-
trolled in practice as part of a preferred surgical procedure.
Finally, there are inevitable interactions between many of these factors,
which lead to additional effects upon the vessel response. Clearly, there still
remains a need for further research on these questions in order to better
understand these effects.

9.6 Angioplasty, Stent, and Endoluminal Graft Implants

More recent methods for treating arterial stenosis include less invasive pro-
cedures in which the vessel is not exposed through a direct surgical approach,
but rather is accessed using a catheter delivery system (Figure 9.20).
7328_C009.fm Page 334 Monday, October 16, 2006 11:05 AM

334 Biofluid Mechanics: The Human Circulation

Dye C.P. Dilatation

Balloon P. Wire
cath.

C.P.
Dye

Wire C.P. & dye

Guiding cath.

FIGURE 9.20
Schematic of a percutaneous balloon angioplasty device showing guidewire (Wire), catheter,
and ports for balloon pressure (Balloon P.), dye injection (Dye), and catheter pressure (C.P.).

With this percutaneous transluminal angioplasty (PTA) technique, a bal-

loon device is placed near the tip of a catheter, which is then inserted through
an artery in an extremity (typically, the common femoral artery in the leg).
The catheter is then positioned within a stenosis site, such as one of the
coronary or carotid arteries, and the balloon inflated at elevated pressures
up to 6 atm in order to enlarge the stenotic lumen. The resulting increase in
vessel lumen size (Figure 9.21) has a dramatic effect on reducing the pressure
drop and, thus, the resistance to blood flow. (See Chapter 1, section 1.8: Fluid
Mechanics in a Straight Tube, Equation 1.74.)
Typically, the initial results with this procedure are very good and far less
traumatic than with surgical bypass procedures, with many patients recov-
ering quickly at lower cost. It has been found, however, that a large number
of these patients experience recurrent symptoms similar to those that were
originally present only a few months or years after the procedure was per-
formed. When this happens, repeat contrast angiography often shows that
the initial atherosclerotic plaque has “reappeared” in the vessel. This is due
to the fact that angioplasty does not remove the plaque, but only displaces
it into the vessel wall, which may exhibit a recoil effect due to its elastic
nature. Progression of the underlying atherosclerosis is also thought to con-
tribute to this phenomenon of restenosis. Therefore, an improvement to this
technique has been to combine PTA with placement of a vascular stent
within the expanded vessel following angioplasty (Figure 9.22).
Vascular stents are made of various metals, such as stainless steel and tita-
nium. They are fabricated in tube or wire forms that are then either cut or
shaped into a mesh or coil-like pattern. Each stent design is produced in a
compact configuration so that it may pass through the stenosis and then be
expanded once in position. Thus, the stent is placed over the balloon compo-
nent of its delivery catheter and then deployed at the site of stenosis, following
7328_C009.fm Page 335 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 335

(a) (b) (c)

FIGURE 9.21
Digital subtraction angiograms showing excellent and long-term immediate results of percuta-
neous transluminal angioplasty in a patient with severe ulcerated internal carotid stenosis.
(From Brown, M.M. (1996) Balloon angioplasty for extracranial carotid disease, in Advances in
Vascular Surgery, Whittemore, A.D., Ed. (1996) Advances in Vascular Surgery, vol. 4, Elsevier,
Philadelphia. With permission.)

PTA with a separate catheter. Once in place, the stent provides a mechanical
scaffold that can help sustain the vessel opening over time (Figure 9.23).
Other variations on this approach involve compressing a coiled wire mesh
inside a small plastic sleeve and then deploying it within the stenosis by
pulling back the sleeve and allowing the stent to spring open (Figure 9.24).
Another approach is to use a unique metal, Nitinol, which has “thermal
memory” properties. This type of stent is first shaped into a coil at warm
temperatures and then straightened into a wire at room temperature (~ 22°C).
Once it is reheated by the body’s blood (~37°C) at the site of stenosis, the
device then returns to its original coiled shape.

FIGURE 9.22
Balloon-expandable coronary artery stent. (From Didisheim, P. and J.T. Watson, Cardiovascular
Applications. Ratner, B.D., Ed. (1996) Biomaterials Science: An Introduction to Materials in Medicine,
Elsevier, Philadelphia. With permission.)
7328_C009.fm Page 336 Monday, October 16, 2006 11:05 AM

336 Biofluid Mechanics: The Human Circulation

FIGURE 9.23
Digital subtraction angiogram immediately after the percutaneous transluminal insertion of a
stent across a carotid stenosis at the bifurcation. (From Brown, M.M. (1966) Balloon angioplasty
for extracranial carotid diseases, in Advances in Vascular Surgery Vol. 4, 1996, A.D. Whittemore,
Ed., Elsevier, Philadelphia. With permission.)

A therapeutic device that evolved subsequent to the use of vascular stents

is the endoluminal stent graft (Figure 9.25). Endoluminal graft implants are
synthetic grafts placed in the circulation system to repair a diseased arterial
segment, but without requiring the use of traditional surgical techniques.
The way this is accomplished is to actually deliver the graft to the site using
an intra-arterial catheter, which then deploys the graft by balloon inflation.
The graft is held in place by the inclusion of stents that are positioned at
either the proximal end only or at both the proximal and distal ends of the
endoluminal graft and which are also deployed by balloon inflation. The
primary application for this device is abdominal aortic aneurysm repair in
patients who are at high risk of mortality during open abdominal surgery.
The graft material is usually a Dacron fabric and the stents are usually
metallic strands in a zigzag configuration. The device is collapsed over the
delivery catheter and covered by a smooth plastic sheath until it is in position
7328_C009.fm Page 337 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 337

FIGURE 9.24
A photo of a Wallstent. The tubular, highly permeable braided mesh has considerable longitudinal
flexibility and radial strength. (From Gwertzman, G. (1995) Clinically tested intravascular stents
and endovascular stent grafts, in Endovascular Stented Grafts for the Treatment of Vascular Diseases,
Marin, M.L., Veith, F.J., and Levine, B.A., Eds. R.G. Landes Co., Austin, TX. With permission.)

for deployment. The catheter used for this procedure is a heavier design
than that used for intra-arterial stent delivery since it carries a larger object
and also because it only passes through the larger arteries of the upper leg
and abdomen. In constructing an endoluminal graft for AAAs, which also
extend down into the iliac arteries from the aorta (and, thus, require distal
anastomoses to each iliac artery), a modular device is used (Figure 9.26).
Here, the main (Y-shaped) component is first inserted between the distal
aorta and one of the iliac arteries. Then, a second (straight) graft segment is
placed in the contra-lateral (opposite) iliac artery and connected to the main
section with additional stents.
Endoluminal grafts have proven to be relatively effective devices, espe-
cially in patients who would not have successfully tolerated surgery. How-
ever, there are several technical difficulties associated with this procedure
that include the need to pass this large object through a potentially diseased
femoral or iliac artery and the need to achieve proper placement of the graft
such that it doesn’t interfere with flow to the renal arteries. The primary
long-term complication experienced with this device has been the develop-
ment of perivascular leaks. These occur because of an incomplete seal
between the graft and the artery, resulting in blood at relatively high pressure
entering the gap between the two vessels. This buildup of blood pressure
could eventually cause a rupture of the aorta very similar to that of an
untreated aneurysm.
7328_C009.fm Page 338 Monday, October 16, 2006 11:05 AM

338 Biofluid Mechanics: The Human Circulation

FIGURE 9.25
The MinTec system for aortic reconstruction. (From Gwertzman, G. (1995) Clinically tested
intravascular stents and endovascular stent grafts, in Endovascular Stented Grafts for the Treatment
of Vascular Diseases, Marin, M.L. Veith F.J. and Levine, B.A. Eds., R.G. Landes Co., Austin, Texas.
With permission.)

9.7 Biomechanics of Stent Implants

The addition of stents to the catheter-based treatment of arterial stenoses has
brought a dramatic improvement in patency rates of these procedures and
has become the method of choice for treating many patients. Since these
procedures are still relatively new, however, there is some uncertainty over
their long-term effectiveness. The main concern involves the possible
responses of the artery wall to the altered fluid and solid dynamic environ-
ment caused by the presence of a small, but exposed, metallic intraluminal
device that is in place for an extended period of time. Part of this response
7328_C009.fm Page 339 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 339

FIGURE 9. 26
Device used in patients with aortoiliac aneurysms with no distal neck. (From Gwertzman, G.
(1995) Clinically tested intravascular stents and endovascular stent grafts, in Endovascular Stented
Grafts for the Treatment of Vascular Diseases, Marin, M.L. Veith F.J. and Levine, B.A. Eds., R.G.
Landes Co., Austin, Texas. With permission.)

is the production of additional support tissue in the form of smooth muscle

cells and fibroblasts, which, in turn, produce elastin and collagen fibers as
well as an extracellular matrix. In some vessels, this response is controlled
and limited in scope, but in others, the process can continue until excess
tissue begins to intrude on the vessel lumen (Figure 9.27).
One implication of the presence of stents in the flow stream is the disrup-
tion of blood flow patterns, especially along the inner surface of the lumen.
Since the stent struts are thin (~0.05 to 0.15 mm), this effect is relatively small
and does not produce gross alterations in the main flow stream. However,
these flow “obstacles” may produce unusually low and reversed shear
stresses on the local level, especially within vortices generated just in front
of and just behind the strut (Figure 9.28), which could contribute to the
development of intimal hyperplasia (see Chapter 6, section 6.5: Wall Shear
Stress and Its Effect on Endothelial Cells).
Another effect is that stents possess a much higher stiffness than the sur-
rounding arterial tissue. This produces a sharp reduction in vessel compliance
in the stented region and also produces impedance mismatches at the
stent–artery interfaces. In addition, stents are typically deployed at very high
pressures (6 atm) and are often over-expanded in order to seat them in place.
This causes the tissue surrounding the stent to be in a permanent state of
elevated wall stress, which may elicit a healing or remodeling response.
One proposed design is to taper the stent stiffness along its length in such
as way as to reduce it from maximum at its center to minimum at its ends
(Figure 9.29).
This compliance-matching stent leads to a more gradual transition of mate-
rial properties and produces a more uniform distribution of circumferential
stress in the artery than earlier designs (Figure 9.30).
7328_C009.fm Page 340 Monday, October 16, 2006 11:05 AM

340 Biofluid Mechanics: The Human Circulation

FIGURE 9.27
Photomicrograph of histological hematoxylin and eosin light microscopic section of atheroscle-
rotic abdominal aorta of a rabbit 8 weeks after a Palmaz stent implantation. (From Becker, G.F.
(1991) Intravascular stents: General principles and status of lower-extremity arterial applica-
tions, in Circulation (Supp. I), 83(2): With permission.)

Flow evaluation of coronary stents

1.2 cm/s

–1E-5 1E-5
(a)

8 cm/s

–0.0001 0
(b)

FIGURE 9.28
Stream lines (m2/s) (for L/D = 3.53, resting conditions, 4 mm vessel) at (a) midsystole and
(b) peak diastolic forward flow. (From Berry, J.L. et al. (2000) Ann. Biomed. Eng. 28: 386–388.
With permission.)
7328_C009.fm Page 341 Monday, October 16, 2006 11:05 AM

Vascular Therapeutic Techniques 341

Transition zone Transition zone

FIGURE 9.29
Illustration of the CMS. The ends of the stent are gradually more flexible toward the ends,
providing a smooth transition in compliance when deployed in an artery. (From Moore, J.E.,
Jr. and Berry, J. L. (2002) Ann. Biomed. Eng. 30: 498–508. With permission.)

0.194 MPa
0.186

0.178
0.170

0.162

0.154
02
0.147
0.139

(a)

8
Circumferential stress (MPa)

CMS
Palmaz
4

0
0 10 20 30 40

–4
Axial position (mm)
(b)

FIGURE 9.30
Color-encoded maximum principal stress in an artery into which the Palmaz stent (top) or the
CMS (bottom) has been deployed. (From Moore, J.E., Jr. and Berry, J.L. (2002) Ann. Biomed. Eng.
30: 498–508. With permission.)
7328_C009.fm Page 342 Monday, October 16, 2006 11:05 AM

342 Biofluid Mechanics: The Human Circulation

Further investigations are currently being made of other design modifica-

tions and also of applying coatings to the stent, which are capable of releasing
various drugs (drug-eluting) that can inhibit IH formation.

9.8 Problems
9.1 A segment of the femoral artery of a dog is replaced with an arterial
graft. The change in radius in response to a change in pulse pressure was
measured in the graft as well as in the femoral artery on the other side
(“contra-lateral”) to yield the following data:

Pulse Pressure Internal Radius Change in Radius Wall Thickness

(mmHg) (mm) (%) (mm)
Artery 50 4.0 15 1.0
Graft 50 5.0 10 1.0

For each of these vessels, compute the compliance and the incremental
elastic modulus at the mean pressure.
9.2 A catheter-tip balloon is designed to expand diseased regions of arteries,
which have a normal (nondiseased) internal diameter of 6 mm. The balloon
has a wall thickness of 0.1 mm and becomes fully distended at 6 mm. It can
withstand up to 6 atm of internal pressure. If the artery can only tolerate a
20% increase over its normal diameter before rupturing, what should the
elastic modulus of the balloon material be to provide full dilatation at
maximum pressure? Is this a maximum or minimum value?
7328_C010.fm Page 343 Monday, October 16, 2006 2:27 PM

10
Fluid Dynamic Measurement Techiques

CONTENTS
10.1 Introduction ..............................................................................................344
10.2 Pressure Measurement............................................................................344
10.3 Blood Flow Measurement ......................................................................347
10.4 Impedance Measurements......................................................................352
10.5 Flow Visualization ...................................................................................356
10.6 Ultrasound Doppler Velocimetry..........................................................359
10.7 Laser Doppler Velocimetry ....................................................................374
10.7.1 General Features ........................................................................375
10.7.2 Probe Volume Specifications ....................................................376
10.7.2.1 Calculation of Probe Volume Dimensions ............376
10.7.2.2 Fringe Patterns...........................................................377
10.7.3 Photodetectors ............................................................................379
10.7.4 Signal Processing .......................................................................379
10.7.5 Phase Window Averaging of LDV Data in Pulsatile
Flow..............................................................................................380
10.8 Magnetic Resonance Imaging and Velocity Mapping
Techniques.................................................................................................381
10.8.1 Slice Excitation ...........................................................................383
10.8.2 Spatial Encoding ........................................................................384
10.8.3 Imaging Procedure and Pulse Sequences ..............................385
10.8.3.1 Spin-Echo....................................................................386
10.8.3.2 Gradient-Echo ............................................................387
10.8.4 Magnetic Resonance Phase Velocity Mapping .....................388
10.9 Computational Fluid Dynamics ............................................................390
10.9.1 Governing Equations ................................................................390
10.9.2 Grid Generation .........................................................................393
10.9.3 Discretization Techniques.........................................................396
10.9.3.1 Finite Difference Method .........................................396
10.9.3.2 Temporal Integration ................................................398
10.9.4 Computational Modeling of Blood Flows .............................399
10.9.4.1 Body Motion ..............................................................399
10.9.4.2 Scale Disparity ...........................................................400
10.9.4.3 Blood Flow Properties ..............................................400

343
7328_C010.fm Page 344 Monday, October 16, 2006 2:27 PM

344 Biofluid Mechanics: The Human Circulation

10.9.4.4 Pulsatility and Reynolds Number..........................400

10.9.4.5 Symmetry ...................................................................401
10.9.4.6 Grid Resolution .........................................................401
10.9.4.7 Validation....................................................................402
10.9.5 Potential for Future Numerical Techniques ..........................402

10.1 Introduction
We have now used models for steady (Chapter 5) and unsteady (Chapter 6)
flow through rigid or elastic tubes to understand the flow dynamics in the
blood vessels. Analysis of the governing equations resulted in expressions
for pressure pulse, flow rate, and velocity profiles. To verify the accuracy of
our models, data on these variables must be obtained from in vitro or, pre-
ferably, from in vivo measurements. Moreover, measurements of pressure,
flow rate, and resistance within specific segments of the circulatory system
are also important diagnostically to the physician. Measurement of detailed
velocity profiles and other flow parameters will also be helpful in under-
standing the factors involved in the initiation of disease processes, such as
thrombus formation and atherosclerosis (see Chapter 6).

10.2 Pressure Measurement

Earlier, we discussed the pressure pulse in the arteries and the systolic and
diastolic pressures. These values are again an important measurement for
the physician to have for diagnostic purposes. An indirect method of pres-
sure measurement can be performed with the familiar pressure cuff method.
In this method, using a sphygmomanometer (sphygmos means pulse), a
pneumatic cuff encircling the upper arm is inflated to a pressure larger than
the blood pressure. Thus, the brachial artery in the arm collapses and
occludes the flow of blood through the artery. As the cuff pressure is slowly
released and when the cuff pressure decreases to a value slightly lower than
the peak systolic blood pressure, the blood squirts through the collapsed
segment of the artery. The flow through the segment is turbulent and gen-
erates a sound referred to as the ‘‘Korotkoff’’ sound, which can be detected
by a stethoscope placed over the brachial artery. The pressure at the initia-
tion of this sound is the systolic pressure. As the cuff pressure is further
decreased, the Korotkoff sound ceases to exist at a point when there is no
constriction of the artery and, hence, flow is no longer turbulent. The pres-
sure corresponding to the cessation of the sound is the diastolic pressure.
7328_C010.fm Page 345 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 345

By replacing the stethoscope with a microphone to detect the sound, the

pressure measurement technique can be automated. The technique is non-
invasive and pressure can be measured with minimal discomfort to the
patient. However, the error in measurement with this technique can be as
high as 10 mmHg.
For more accurate determination of blood pressure, direct methods that
require gaining access to the blood vessel through catheterization is required.
Usually, a fluid-filled catheter is introduced into an artery and is connected
to a pressure transducer where the fluid comes in contact with the transducing
element. The pressure transducers described above work on the principle
that the resistance of that element changes in proportion to the pressure
applied on the element. This resistance change is converted into an electrical
output signal. For a bridge containing four active elements, the output volt-
age can be shown to be:

∆R
e0 = E
R

Commercial transducers have a transducer sensitivity factor provided, and

the sensitivity factor F is given as volts output per volt of excitation per unit
of the physical quantity measured. Thus, the output voltage e0 is given by

e0 = FEQ (10.1)

where E is the excitation voltage and Q is the quantity being transduced,

such as pressure. Thus, the measurement of any physical quantity depends
upon the accuracy of the sensitivity factor, the accuracy of the excitation
voltage, and the accuracy of the measurement of the output voltage. Many
transducers also have a calibration resistor incorporated in the circuit.
The fluid-filled catheter system is commonly used in the clinical setting
for continuous monitoring of the patient’s blood pressure. However, the
catheter-tipped transducer, which avoids the fluid-filled flexible tubing, has
a higher frequency response and is preferred in a research setting where an
accurate reproduction of the pressure pulse is important. The fluid-filled
catheter is filled with heparinized saline to prevent clotting of the blood in
the catheter, and periodic flushing of the catheter is necessary to prevent
blood coagulation. Also, care must be taken to prevent trapped air bubbles
in the catheter because the air bubbles will distort the pressure pulse due to
damping. Figure 10.1 compares the pressure signals obtained with a fluid-
filled catheter (with and without air bubbles) with pressure signals obtained
from a catheter-tipped transducer. When the fluid-filled catheter is intro-
duced in a large-diameter blood vessel, due to the flailing motion of the
catheter, low frequency oscillations are superimposed on the pressure signal
(Figure 10.2).
7328_C010.fm Page 346 Monday, October 16, 2006 2:27 PM

346 Biofluid Mechanics: The Human Circulation

ECG II

1 mv

–175 mm Hg

Strain-gauge –100 mm Hg
6 Fr catheter
–175 mm Hg

Catheter-tip –100 mm Hg
transducer
No bubble in transducer

1 second

(a)

ECG II
1 mv

–175 mm Hg

Strain-gauge –100 mm Hg
42 milliseconds
6 Fr catheter
–175 mm Hg

–100 mm Hg
Catheter-tip
transducer
Bubble in transducer

1 second

(b)

FIGURE 10.1
Comparison of pressure signals recorded with fluid-filled catheter transducer and catheter-
tipped transducer. Note the similarity of the signals between the two transducers in the absence
of air bubble in the catheter. Changes in pulse shape and a time delay are observed with the
presence of the air bubble with the fluid-filled catheter. (From Geddes, L.A. (1984) Cardiovascular
Devices and Applications, John Wiley & Sons, New York. With permission.)
7328_C010.fm Page 347 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 347

(a)

(b)

FIGURE 10.2
Distorted pressure signals in a fluid-filled catheter due to catheter whip: (a) good quality
pressure recording and (b) the distorted signal due to catheter whip. (From Geddes, L. A. (1984)
Cardiovascular Devices and Applications, John Wiley & Sons, New York. With permission.)

10.3 Blood Flow Measurement

Ultimately, the delivery of essential nutrients and oxygen to tissues and the
removal of waste products is the overriding purpose of the circulation.
Therefore, measurement of the volume rate of blood flow through the heart
and vessels is very important in assessing the function of the system. The
electromagnetic flow (EMF) meter is one device currently used for such
measurements. It operates by determining the mean velocity of flow through
a blood vessel of a known cross-sectional area and then derives the flow rate
through the vessel as the product of these two values. This measurement
technique theory is based on Faraday’s law. According to this law, when a
conductive fluid, such as blood, flows between the lines of force of a magnetic
field, an electromagnetic force is generated in the fluid, which is perpendicular
to the direction of the magnetic field as well as the direction of motion of the
fluid (Figure 10.3). The voltage generated between the electrodes, Ef (volts), is
given by the relationship

E f = B
V

where B is the flux density of the magnetic field (Webers/m2),
is the spacing
between electrodes (m) and V is the mean flow velocity (m/s). Since the flow
rate through the vessel, Q (m3/s), is related to the area of cross section A
(m2) and the mean velocity by

Q = VA

it can be computed from the relationship

A
Q= E (10.2)
Bl f
A
The ratio B

is a constant specified for a given flow meter system.
7328_C010.fm Page 348 Monday, October 16, 2006 2:27 PM

348 Biofluid Mechanics: The Human Circulation

Flow meter

Range

Zero

Signal voltage
Magnet current

Electromagnet

Magnetic field

Flow

Signal sensing Blood vessel

electrodes

FIGURE 10.3
Schematic of an electromagnetic flow probe used for blood flow measurement.

The electromagnetic flow probe, which is attached to the blood vessel,

consists of an electromagnet that generates the magnetic force and two
electrodes to detect the flow signal. These are encapsulated in probes of
varying sizes to adapt to the various vessels in the circulation system. The
intracorporeal probe shown in Figure 10.4 is inserted by cannulation of the
blood vessel or is used for flow measurements in extracorporeal circulation.
The intracorporeal probe has an opening slot so that it can be fit snugly
around the vessel as shown in Figure 10.5.
The relationship derived in Equation 10.2 assumes a DC (direct current)
magnetic field, which would produce a DC flow signal. However, problems
from electrode offset potentials, amplifier drift, etc. make it difficult to delin-
eate the actual flow signal, and AC (alternating current) excitation is used
in the common flow meters. The flow meters use either a square wave
excitation or sine wave excitation. In AC excitation, the time variation of the
magnetic flux, dB/dt, must be included in Equation 10.2. In the commercial
systems, measurements are obtained when the time varying component is
zero. In the sine wave excitation, the dB/dt component is effectively zero
7328_C010.fm Page 349 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 349

FIGURE 10.4
Typical electromagnetic flow probes: (a) Extracorporeal. (Courtesy of Carolina Medical Electronics
Inc., King, NC.)

only for a relatively shorter time and the voltage-sensing circuits must be
gated for a short time. On the other hand, with square wave excitation, the
time varying component is effectively zero, except during the short periods
in which the signals are switched. The signal-to-noise ratio is proportional
to the peak-to-peak value of excitation, and for the same signal-to-noise ratio,
the square wave excitation requires twice as much power as the sine wave
excitation. Hence, the meters for the square wave devices are larger and
must operate at higher temperatures. Shown in Figure 10.6 is a typical flow
rate signal from an in vitro pulse duplicator distal to a heart valve. Measure-
ments were made with a Carolina square wave electromagnetic flow meter
and an in vivo metric extracorporeal flow probe. The time-averaged flow rate
(cardiac output) and other information, such as the amount of back flow,
can be determined from these curves.
Even though the electromagnetic flow meter is an ideal device for in vitro
testing, there are several drawbacks in using it for in vivo measurements.
One is that this is an invasive technique in which the vessel must be exposed
in order to fit the flow probe around the vessel. Secondly, if left in the body
for an extended length of time, problems of contact of the electrodes with
the arterial wall, such as protein or thrombus coating and subsequent dete-
rioration of the signals, may occur. Moreover, this device actually measures
the mean flow velocity and, thus, the area of the probe lumen must be accu-
rately known to determine the flow rate. The EMF theory also assumes that
the fluid has a flat velocity profile at the mean velocity magnitude. Since in
7328_C010.fm Page 350 Monday, October 16, 2006 2:27 PM

350 Biofluid Mechanics: The Human Circulation

FIGURE 10.5
Typical electromagnetic flow probes: (b) Intracorporeal. (Courtesy of Carolina Medical Electronics
Inc., King, NC.)

blood vessels the lumen is not an ideal circular cross section and the flow is
asymmetric, some errors will occur in measuring the flow rate using this
technique.
A more recent flow-measuring device is the Transit Time Flow Meter.
This instrument is based on the fact that sound waves travel through a
fluid at slightly greater or lower velocities depending upon whether the
flow is forward or reverse, respectively. The transit time flow meter utilizes
ultrasonic energy in the kilohertz range (as opposed to ultrasound imaging
7328_C010.fm Page 351 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 351

400

300

200
Flow rate (ml/sec)
100

–100

–200

–300
0 0.2 0.4 0.6 0.8 1
Time (sec)

FIGURE 10.6
A typical flow rate signal distal to a heart valve from an electromagnetic flow meter in a pulse
duplicator.

and Doppler devices that operate in the megahertz range), which is trans-
mitted through a vessel along the axis of the flow in alternate directions
(Figure 10.7). The device then detects small differences (in the order of
nanoseconds) in transit times between the two signals, tf and tr, as

tf =
(c + u)

tr =
(c − u)

T, R R, T

u d
θ

FIGURE 10.7
Transit-time flow probe with two ultrasonic crystals alternately transmitting and receiving
signals passing through a fluid and reflecting off of a backplate.
7328_C010.fm Page 352 Monday, October 16, 2006 2:27 PM

352 Biofluid Mechanics: The Human Circulation

where

= 2d (sin θ)

It then computes the flow velocity, u, from

c 2 ∆t
u=
(2 d cos θ)

where ∆t = tr − t f .
The volume flow rate is determined by multiplying this velocity by the
known vessel’s cross-sectional area. While this device is also invasive
(i.e., the probe must be placed around the vessel), it does not depend upon
any special features of the vessel (e.g., electrical conductivity) other than the
presence of reflectors in the form of red blood cells. Since the flow rate is
proportional to the transit time of the sound signals, it is a very linear and
easily calibrated device, which is very stable (i.e., not susceptible to other
electrical signals) and which can be left in place for long-term (i.e., days to
weeks) experiments.

10.4 Impedance Measurements

In developing the steady flow models for blood flow, we discussed the
resistance to blood circulation as being given by the relationship

∆p
R=
Q

When measuring the resistance across the systemic circulation, the mean
pressures at the aorta and vena cava are used along with the mean flow rate
(cardiac output). Using a familiar electrical analogy, the resistance in DC
circuits is given by R = E/I, where E is the voltage and I is the current in a
branch of the circuit. In AC circuits, the corresponding measure is the elec-
trical impedance. Thus, impedance to flow in the circulation under unsteady
flow can be defined as similar to resistance under steady flow conditions.
Milnor (1989) defines three different impedance relationships as

Longitudinal impedance:

− dp
ZL = dz
(10.3)
Q
7328_C010.fm Page 353 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 353

Input impedance:

p
Zz = (10.4)
Q

Transverse impedance:

p
ZW = − dQ (10.5)
dz

Longitudinal impedance is the ratio of the pressure gradient to the flow

and, therefore, is directly analogous to the vascular resistance defined in
steady flow. The longitudinal impedance depends only on the local proper-
ties of the vessel across which the pressure gradient is measured. Input
impedance, on the other hand, is the ratio of the pressure and flow at a
particular site in the vasculature and is dependent on the local properties as
well as those in the segments distal to it. The input impedance at the end of
the vessel is called the terminal impedance, Zt. If the distal bed were just a
continuation of the vessel in which measurements are made (i.e., without
any change in characteristics), then this value is termed the characteristic
impedance, Zo.
From the theoretical models developed for pulsatile blood flow in elastic
tubes by Womersley and others, the impedances can be computed from the
relationships given above. The impedances will be complex quantities, which
depend upon the Womersley parameter as well as the elastic properties of
the tube. Thus, the complex impedance can be resolved into real and imag-
inary quantities where the real part is the resistive component and the
imaginary component is the reactive component of the impedance. The
computed results can also be presented as the modulus and phase lag.
Several models have been used to theoretically calculate the input or char-
acteristic as well as the longitudinal impedance (Milnor, 1989).
To determine the input impedance experimentally, the pressure, as well as
the flow, is measured simultaneously at the selected point in the circulation.
The pressure can be measured with high fidelity pressure transducers dis-
cussed earlier and the flow can be measured by an electromagnetic flow
meter or by ultrasound methods. The frequency response of the flow and
pressure transducers must be determined and the distortion of modulus and
phase must be taken into account in computing the impedance (O’Rourke,
1982). The signals are then subjected to a frequency analysis and the input
impedance at any frequency component is the ratio of the pressure and the
flow at that frequency. Thus, at a frequency, n, the input impedance will be
given by

pn
Zz n = (10.6a)
Qn
7328_C010.fm Page 354 Monday, October 16, 2006 2:27 PM

354 Biofluid Mechanics: The Human Circulation

and

θn = φ n − ξn (10.6b)

where φn is the phase angle of the pressure and ξn is the phase angle of the
flow. Frequency analysis, or decomposition, is usually performed by Fourier
analysis, where the phase and amplitude of each frequency component is
obtained. A basic assumption of this technique is that the system is linear
so that there is no interaction between signals of different frequency. A
limitation of the characteristic impedance is that it cannot be directly mea-
sured in vivo. However, in larger arteries, the input impedance is essentially
independent of the frequency except for a drop in the impedance at low
frequencies. Thus, the characteristic impedance, Zo, is computed by averag-
ing the modulus of the input impedance across the frequency spectrum,
neglecting the low frequencies where the steep drop in impedance occurs.
The input impedance in the ascending aorta for a dog and a human is
given in Figure 10.8, where it is seen to fall steeply from 0 to 2 Hz and then
fluctuate with frequency thereafter. Such an input impedance spectrum usu-
ally shows a minimum of the impedance modulus between 2 and 8 Hz. The
phase angle of the input impedance is usually zero at the first minimum of
the modulus. Oscillations in the modulus of the impedance depend upon
reflections of the waves in the peripheral arteries as they encounter discon-
tinuities (i.e., branches, end-organs, etc.), and vasoconstriction will increase

1400
1.8
Modulus (103 dyn sec cm–5)

Modulus (dyn sec cm–5)

1.6 1200
1.4
1.2 300
1.0
0.8 200
0.6
0.4 100
0.2 z0
0 0
2 4 6 8 10 12 2 4 6 8 10
Frequency (Hz) Frequency (Hz)
2
Phase (rad)
Phase (rad)

0 0
–1
–2

(a) (b)

FIGURE 10.8
Input impedance in the aorta of dog (A) and human (B). (Redrawn from Milnor, W.R. (1989)
Hemodynamics, Williams and Wilkins, Baltimore, MD. With permission.)
7328_C010.fm Page 355 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 355

400

Modulus (dyn sec cm–5)

300

200

100

F (Hz)
0
Phase (rad)

0.5 2 4 6 8 10

–0.5

FIGURE 10.9
Input impedance in the main pulmonary circulation of a dog (Redrawn from Milnor, W.R.
(1989) Hemodynamics, Williams and Wilkins. Baltimore, MD. With permission.)

the reflections and, hence, increase the oscillations in the impedance. The
ratio of Zz/Z0 and also the ratio of

Zz (max) − Zz (min)
Z0

are useful measures of the oscillations in the impedance. In this ratio, the
first maximum and first minimum of the modulus are used. Under normal
resting conditions, this ratio is about 0.6 to 0.85 in man and dogs.
The input impedance in the main pulmonary circulation is shown in
Figure 10.9. As can be observed, the shape of the profile is similar to that for
the aorta even though the magnitude is smaller. The first minimum of the
modulus occurs between 2 and 4 Hz and the next maximum occurs at 6 to
8 Hz. The characteristic impedance in pulmonary circulation is obtained
by averaging the impedance modulus at higher frequencies and it is about
190 dyne-sec/cm5 in dogs and about 23 dyne-sec/cm5 in man.
As mentioned earlier, the impedance will change due to alterations in the
vascular tone as well as changes in the elastic properties of the distal vessels.
Thus, measurements of vascular impedance can be used for diagnostic pur-
poses to determine changes in the vasculature in the distal vessels. For
example, pulmonary vascular hypertension increases the vascular imped-
ance to as much as three times the normal value and, as a consequence, the
7328_C010.fm Page 356 Monday, October 16, 2006 2:27 PM

356 Biofluid Mechanics: The Human Circulation

minimum of the modulus is displaced from 3 to nearly 8 Hz. Systemic

vascular hypertension is also known to increase the input impedance from
normal values in a similar way. Further discussion on the impedance char-
acteristics with arterial disease can be found in Milnor (1989).

10.5 Flow Visualization

A useful technique for obtaining visual information about the flow charac-
teristics in various cardiovascular anatomies and devices is flow visualiza-
tion. Flow visualization, widely used in the automotive and aircraft
industries, is performed by inserting a visible marker into the flow stream
and then making a photographic record of the marker’s movement. With
this technique, it is possible to obtain a general overview of the flow field
as well as to identify fine details of flow structures, such as jets, separation
zones, secondary motions, etc., as well as providing an indication of the
stability of the flow (i.e., laminar vs. turbulent). This information can be
obtained relatively simply and cheaply and serves as a good starting point
for designing subsequent experiments for obtaining more quantitative data.
The key assumption of a flow visualization technique is that the marker
accurately follows the actual flow movement. To do so, there must be a
balance of the inertial, viscous, buoyant, and gravitational forces acting on
the marker and it must not disturb the flow. The best way to satisfy these
requirements is to use small, neutral, density particles. Advantages of flow
visualization over other techniques are its simplicity and ability to provide
full-field (i.e., 2-D) information at any time. Disadvantages are its require-
ment for optically transparent fluids and materials and, traditionally, its
nonquantitative nature, although newer advances with Particle Image
Velocimetry (PIV) now make it possible to derive quantitative information
from flow visualization data as well.
Most flow visualization techniques provide information in the form of
either pathlines or streaklines. Pathlines represent the actual trajectory or
path followed by a single particle in the flow field while streaklines show
the present positions of all particles that have passed through some common
point. A related concept is that of “streamlines,” which are curves superim-
posed on the flow field that are tangent to the velocity vectors of particles
at a particular position and at a specific instant in time. In some cases,
timelines may also be obtained, which show the movement of a region of
flow that occurs between successive time intervals.
Depending upon the application and the nature of data desired, there are a
variety of flow visualization methods used that are classified by the type of flow
marker employed. For example, a colored dye is an effective way of obtaining
streaklines in a flow system (Figure 10.10). This method can be performed with
virtually any visible dye (e.g., ink, food coloring, fluorescein, etc.), which
7328_C010.fm Page 357 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 357

Constant infusion pump

for near wall flow
visualization

Internal carotid (INT)

Common carotid (COM)
steady or pulsatile flow
from constant head tank
Bifu
ang rcatio
le ≈ n
40°
External
carotid
(EXT)

Fine needle injection

into selected core
streamlines

FIGURE 10.10
Schematic of the experimental setup for flow visualization using colored dye. LoGerfo et al.
(1985) Structural details of boundary layer separation in a model human carotid bifurcation
under steady and pulsatile flow conditions, in J. Vasc. Surg., 2: 263–269.

dissolves in the fluid medium (usually an aqueous solution). The dye can then
be injected at a specific location, generally using a fine needle, and then its path
tracked using photographs or video recordings. One disadvantage of dye flow
visualization is the eventual buildup of the dye in the fluid, which then clouds
the visual field. Another is that it is not useful for either pulsatile or turbulent
conditions. Solid particles (e.g., aluminum flakes or microspheres) are often used
to obtain pathlines, especially if either the light source or the camera shutter is
strobed (Figure 10.11). The particles must be chosen so that they are good
reflectors of light, have densities similar to that of the fluid medium, and are
small in comparison to the finest flow structure being imaged.

FIGURE 10.11
An example of flow visualization using solid particles. (From Lutz, R.J. et al. (1983), Comparison
of steady and pulsatile flow in a double branching arterial model, in J. Biomech. 16: 753–766,
With permission.)
7328_C010.fm Page 358 Monday, October 16, 2006 2:27 PM

358 Biofluid Mechanics: The Human Circulation

Doppler
Nitrogen laser frequency
analyzer

Pulsatile pump
Test section

35 mm
camera

Microcomputer

(a)

Video terminal

Input from pump

group cycle detect electronics
Disk for
data storage
CROMEMCO Z-2D
computer system
Serial data link Spectrum
to PDP 11/34 analyzer

From doppler
system

Camera Extra S-100 boards

and flash within include. Laser interface
unit
interface unit • Programmable
timer board
• Laser, camera,
flash and pump
To camera interface board To laser
and flash
(b)

FIGURE 10.12
Schematic of the experimental setup for flow visualization using a photochromic dye technique.
(From Poots, K. et al., (1986), A new pulsatile flow visualization method using a photochromatic
dye with applications to Doppler ultrasound, in Ann. Biomed. Eng. 14: 203–218, With permission.)
7328_C010.fm Page 359 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 359

A technique sometimes used in an aqueous fluid medium is called

Hydrogen Bubble flow visualization (see Chapter 5, section 5.6: Flow through
Arterial Stenoses and Aneurysms). This technique actually generates flow
markers within the fluid by the process of electrolysis, thereby creating H2
and O2 molecules at a cathode and anode, respectively. Advantages of this
method are that no external markers need to be added to the fluid and there
is no buildup, such as with dye. Disadvantages of the technique are that
electrodes must be introduced into the fluid, which must be an electrical
conductor, and that the density of the gas particles generated is much less
than that of the fluid medium. Despite these drawbacks, however, this tech-
nique may be adequate for cases where high flow velocities exist and the
transit time of the gas bubbles through the imaging field is brief. (Note:
Hydrogen bubble flow visualization was the technique used in earlier images
of symmetric, carotid stenosis models.)
A fairly ingenious flow visualization approach is the use of a transparent
compound that is mixed into the fluid medium and then can be selectively
converted to a colored state upon activation by laser light illumination
(Figure 10.12). This Photochromic Dye technique utilizes the chemical TNSB,
1′, 3′, 3′-trimethyl-6-nitroindolene-2-spiro-2-benzopyran which has the property
of becoming opaque when exposed to the ultraviolet light emitted from a
nitrogen laser. Images of the movement of regions of the flow rendered visible
by the laser beam activation can then be recorded by standard photographic
methods. Advantages of this technique are that the laser beam can be posi-
tioned at any site in the flow field, the chemical reaction is completely revers-
ible with a half-time of only a few seconds (i.e., there is no accumulation of
dye) and the dye is clearly seen and recorded. The major disadvantage of this
technique is that TNSB is not water-soluble and must instead be dissolved in
a hydrocarbon-based fluid, such as kerosene, a flammable and toxic substance.
Since each of the above is an optical technique, it is critical to provide a
transparent, distortion-free system for making recordings. A common
approach is to place the test model within a flat surfaced view box filled
with a fluid having a refractive index similar to that of the model construction
material. In this way, light penetrating through the viewing area will follow
a straight path and not be diffracted by irregular or curved surfaces.

10.6 Ultrasound Doppler Velocimetry

While several of the previous techniques are invaluable for providing impor-
tant qualitative and quantitative information about flow conditions within
in vitro models, there are some cases where they cannot be used due to their
optical nature and their requirement for transparent models and fluids.
Examples of this would be testing of an actual prototype device that is
opaque or translucent, use of blood as a test fluid, or testing in an in vivo
7328_C010.fm Page 360 Monday, October 16, 2006 2:27 PM

360 Biofluid Mechanics: The Human Circulation

(a) S R

(b) S R VR

(c) S R VR

FIGURE 10.13
Doppler effect caused by a moving receiver. Atkinson, P. and Woodcock, J. (1982) Doppler
Ultrasound and Its Use in Clinical Measurement, Academic Press, London.

animal model. An approach that can be used to make velocity measurements

in these cases is ultrasound Doppler velocimetry. Here, the term “ultrasound”
denotes frequencies that are much higher than those of audible sound (i.e.,
>20,000 Hz).
Ultrasound Doppler velocimetry is based on a phenomenon first described
by Christian Doppler of Austria in 1842. It was his observation that the
apparent tone of a sound changes if the source and the listener are moving
relative to each other. He demonstrated this with a musician riding on a
train car that passed a second musician standing along the tracks. As the car
approached and then departed, the second musician recorded the perceived
tone, or note, of the sound. The result was an apparent increase in tone as
the car approached and a decrease in tone on departure. This “Doppler-shift”
effect can be described mathematically to show that the change in frequency
is proportional to the relative speed between source and observer.
Specifically, if a source emits a sound at a frequency, fs, and a receiver is
moving at a velocity, Vr , relative to the source (Figure 10.13), then the number
of wave peaks received per unit time will equal the number of peaks trans-
mitted from the source plus the extra number of peaks intercepted due to
the advancing position of the receiver, or

Nr = Ns + ∆N (10.7)

Since the number of wave peaks in a given time interval, ∆t, would be
equal to the corresponding frequency times ∆t and the additional peaks
would be the distance traveled divided by the wavelength, λ,

Vr ( ∆t)
fr ( ∆t) = fs ( ∆t) + (10.8)
λs
7328_C010.fm Page 361 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 361

Therefore,

Vr
fr = fs + (10.9)
λs

Rewriting the wavelength, λs, in terms of its frequency, fs, and the speed
of sound in the medium, c (= 1540 m/s for water and 1560 m/s for blood)

c
λs = (10.10)
fs

we can now solve for the frequency difference, fd

Vr  Vr 
fd = fr − fs = = f (10.11)
λ s  c  s

The same analysis can be carried out for the case of a moving source,
yielding

V
fd =   fs (10.12)
 c

Both of these equations show that the amount of frequency shift caused
by the motion between the source and receiver is directly proportional to
the relative velocity between them.
In practice, most fluid velocity measurements are made using a sound
wave, which is transmitted to an object and then echoed back to the source
where it is received. Examples of this would be a particle reflector in the
fluid medium (in vitro) or a red blood cell in the blood (in vivo). In this
configuration, the frequency-shift, ∆f, becomes

2Vfs
∆f = (10.13)
c

Finally, ultrasound Doppler transducers are usually not placed directly in

the flow stream in order to avoid either disturbing the flow (in vitro) or
becoming invasive to the body (in vivo). Thus, they are generally positioned
outside the vessel and directed at an angle, θ, to the flow axis. In this case,
the velocity detected represents only a component of its axial magnitude and,
thus, a modified Doppler shift equation is used.

2Vfs
∆f = cos θ (10.14)
c
7328_C010.fm Page 362 Monday, October 16, 2006 2:27 PM

362 Biofluid Mechanics: The Human Circulation

To obtain the axial flow velocity, the Doppler shift equation is simply
rewritten in terms of V

∆f c
V= (10.15)
fs 2 cos θ

or

V ∆f
= (10.16)
c fs 2 cos θ

In this last expression, we can see the direct proportionality between the
ratio of reflector velocity to sound wave speed and the ratio of frequency
shift to source frequency. Also note in Equation 10.14 that a forward velocity
will produce an up-shifted frequency, while a reverse velocity will produce
a down-shifted frequency; thus, the device has directional capability.
In order to obtain the flow velocity from the Doppler shift waveform, the
received signal must be processed for its frequency content and, in particular,
for the frequency shift information. This outcome is accomplished in three
stages. The first step, called Phase Quadrature Demodulation (Figure 10.14),
extracts the frequency shift of the received signal from the frequency of the
transmitted signal by mixing the two. This process produces a combination of
signals at both summed and differenced frequencies. Those that are summed
are all in the range of fs , while those that are differenced are in the range of ∆f.

Transmitting Master π/2 phase

amplifier oscillator shifter
(+gate)

Receiving Coherent Coherent

Transmitting amplifier demodulator demodulator
transducer

Receiving
transducer
Direct channel Quadrature channel
output output

Vf
Vr

FIGURE 10.14
Phase-quadrature demodulation. Atkinson, P. and Woodcock, J. (1982) Doppler Ultrasound and
Its Use in Clinical Measurement, Academic Press, London.
7328_C010.fm Page 363 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 363

D Q D Q D Q

π/2 π/2 Sin wp f Cos wp f

Logic phase phase Multiplier Multiplier
Switch
unit shifter shifter
Ds Qc
Dπ/2 Qπ/2
To To
forward reverse
flow flow Summing Summing
channel channel amplifiers amplifer
Dπ/2 + Q D + Qπ/2
Reverse flow Forward flow Forward (wp + wf ) and
component wr component wf reverse (wp – wr) flow
components
(a) (b) (c)

FIGURE 10.15
Directional processing in the time domain. Atkinson, P. and Woodcock, J. (1982) Doppler
Ultrasound and Its Use in Clinical Measurement, Academic Press, London.

By running all signals through a low-pass filter (e.g., one with a cut-off
frequency ≥2∆f ), only those signals with Doppler-shifted frequencies will
remain. The second step, called Phase Domain Demodulation (Figure 10.15),
separates out those signals that were up-shifted in frequency (i.e., forward
velocity) from those that were down-shifted in frequency (i.e., reverse velocity).
The final step involves actually quantifying the frequencies of those analog
signals. This can be done using one of two common signal-processing methods —
either zero crossing detection or Fourier transformation.
Zero-crossing detection [ZCD] or counting [ZCC] simply counts the num-
ber of times an oscillatory signal crosses the zero baseline (Figure 10.16).
Thus, a signal with high frequency has a larger count than a signal with low
frequency over a fixed period of time. This is a simple and inexpensive
approach, which can be built into any ultrasound Doppler device. However,
it doesn’t give a true mean frequency, as its output is actually proportional
to the Root Mean Square (rms) value of the signal. An alternate approach to
the ZCD/ZCC is to perform a fast Fourier transformation [FFT] on the signal
(Figure 10.17). This process converts the signal from a time domain to a
frequency domain representation in terms of its harmonics. By doing this, the
relative magnitude of each harmonic is determined and a resultant spectrum
of frequencies can be obtained (Figure 10.18). The result is that the frequency
shift (α particle velocity) and the power of the signal (α number of particle
reflectors present) can be obtained over time.
While all ultrasound Doppler velocimeters are based on the same funda-
mental principles and utilize virtually the same signal processing techniques,
7328_C010.fm Page 364 Monday, October 16, 2006 2:27 PM

364 Biofluid Mechanics: The Human Circulation

Doppler signal Positive-going

threshold
Negative-going
threshold
Zero crossing output

Output
pulses

Set Reset S R S R S R

Time

FIGURE 10.16
SET–RESET zero-crossing counter. Atkinson, P. and Woodcock, J. (1982) Doppler Ultrasound and
Its Use in Clinical Measurement, Academic Press, London.

they are typically constructed in two different configurations or modes:

(1) Continuous-wave Doppler devices and (2) pulse Doppler devices.
Continuous-wave (CW) devices (Figure 10.19) operate by transmitting a high
frequency (generally >1 MHz) sound wave from a crystal and then collecting
the returned echo from a moving object with a second crystal. Because the
crystals operate independently, each one either continuously transmits or
continuously receives signals, which are then compared to determine the
instantaneous frequency shifts and, thus, velocities of the target objects. Since
the transmit and receive sound beams virtually overlap with each other, the

Doppler signal Anti-aliasing

input filters
(analogue)

Internal timing
and logic unit
Synchronous
ADC

p( fi)
Input buffer Processor Display Spectrum
memory memory generator Display
fi outputs

FFT Output buffer

processer memory

FIGURE 10.17
The fast Fourier transformation (FFT) analyzer. Atkinson, P. and Woodcock, J. (1982) Doppler
Ultrasound and Its Use in Clinical Measurement, Academic Press, London.
7328_C010.fm Page 365 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 365

Power

uency
Freq

Tim
e

FIGURE 10.18
Three-dimensional plot of time-variable spectrum. Hatte, L. and Angelsen, B. Doppler Ultrasound
in Cardiology: Physical Principles and Clinical Applications. (1982) Lea & Febiger, Phileadelphia.

Transmitting Master
amplifier oscillator

Transmitting
transducer Receiving
Demodulator
amplifier
Receiving
Ultrasonic transducer
beam

Doppler difference
signal

Moving targets
(blood) Blood vessel

FIGURE 10.19
The continuous-wave flow meter. Atkinson, P. and Woodcock, J. (1982) Doppler Ultrasound
and Its Use in Clinical Measurement, Academic Press, London.
7328_C010.fm Page 366 Monday, October 16, 2006 2:27 PM

366 Biofluid Mechanics: The Human Circulation

CW Doppler is characterized by having a relatively large sample volume,

i.e., region within which motion can be detected. Thus, it is generally easy to
obtain a clear recording from the vessel, but with no information as to the
specific location of that signal within the entire cross section. Furthermore,
since the velocities from many particles are being detected simultaneously,
the resultant signal contains a wide range or spectrum of shifted frequencies.
Thus, in practice, the signal is usually averaged or processed by simply
identifying the maximum velocity for use.
These recordings have proven valuable, especially for clinical diagnostic
purposes, because of the ability of ultrasound to readily penetrate soft
tissues. We mentioned earlier (Chapter 5, section 5.2: Application of the
Bernoulli Equation), how important pressure measurements may be indi-
rectly obtained through the use of blood flow velocity measurements. In
valvular stenosis, for example, an obstruction to blood flow is provided by
a valve that does not open to its full extent. Such an obstruction results in
an elevated pressure gradient across the valve, thereby placing additional
pumping requirements on the heart, which can result in hypertrophy. Since
the systemic pressure is essentially independent of valvular disorders, the
elevated pressure is reflected mainly in the proximal chamber (such as the
left ventricle in aortic stenosis) and can lead to regurgitation in associated
valves (such as the mitral valve). Therefore, it is the pressure gradient across
the valve that best characterizes the severity of the lesion. Traditionally, this
gradient was measured invasively by catheterization techniques. How-
ever, since conservation of energy requirements state that the pressure drop
(or loss of potential energy) is accompanied by an increase in velocity (or
kinetic energy), elevated velocities can be used to calculate decreased pres-
sures. The physiologic impact of the stenosis is most accurately assessed by
the maximum pressure drop or the maximum velocity. Since this maximum
velocity may occur at the level of the leaflets or distal to them — if a
significant vena contracta exists — a technique with which maximum velocity
could be obtained regardless of its location along the entire path of blood
would be most effective in obtaining this clinically relevant quantity.
Continuous-wave Doppler fits such specifications perfectly. Because of the
continuous emission/reception of signals, the locations of the velocity measure-
ments cannot be defined due to lack of spatial resolution. Therefore, for a given
continuous-wave spectrum, velocities present within the entire range of the
ultrasound beam are displayed. Consequently, by reading the peak signal dis-
played at a given temporal location on the spectrum, one automatically obtains
the maximum velocity, regardless of where it occurs along the ultrasound beam.
Clinically, then, if the continuous-wave beam is passed down the ascending
aorta and through the barrel of the stenotic aortic valve, for example, the
maximum systolic velocity is immediately obtained and can be converted to
the maximum pressure gradient by use of Equation 1.54. Figure 10.20 shows a
continuous-wave spectrum of a patient with mitral valve stenosis.
Techniques developed to replace catheterization in the assessment of
stenosis using continuous-wave Doppler have been quite successful over
7328_C010.fm Page 367 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 367

2 –m/s

FIGURE 10.20
Continuous-wave spectrum of patient with mitral valve stenosis. This patient has a peak
mitral valve ejection velocity of 2 m/s, which is quickly and easily obtained by continuous-
wave Doppler. The velocity then can be converted into a pressure gradient of 16 mm Hg
via the simplified Bernoulli equation to provide an estimate of the severity of the stenosis.
Traditionally, such a pressure gradient would have been measured painstakingly by cathe-
terization. Hatte, L. and Angelsen, B. Doppler Ultrasound in Cardiology Physical Principles and
Clinical Applications. (1982) Lea & Febiger, Philadelphia.

the past decade. The only limitation to such a modality is the lack of range
resolution, which actually is an advantage for the assessment of valvular
stenosis and, because of the continuous nature of ultrasound emission in
this modality, there is no maximum velocity limit.
The pulse Doppler (Figure 10.21), on the other hand, uses only a single
crystal that transmits a signal for a brief time (i.e., a “burst”) and then pauses
to listen for returning echoes from previous transmissions. In this way, the
device cannot only acquire the frequency shift (i.e., velocity) information,
but it can also determine the location of the moving reflector by recording
the time required for the sound waves to travel out and back from the
reflector. This capability is known as range resolution and is very useful
when it is important to be able to detect velocities at specific locations. The
resolution of the measurement along the beam axis is dependent upon the
length of the signal burst sent out (Figure 10.22). Typically, the number of
cycles transmitted in a pulse Doppler burst, n, is of the order of four to six,
so that there is sufficient information in the echo for Doppler-shift frequen-
cies to be detected. Thus, the sample volume length along the beam axis and
the device resolution is nλ. Consequently, high spatial revolution is achieved
by using a high frequency system. Situations where this feature is useful
would be in obtaining a velocity profile across an artery or in determining
the maximum blood velocity through a stenosed heart valve.
As mentioned above, the relationship used to convert Doppler velocities
to stenotic pressure gradients is generally the simplified Bernoulli equation

p1 − p2 = 4V22 (1.55)
7328_C010.fm Page 368 Monday, October 16, 2006 2:27 PM

368 Biofluid Mechanics: The Human Circulation

PRF
generator

Transmitting Transmission Master

amplifier gate oscillator

Receiving Range
Demodulator
amplifier gate

Transducer Sample
and hold
Delay gate
generator
Selected
range
Audio
filters
Sample
volume
Output

FIGURE 10.21
Pulse-Doppler layout. Atkinson, P. and Woodcock, J. (1982) Doppler Ultrasound and Its Use in
Clinical Measurement, Academic Press, London.

Transducer

(a) Transmit

Pulse

(b) Wait Target

Scatter

(c) Sample

Echo

Time

FIGURE 10.22
Pulse-Doppler basic principles. Atkinson, P. and Woodcock, J. (1982) Doppler Ultrasound and Its
Use in Clinical Measurement, Academic Press, London.
7328_C010.fm Page 369 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 369

where pi and p2 are in mmHg, V2 is the continuous-wave maximal velocity

in m/s, and the constant term, 4, has units of [mmHg/(m/s)2].
To derive such a simplified equation, three deletions were made from the
complete Bernoulli (energy balance) Equation 1.51. The first two, the absence
of acceleration and viscous effects, have been demonstrated to be valid
omissions both in vitro and in vivo for the obstructive geometries generally
presented in valvular stenosis. The third, neglecting of proximal velocities,
is not always clinically valid. If such an assumption is not made, the Bernoulli
equation appears as

1
p1 − p2 = ρ(V22 − V12 ) (1.53)
2

where V2 is the maximal distal velocity and V1 is the proximal velocity. The
distance between points 1 and 2 is the distance over which the pressure drop
occurs.
Since the distal velocity V2 in a stenotic jet usually significantly exceeds
any proximal velocity (and this is especially true for the values squared), the
proximal velocity is neglected to enable ease of application in the form of
Equation 8.1 with continuous-wave Doppler. However, in some cases such
as combined valvular insufficiency and stenosis for the same valve, blood
cell velocities are already elevated beyond normal at a location proximal to
the stenotic valve. These situations can be identified with 2-D echocardio-
graphy and, in such cases, measurement of the proximal velocity is necessary
using pulse Doppler techniques to correct, in effect, Equation 1.55.
Another useful in vivo application of pulse Doppler velocimetry is in
determining the velocity profile across a vessel. This can be important in
evaluating complex flow patterns, such as exist in the aortic arch (Figure 10.23),
branches and bifurcations, and in key arteries, such as the coronaries and
carotids (see Chapter 6, section 6.5: Flow Through Curved Arteries and
Bifurcations). It is also helpful in pathologic situations, such as arterial
stenoses and aneurysms, where the flow patterns are greatly altered by
sudden changes in geometry. Once a velocity profile is obtained (either in
real time or in the mean), additional parameters may be derived from it,
such as the wall shear stress (including the OSI (oscillatory shear stress) and
shear stress gradients) and also local residence times.
While range resolution is a valuable feature, the act of pulsing the trans-
mitted signal imposes certain limitations on the operation of the system.
First, the maximum range of the device is directly restricted by the time
period between pulses since the distance traveled to the target is the product
c·∆t. The period between successive transmissions is determined by the
instrument’s pulse repetition frequency, PRF, and can be expressed as 1/PRF
(s/cycle). Thus, a pulse Doppler device’s maximum range, Zmax, becomes

c( ∆t) c
Zmax = = (10.17)
2 2(PRF)
7328_C010.fm Page 370 Monday, October 16, 2006 2:27 PM

370 Biofluid Mechanics: The Human Circulation

4
4 5
1A 2
1A 2 5
3c 3c
1
1 3
3
2B
2B

90 ms 240 ms
(a) (b)

4 4
1A 2 5 5
1A 2
3c 3c
1 1
3 3
2B 2B

140 ms 320 ms
(c) (d)

4
5 4
1A 2 2
1A 2 5
3c
3c
1
1
3
3
2B
2B

170 ms 380 ms

(e) (f )

2
V1 4 5
50 cm.s–1 1
3
1A 3c
Vt 2B

Calibration Location of the

probes

FIGURE 10.23
Velocity profiles in the aorta of a dog using ultrasound Doppler velocity measurement tech-
nique: (a) 90 msec and (b) 320 msec after the opening of the valve. 1A, 2B, and 3C represent
profiles in perpendicular planes corresponding to the profiles shown in the plane of the figure.
(Redrawn from Farthing, S. and Peronneau, P. (1979) Flow in the thoracic aorta. Cardiovas Res.
13(11): 607–20. With permission.)
7328_C010.fm Page 371 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 371

Secondly, the pulsed nature of the device also limits the amount of infor-
mation that can be obtained from a measurement since a returning echo is
only received once every 1/PRF seconds. Therefore, the frequency of a Dop-
pler-shifted signal, ∆fmax, must be within certain values in order for it to be
accurately determined. Exceeding this limit, also known as the Nyquist
sampling limit, will cause “aliasing” to occur where the frequency of the
signal appears to be lower than it really is. Since the Nyquist theory requires
a sampling rate, which is at least twice that of the signal frequency, the pulse
Doppler can only accurately collect data with frequency shifts of ∆fmax <
(PRF/2). This, in turn, imposes a limitation on the maximum velocity, Vmax,
obtainable through the Doppler-shift Equation 10.15

(PRF/2)c
Vmax < for θ = 0° (10.18)
2( fs )

If we combine these two parameters, we obtain a Range × Velocity limit

for the device, or

 c   (PRF)c  c2
Zmax (Vmax ) <     < (10.19)
 2(PRF)   4( fs )  8( fs )

It is interesting to observe that, although each of these parameters (Zmax

and Vmax) is limited by the PRF, its product is not. Instead, it depends simply
on the wave speed, c, and the source frequency, fs. The implications of this
are that for a given flow location, or depth, only a limited range of velocities
can be accurately detected, or vice versa, without producing aliasing
(i.e., underestimates of velocity). And, the only recourse to extending this limit
(for the same fluid medium) is to change the source frequency, fs, to a lower
value. This can be done, but it turns out that this impacts other features of
the system, such as the attenuation of the sound through the medium (∝ fs)
and the size of the sample volume [∝ (1/fs)]. As an example, a pulse Doppler
velocimeter that transmits a 1 MHz signal through a water-based medium
(c = 1540 m/s) will have a ”range × velocity” limit of 0.30 m2/s, while one
operating at 10 MHz would have a product limit of 0.03 m2/s. The implication
of this is that at a distance of 10 cm from the transducer, the 1 MHz device
could measure velocities up to about 30 cm/s without aliasing, while the
signal from the 10 MHz device would alias at velocities above 3 cm/s.
In light of these considerations, a decision must be made in practice regard-
ing the tradeoff between these two factors. Increasing pulse repetition fre-
quency allows higher velocity measurements, but increases the chance of
range ambiguity. Decreasing pulse repetition frequency results in the oppo-
site, so the nature of the clinical situation at hand must be examined con-
stantly when making this decision. In general, the normal upper limit of
pulsed wave Doppler measurements is around 2 m/s. Using high PRF,
velocities of around 4 to 5 m/s can be obtained.
7328_C010.fm Page 372 Monday, October 16, 2006 2:27 PM

372 Biofluid Mechanics: The Human Circulation

02: 53: 24 PM
V219 24
Depth = 180
ADLT HRT

PWR = 0dB
LV 49dB B/0/F

RA LA

(a) (b)

WFU BOSH CDI

CMU NCBH

4 MAP
??
select Green
TAG

7Flow DIR
±, +, –
8 Playback
select

0 Color
ON/OFF

(c) (d)

FIGURE 10.24
Linear array transducer. Kremkau, F.W. (2002) Diagnostic Ultrasound Principles and Instruments,
6th ed., W.B. Saunders Co., Philadelphia.

Newer ultrasound Doppler techniques have greatly expanded the capa-

bilities of a single crystal device by incorporating a large number of crystals
into a single array (either linear or annular) that allows detection of veloc-
ities at many points in a 2-D field of view (Figure 10.24). The resultant
device combines velocity detection (Doppler) with an imaging capability
(called Brightness, or B-Mode), thus forming an ultrasound Duplex sys-
tem (Figure 10.25) that can provide flow information at known anatomic
locations. In order to display all these data, the device displays point
velocities in a color-coded presentation where velocity amplitudes vary as
intensities and flow directions are typically displayed as reds (toward the
transducer) or blues (away from the transducer). The resultant ultrasound
Color Doppler Flow Mapping (CDFM) systems are now widely used
clinically in many cardiovascular diagnostic applications.
One of the main advances of the CDFM systems is that their transducers
have many crystals (from 64 to 128 or 256) arranged along the face of the
7328_C010.fm Page 373 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 373

Conventional
doppler signal processor

Low pass Digital Color

Autocorrelator processor
filter scan converter

Velocity &
variance
Array calculator
Display

2-D B-mode
imager

FIGURE 10.25
Schematic of a color Doppler imaging system. Shung, K.K. (1992) Principles of Medical Imaging,
Academic Press, Inc. San Diego, CA.

transducer. Furthermore, by gating the signal, many sample volumes

(typically 64) can be arranged along a single line extending perpendicular
to the face of the transducer. This line may then be swept laterally or
rotationally back and forth at a frequency of 7 to 30 frames per second to
form either a block or sector scan. Since the scanning frequency is rela-
tively high compared to characteristic cardiac cycle frequencies, which are
on the order of 1 Hz, temporally varying blood flow velocities are dis-
played throughout the two-dimensional sector in what essentially appears
as real time. These velocities are then color coded (in RGB [red, green,
blue] format), usually with red for flow towards the transducer (i.e., up-
shifted Doppler frequencies) and blue for flow away from the transducer
(i.e., down-shifted Doppler frequencies) in order to compress all this infor-
mation into a single display. Within a certain pixel location, increasing
image intensity corresponds to a higher velocity in that beam direction.
Therefore, a cardiac CDFM image consists of moving white or grey ana-
tomic structures (i.e., B-mode images) on a black background, with super-
imposed color-coded blood flow (i.e., pulse Doppler) — all imaged in
quasi real time. It should be noted that, in addition to these very sophis-
ticated signal processing and display techniques, the CDFM device still
offers the capability of single-point pulse Doppler measurements by sim-
ply tracking the reflected echo from one selected crystal and then gating
it to a specific axial location. Using this feature, it is then possible to better
interrogate a point of particular interest (e.g., within a cardiac septal defect
or stenosed valve, or downstream of a stenosis, etc.) and obtain accurate,
pulsatile waveforms.
7328_C010.fm Page 374 Monday, October 16, 2006 2:27 PM

374 Biofluid Mechanics: The Human Circulation

In its present state, however, the full, 2-D CDFM display provides only
semiquantitative velocity information. This occurs because the system uti-
lizes a faster autocorrelation technique in place of direct spectral analysis
due to the greatly increased amount of data processing requirements of
CDFM compared to conventional Doppler. Also, pulse repetition frequencies
are set very low to avoid range ambiguity and to obtain greater depth
penetration (Equation 10.16) and, thus, Nyquist velocity limits also become
very low. With these low Nyquist velocities, aliasing frequently occurs with
physiologic flows, especially in high velocity situations, such as mitral regur-
gitation or aortic stenosis. The concept of aliasing in color flow results in a
‘‘wrap-around’’ of color to a value, which corresponds to flow in the opposite
direction. In these cases, multiple aliasing effects have restricted color image
quantification to a simple measurement of the color boundary location, since
quantitative distinction of velocities within the jet is virtually impossible for
high-velocity flows.

10.7 Laser Doppler Velocimetry

Laser Doppler Velocimetry (LDV) (also referred to as Laser Doppler
Anemometry [LDA]) is a technique used to accurately measure fluid velocity
by detecting the Doppler frequency shift of laser light that has been scattered
by small particles moving within the fluid. The LDV provides precise veloc-
ity data within a small volume element and, by using a traverse system to
move the laser probe volume point-by-point, it is possible to perform an
area investigation. Since only light is used as the measuring tool, the tech-
nique is noninvasive. Furthermore, it can reach difficult measurement loca-
tions without disturbing the flow. LDV offers a wide flexibility and is used
in many applications, such as transonic and supersonic flows, boundary
layers, and flames.
Over the past 3 decades, LDV systems have been used extensively for
in vitro biofluid mechanics studies. These studies have been performed under
both steady and pulsatile flow conditions to measure one-, two-, and three-
dimensional flow fields. From these data, it is possible to compute parame-
ters, such as velocity fields, turbulent normal stresses, and laminar and
turbulent shear stresses. Examples of such studies include:

1. Measurements in the immediate vicinity of prosthetic heart valves

2. Flow fields in arterial bifurcation geometries
3. Flow fields downstream in arterial stenoses
4. Flow fields in vascular bypass geometries
7328_C010.fm Page 375 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 375

10.7.1 General Features

A typical LDV system uses a plasma tube, such as an argon-ion gas laser,
to produce the laser beam. This intense laser beam is directed toward a fiber
optic drive unit through a Bragg Cell, which splits the incoming laser into
two parallel beams of equal power, but of different frequency. The Bragg
cell adds a frequency shift to one beam of each focal pair; i.e., one beam
(called the zero order beam) retains the frequency of the incident beam,
while the other beam (called the first order beam) is typically shifted by
40 MHz in frequency. The two beams are then manipulated by dispersion
prisms that separate them into individual colors (i.e., wavelength), for
example, green (514.5 nm), blue (488 nm), and violet (476.4 nm), in the case
of the argon-ion laser. This process creates two beams for each color, for a
total of six beams. Each color pair forms a plane of measurement and the
optical arrangement is such that the three color pairs are perpendicular to
one another.
Fiber optic transceiver probes are coupled to the fiber drive, allowing the
crossing of the individual color beam pairs to produce an ellipsoidal probe
volume. Thus, a two-component measurement entails the crossing of four
laser beams, while a three-component measurement entails the crossing of
six laser beams. Because a single transceiver is required for two-component
measurement, a three-component investigation requires the use of two
transceivers placed at right angles to each other (Figure 10.26). The back-
scattering mode is commonly used in a two-component study, where a
single optic fiber probe can act as both a transmitter as well as a receiver.
An additional receiver is usually required in the forward and side scattering

Fiber optic
2-D transceiver
blue and green

Fiber optic Fiber optic

1-D transceiver
receiver violet
Model

Table

FIGURE 10.26
Transceiver and receiver orientation around a flow model.
7328_C010.fm Page 376 Monday, October 16, 2006 2:27 PM

376 Biofluid Mechanics: The Human Circulation

modes for a three-component investigation. The spatial resolution of this

technique is determined by the beam width and the angle of the beam
intersection.

10.7.2 Probe Volume Specifications

Figure 10.27 shows the intersection of two arbitrary color beams, which
creates a probe volume with the shape of an ellipsoid. Because the beams
are coherent sources, the probe volume is a stable interference pattern char-
acterized by alternating light and dark fringes. In this figure, lm represents
the length dimension of the probe volume, dm is the diameter dimension, df
is the fringe spacing, α is the half-angle of the beam intersection, d is the
prefocal beam spacing, and f is the beam focal length.

10.7.2.1 Calculation of Probe Volume Dimensions

In the case where the beam pairs intersect in air, α can be calculated from
d and f using the following equation

d
tan α = 2
(10.20)
f

Flow direction
Reference beam
d α
α
Bragg beam Probe volume
(exaggerated)
Transceiver Lens

Exit

df
dm

Entrance
lm

Particle

FIGURE 10.27
Probe volume formed by the intersection of a laser beam pair.
7328_C010.fm Page 377 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 377

The dimensions of the ellipsoidal probe volume, dm and lm, formed by a pair
of intersecting beams, are given by

4λ f
dm = (10.21)
πDe E

dm
lm = (10.22)
tan α

where λ is the wavelength of the beam in air, f is the focal length of the
transceiver lens, De is the original beam diameter, and E is the beam expan-
sion ratio.

10.7.2.2 Fringe Patterns

Because the beams are coherent sources, the probe volume is a stable
interference pattern characterized by alternating light and dark fringes
(Figure 10.27). When a particle passes through the probe volume formed
by the intersection of all beam pairs, it scatters the light from within the
probe volume, generating a signal called a Doppler Burst. The time
required for a particle to traverse the probe volume, called its gate time,
is recorded for each valid particle. In order to measure multiple velocity
components of a given particle, the probe volumes formed by each beam
pair are aligned so that they intersect at the same spatial location.
Within the destructive fringes (dark fringes) of the probe volume, the light
intensity is zero. Within the constructive fringes (bright fringes) of the probe
volume, however, the light intensity varies in a Gaussian fashion from the
probe volume entrance to the exit. Thus, the signal amplitude modulates as
the particle alternately intersects dark and bright fringes. Consequently,
when a particle crosses a dark fringe, the intensity is zero, and when it crosses
a light fringe, the intensity is high. The output voltage then peaks when the
particle crosses the central light fringe of the probe volume where the inten-
sity is the highest. The crossing of the particle through the alternate light
and dark fringes gives a characteristic “on–off intensity” Doppler burst as
illustrated in Figure 10.28. The variation in light intensity causes a low
intensity background pedestal to be superimposed on the burst. This ped-
estal is removed during signal processing.
The frequency relationship between a forward scattered light wave and
the incident light wave is given by the following vector equation

1
f s = fl + V ⋅ ( js − ji ) (10.23)
λ

where fi is the frequency of incident light, fs is the frequency of scattered

light, ji is the unit vector in incident direction, js is the unit vector in
7328_C010.fm Page 378 Monday, October 16, 2006 2:27 PM

378 Biofluid Mechanics: The Human Circulation

Light intensity

Time

FIGURE 10.28
Schematic of a raw Doppler burst.

scattering direction, V is the velocity vector, and λ is the wavelength of

incident light.
The Doppler frequency, fd, is given by

fd = fs − fi (10.24)

Therefore,

1
fd = V ⋅ ( js − ji )
λ

The distance between the fringes is fixed by the wavelength of the laser
beam and the angle between the intersecting beams. The light scattered by
the particle passing through the probe volume is modulated with a fre-
quency, which is related to the velocity of the particle by the equation

λ
V = fd (10.25)
2 sin α

where V is the component of local velocity, which is normal to the bisector

of the beam intersecting angle. Alternately,

V = fd d f (10.26)
7328_C010.fm Page 379 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 379

and

λ
df = (10.27)
2 sin α

It can be seen from Equation 10.25 that the Doppler frequency is directly
proportional to the velocity. The constant of proportionality is a direct func-
tion of the wavelength of the laser and of the optical arrangement of the
experiment.

10.7.3 Photodetectors
Particles moving through the measuring volume scatter light of varying
intensity. However, the light signals are usually too weak to be analyzed
and, thus, photomultipliers (PM) are needed to amplify and convert the
input to a voltage output for analysis. A photomultiplier tube consists of a
photocathode, a multiplier chain, and an anode. When photons from the
scattered laser beam of a passing particle hit a PM tube, they are converted
into electrons by means of the photoelectric effect. These electron emissions
are amplified and accelerated by a string of successive electron absorbers
called dynodes to produce enhanced secondary emission. Finally, the anode
at the end of the tube is used to collect the resulting voltage that is large
enough to be picked up and analyzed. Some laser systems make use of
photodiodes for the signal amplification; however, the main advantage of
using photomultiplier tubes over photodiodes is the high gain and low noise
levels, which allow the system to operate with a light signal of low intensity.
The light signal picked up by the photodetectors contains the Doppler signal
as well as the frequency shift from the Bragg cell.

10.7.4 Signal Processing

The optical frequency shift introduced by the Bragg cell causes the fringe
pattern in the probe volume to move up on the frequency axis, thereby
displacing the zero velocity away from the zero frequency. This frequency
shift is necessary in order to discriminate between negative and positive
velocities, which might otherwise produce the same detected frequency shift
(since frequency cannot be negative). It is also necessary because the Doppler
frequency ( fd) of the particle is very small compared to the frequency of the
laser beams. For example, the frequency of visible light is on the order of
1015 Hz, while the Doppler frequencies can be on the order of 106 Hz. Thus,
the frequency shift permits the resolution required to process the positive
and negative velocity fluctuations that typically occur in each of the mea-
sured directions.
7328_C010.fm Page 380 Monday, October 16, 2006 2:27 PM

380 Biofluid Mechanics: The Human Circulation

Voltage

Time

FIGURE 10.29
Down-mixed filtered Doppler signal.

The raw Doppler signal and a superimposed pedestal are created when-
ever a particle crosses the probe volume. The resultant signal thus consists
of (1) the true Doppler frequency of the particle, (2) the 40 MHz Bragg shift,
and (3) the frequency of the pedestal. The signal is passed through a pream-
plifier and through real-time signal analyzers (RSA). In the RSA, a high-pass
filter is applied to the signal to remove the low intensity pedestal. The signal
is then down-mixed, a process that removes the 40 MHz shift and leaves the
Doppler frequency fD. After down-mixing, the signal is passed through a
low-pass filter, which eliminates high frequency noise. The down-mixed,
filtered Doppler signal is illustrated in Figure 10.29.

10.7.5 Phase Window Averaging of LDV Data in Pulsatile Flow

The first step in processing laser Doppler velocimetry measurements made in
simulated cardiac pulsatile flow is to divide the heart cycle into discreet inter-
vals, called phase windows (Figure 10.30). Within these phase windows, the
flow is assumed to be steady, allowing the use of ensemble averaging to describe
the statistics of the flow. In choosing the size of the phase windows, there is a
tradeoff between temporal resolution and statistical accuracy. The choice of a
relatively small phase window allows a fine definition of the temporal charac-
teristics of the flow and increases the strength of the steady flow assumption.
However, a relatively small phase window also contains a relatively small
number of discreet velocity samples for ensemble averaging, and limits statis-
tical accuracy. Conversely, choosing a large phase window limits the temporal
definition of the flow and decreases the strength of the steady flow assumption,
but increases statistical accuracy of the velocity measurements.
In typical measurements, velocities are usually phase-averaged within
10 to 30 ms time windows. The averaging procedure is summarized as
7328_C010.fm Page 381 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 381

1.2 Velocity measurements

Phase window
Velocity (m/s)

0 860
Cycle time (ms)

FIGURE 10.30
Phase window averaging.

follows. The mean and standard deviation of the measurements are calcu-
lated as

a n
∑ ∑ uijGij
i =1 j =1
mean = a n (10.28)
∑ ∑ Gij
i =1 j =1

a n
∑ ∑ (Uij − mean)2 Gij
i =1 j =1
S.D. = a n (10.29)
∑ ∑ Gij
i =1 j =1

where a represents the total number of cycles during which measurements

from a specific location are collected, n represents the number of measure-
ments at that location with a specific combination of cycle number and phase
window, and Gij is the gate time associated with instantaneous velocity, Uij.

10.8 Magnetic Resonance Imaging and Velocity

Mapping Techniques
Magnetic resonance imaging, or MRI, is a widely used clinical imaging
modality that provides high quality images with detailed anatomical infor-
mation. In addition, either qualitative or quantitative flow information can
7328_C010.fm Page 382 Monday, October 16, 2006 2:27 PM

382 Biofluid Mechanics: The Human Circulation

be obtained from changes in signal intensity or directly from the phase of

the signal, respectively.
The main component of an MRI examination is the scanner. A strong static
magnetic field is created by super-conducting materials, which are kept at
very low temperatures by cooling with liquid nitrogen and helium. A com-
bination of coils is used to cause small variations, or gradients, in the static
magnetic field in any orientation in space. These gradients are responsible
for the creation of an image.
The principles of MRI have been extensively published and, therefore in
this chapter, only the main points will be discussed briefly. Consider a
nucleus with an odd number of particles (sum of protons and neutrons).
Because of its charge and spin, a magnetic field is created. If the nucleus is
placed inside a static magnetic field, the vector of the magnetic moment will
tend to align parallel (the low energy state) or antiparallel (the high energy
state) to the vector of the magnetic moment of the applied static magnetic
field. The combination of the external magnetic field and the angular
momentum will force the nucleus to a composite motion called precession
(Figure 10.31). The frequency of the precession, called the Larmor frequency,
depends on the strength of the magnetic field and on the type of material,
and is given by the following equation

Φ = γ B0 (10.30)

where Φ is the angular frequency of the precession or Larmor frequency, γ

is the gyromagnetic ratio (Tesla−1·s−1) and Bo is the strength of the static

Precession

Bo Mz
M
M

Proton Mxy

M: Magnetic moment of proton

Bo: External magnetic field

FIGURE 10.31
The combination of the angular momentum and external magnetic field results in a complex
motion called precession.
7328_C010.fm Page 383 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 383

magnetic field (Tesla). Because hydrogen with a single proton in its nucleus
is most commonly used as the target atom in clinical MRI, the term “nucleus”
will be replaced hereafter by “proton.”
Of the two possible orientations for the proton’s magnetic moment vector,
the parallel orientation prevails slightly over the antiparallel. It is this net
magnetic moment of the whole mass of protons directed parallel to the
magnetic moment of the static magnetic field that is actually responsible for
detecting a signal at the end of the acquisition procedure.
An advantage of MRI compared to other imaging techniques is that any
region in the body at any orientation can be imaged without limitations
related to acoustic windows or to the depth of the region of interest (in
contrast to ultrasound, for example). The imaging procedure consists of
(1) slice selection, or excitation, (2) encoding in the phase and frequency direc-
tions, or spatial encoding, (3) signal read-out, and (4) image reconstruction.

10.8.1 Slice Excitation

The first step during the MRI examination is to select and energetically excite
the region to be imaged. Since protons within a magnetic field, Bo, precess
with the Larmor frequency, φ, they can be energetically excited to move
between the parallel and the antiparallel energy states by another field with
a frequency equal to ƒ. This excitation is achieved by the application of
another magnetic field, B1, which acts very rapidly as a pulse with a fre-
quency in the radio wave band called “RF pulse.” Before the RF pulse is
applied, the sum of the vectors of the magnetic moments of all protons in a
sample volume is the net magnetization vector M aligned with the static
magnetic field along the bore of the scanner. This direction is conventionally
referred to as the “z direction.” By applying the RF pulse, protons are forced
to move between the lower and higher energy states. Since the number of
protons with a magnetic momentum vector parallel to Bo is larger than that
with antiparallel orientation, the result is a net transition from the lower
energy state to the higher energy state or an excitation. Therefore, M is
‘tipped’ (i.e., rotated) from the z-axis towards the x–y plane (Figure 10.32).
The angle that M is tipped with respect to the z-axis is called “flip angle.”
To ensure that a specific thin slice in the body will be excited, the pre-
cession frequency of the protons to be imaged can be varied by proper
application of magnetic field gradients. These gradients create a spatial
variation in the strength of the static magnetic field, Bo ± B(r), where r is
the direction of slice selection. Therefore, the Larmor frequency varies
depending on the position along the slice selection direction. Upon appli-
cation of the RF pulse, only those protons within a slice having a Larmor
frequency equal to that of the RF pulse will be excited. Therefore, axial
resolution is achieved since these are the only protons that will contribute
to the resulting image.
7328_C010.fm Page 384 Monday, October 16, 2006 2:27 PM

384 Biofluid Mechanics: The Human Circulation

Excitation
z
Bo

y
Mz

B1

M: Magnetic moment of proton

Bo: External magnetic field
B1: RF field

FIGURE 10.32
Proton excitation: By applying an RF pulse, B1, the magnetization vector is tipped from the
z-axis towards the x–y plane.

10.8.2 Spatial Encoding

At the end of the application of the RF pulse, all protons within a thin slice
are excited such that the magnetic moment, M, is tipped from the z direction
towards the x–y plane. After the excitation, the protons relax by releasing
the excessive energy in the form of a detectable signal through a process
called “free induction decay” (FID), as shown schematically in Figure 10.33.
If signal read-out were the only event to follow slice excitation, the protons
would relax by emitting the excessive energy in the form of a signal with a
single frequency, since all protons under the same magnetic field resonate
with the same Larmor frequency. The reconstructed image would contain
no information about structures, function, or flow. Therefore, it is necessary
to encode the position of each proton or, in practice, a small group of protons
within a volume element (voxel) in order to ensure a strong signal. The slice
can then be divided into voxels, each of which will correspond to a different
resonant frequency; thus, it will be spatially encoded. To achieve such an
encoding, the signal information in each voxel must be unique. This is
accomplished by applying proper magnetic field gradients in the two planar
directions of the slice. One of those directions is used for phase encoding
and the other for frequency encoding.
During the phase encoding step, the magnetic field gradient is applied
solely to differentiate the precession frequency in each row of voxels for
a short period of time. At the end of this step, all rows return to the original
precessing frequency, but now include a small differentiation with respect
7328_C010.fm Page 385 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 385

Free induction decay

Bo z

y
Mz

FID

FIGURE 10.33
Free Induction Decay: After termination of the RF pulse, the excited protons emit their excessive
energy while returning to their original condition. The magnetization vector returns to the
z direction alignment. It is during this period that the image data is actually acquired.

to the phase of this precession from row to row. After phase encoding,
another magnetic field gradient is applied in the frequency encoding
direction. This creates a change in the precession frequency along each
row. Since each row already had an identity in space through the differ-
entiation in the phase of precession, the second encoding causes an iden-
tification of each separate voxel with a unique frequency and phase. This
is also the stage when the receiving coil reads the signal. This signal fills
the frequency k-space and, through 2-D Fourier transform, the final image
is reconstructed.

10.8.3 Imaging Procedure and Pulse Sequences

To obtain an initial idea about the location of the region of interest, a series
of “scout” or “localizer” images is acquired. This is usually a multislice
acquisition procedure. In the scout images, it is possible to isolate and
observe an organ or a vessel and plan the main acquisitions in the region
of specific interest. Depending on the type of information needed, imaging
can be performed using a variety of pulse and gradient sequences. Here,
we will consider two pulse sequences that may be used — spin-echo and
gradient-echo.
7328_C010.fm Page 386 Monday, October 16, 2006 2:27 PM

386 Biofluid Mechanics: The Human Circulation

10.8.3.1 Spin-Echo
In this imaging modality, a 90° RF pulse is used to tip the net magnetization
vector M by 90°, moving it completely into the x–y plane (Figure 10.34).
Then, free induction decay occurs while the individual spins begin to precess
out-of-phase or dephase due to local magnetic field inhomogeneity. Since
dephasing reduces the amplitude of the received signal, rephasing of the
spins is desirable. This can be achieved by the application of an additional
180° RF pulse that forces the spins to rephase, producing an echo. The
emitted signal is read when this echo occurs. The components of the spin-
echo sequence are shown in Figure 10.35. It is widely used clinically, espe-
cially to obtain anatomical information anywhere in the body. In the heart,
it is used mainly to observe vessel walls that appear bright compared to
blood that appears dark (also referred to as MRI Angiography).

Y
90 deg pulse

180 deg pulse

Echo

FIGURE 10.34
Spin-echo sequence: Initially, the magnetization vector (bold upward arrow) is tipped 90°. Then,
spins start to dephase until another 180° pulse rephases them and an echo is generated.
7328_C010.fm Page 387 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 387

90 180 Echo

RF pulses
and echo

Slice select
direction

Phase encoding
direction

Frequency
encoding direction
(read-out)

FIGURE 10.35
Diagram of a typical spin-echo pulse sequence.

10.8.3.2 Gradient-Echo
This is the most commonly used pulse sequence for cardiovascular imaging,
especially to obtain flow information, either qualitatively or quantitatively.
In gradient-echo, the RF pulse tips the magnetization vector M to an angle
less than 90°. This provides a faster acquisition procedure, since less time
is required for relaxation, which is important for cine cardiac imaging. After
excitation, the relaxation process and the dephasing of the individual spins
begin. Application of an additional magnetic field gradient in the frequency
encoding direction causes the dephasing to accelerate. Then, another gra-
dient of opposite polarity rephases the spins and an echo is formed and
read. The great advantage of the gradient-echo pulse sequence is its speed
compared to spin-echo, since the flip angle is usually much less than 90°.
A cine acquisition throughout a specific time period then can be achieved.
This is very important when imaging structures that are in motion, such as
the heart or arteries. Gradient-echo sequence components are shown in
Figure 10.36.
7328_C010.fm Page 388 Monday, October 16, 2006 2:27 PM

388 Biofluid Mechanics: The Human Circulation

RF Echo

RF pulses
and echo

Slice select
direction

Phase encoding
direction

Frequency encoding
direction (read-out)

FIGURE 10.36
Diagram of a typical gradient-echo pulse sequence.

10.8.4 Magnetic Resonance Phase Velocity Mapping

The result from the conventional imaging acquisition and display process
is the reconstruction of an image where contrast is based on variations of
the magnitude or intensity of the received signal. MRI can also be used
to directly measure the velocity of moving protons based on the following
principle. When a proton moves along a magnetic field gradient, it
acquires a phase shift in its precession proportional to the strength and
duration of the magnetic field gradient. This shift is expressed by the
following equation

Φ= ∫ γ G(t)[r(t)]dt (10.31)

where Φ is the phase of the signal, γ is the gyromagnetic ratio, G(t) is the
vector of the magnetic field gradient (mTesla/m), and r(t) is the position
vector of the protons (in meters). For the case of motion in only a single
7328_C010.fm Page 389 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 389

Magnetic field gradient strength

Time

FIGURE 10.37
A schematic of a Bipolar Gradient, Gz (t): The total area of the two lobes is zero, resulting in a
linear relationship between signal phase and proton velocity.

direction where acceleration and higher order motion terms are neglected,
the position vector is:

r(t) = Z0 + u ⋅ t (10.32)

where Z0 is the initial position of the proton (in meters), u is the velocity of
the proton in the z direction (m/s), and t is the time of the motion (in seconds).
Combining Equation 10.31 and Equation 10.32 results in

Φ= ∫ γ G (t)Z dt + ∫ γ G (t)u ⋅ tdt

z 0 z (10.33)

By applying a bipolar gradient Gz(t), as shown in Figure 10.37, the first term
on the right side of Equation 10.33 becomes zero. In addition, if the particle
velocity is assumed constant, then

∫
Φ = γ u Gz (t)tdt (10.34)

As seen from Equation 10.34, a linear relationship exists between the phase
of the signal and the velocity of protons. Thus, if the phase is obtained, the
velocity can be determined by

Φ
u= (10.35)
 γ G (t)tdt 
∫ z 
7328_C010.fm Page 390 Monday, October 16, 2006 2:27 PM

390 Biofluid Mechanics: The Human Circulation

The velocity can be measured in all three Cartesian coordinates in space.

By using the bipolar velocity-encoding gradient, in which the total area
under the lobes is zero, stationary protons will not have any resulting phase,
whereas moving protons will. This linear relationship between signal phase
and proton velocity is the principle of phase velocity mapping (PVM).

10.9 Computational Fluid Dynamics

As pointed out in previous chapters, the dynamic character of blood flow is
a predominant factor in the development of disease (e.g., atherosclerotic
plaques that preferentially develop in regions exposed to a low mean shear
stress and cyclic reversal of flow direction) as well as the functioning of
implanted devices (e.g., platelet activation and thrombus formation around
prosthetic heart valves due to their abnormal hemodynamics). Thus, it
becomes very important to understand blood flow through the cardiovas-
cular system to get a new insight into disease progression and clinical com-
plications. Many of these features can be delineated through the use of
mathematical modeling techniques.
In Chapter 1, the motion of fluid was described with a system of differential
equations called the Navier–Stokes equations. Solution of these equations
can be obtained analytically only for the simplest physical problems —
generally those limited to laminar, steady flows with large fluid viscosity or
small velocity; in other words, those with relatively low Reynolds numbers.
For more complex situations, solution of the Navier–Stokes equations must
be obtained computationally. Computational Fluid Dynamics (CFD), there-
fore, is a method to numerically predict a discrete solution for the nonlinear
Navier–Stokes equations and, thus, determine the dynamics of blood flow.
CFD has been used to reproduce and predict the pulsatile complex flow field
in the cardiovascular system. The development of computational models of the
fluid mechanics in the cardiovascular system is motivated by both medical and
economic concerns. Accurate and descriptive computational models serve as
an excellent tool for scientific and medical research. Additionally, computa-
tional models provide medical device developers, such as prosthetic valve
manufacturers, with a means of validating and predicting device performance
to improve the product development process and make it more cost effective.
Finally, fast and accurate computational models can be used in clinical settings
if they are combined with medical imaging and other cardiovascular diagnostic
techniques to provide real-time information to improve patient care.

10.9.1 Governing Equations

A fluid whose shear stress is linearly proportional to the velocity gradient
in the direction perpendicular to the plane of shear is considered Newtonian.
7328_C010.fm Page 391 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 391

The behavior of such a fluid was described in Chapter 1 with Equation 1.1.
Blood is a fluid that does not obey the relation of Equation 1.1 and is called
non-Newtonian because its viscosity changes with the applied shear forces.
However, non-Newtonian fluid behavior is extremely complex, and the
assumption of a Newtonian fluid is commonly made in order to predict flow
dynamics using numerical simulation tools. Another simplification com-
monly used is to consider the fluid as being incompressible. In this case, the
fluid density is assumed to be constant throughout the flow field and over
time. Despite the fact that in reality all fluids are compressible to some extent,
this idealization holds in most cases.
The equations that govern the motion of Newtonian fluids are called the
Navier–Stokes equations and are given by Equation 1.43a to Equation 1.43c.
These equations, derived from the conservation laws of mass, momentum,
and energy, are the so-called continuity equation, and momentum and
energy balances. The governing equations for an incompressible isothermal
flow of a constant density fluid r can be expressed in Cartesian coordinates,
using Einstein notations, as follows.
Continuity Equation

∂ui
=0 (10.36)
∂xi

Momentum Equation

∂ui  ∂υ  ∂ui ∂2 u ∂P ∂υ ∂uj

+  uj −  − υ 2i = − − =0 (10.37)
∂t  ∂x j  ∂x j ∂x j
∂xi ∂x j ∂xi

where ui is the velocity in the ith direction, xj is the jth direction coordinate,
P is the pressure divided by the density ρ, υ is the fluid kinematic viscosity
(defined as υ = µ ρ ), and t denotes time.
Since the fluid density is assumed constant, the continuity and momen-
tum equations constitute a closed system of equations, i.e., there are two
equations for the two unknowns — velocity and pressure. In compressible
flow computations, the energy equation is decoupled from the momentum
and continuity equation and discarded. Therefore, the continuity and
momentum equations are the only equations solved to obtain velocity and
pressure fields. The energy equations are only solved if heat transfer is of
interest.
The remaining equations can be written in their noncompact forms as
Continuity Equation

∂u ∂v ∂w
+ + =0 (10.38)
∂x ∂y ∂z
7328_C010.fm Page 392 Monday, October 16, 2006 2:27 PM

392 Biofluid Mechanics: The Human Circulation

Momentum Equations

∂u ∂  2 P ∂u  ∂  ∂u  ∂  ∂u 
+ u + − υ  +  uv − υ  +  uw − υ  = 0
∂t ∂x  ρ ∂x  ∂y  ∂y  ∂z  ∂z 
(10.39a)

∂v ∂  ∂v  ∂  P ∂v  ∂  ∂v 
+ uv − υ  +  v 2 + − υ  +  vw − υ  = 0
∂t ∂x  ∂x  ∂y  ρ ∂y  ∂z  ∂z 
(10.39b)

∂w ∂  ∂w  ∂  ∂w  ∂  2 P ∂w 
+  uw − υ  +  vw − υ + w + − υ =0
∂t ∂x  ∂x  ∂y  
∂y  ∂z  ρ ∂z 
(10.39c)

where x, y, and z are the Cartesian coordinates; u, v, and w are the respective
velocity components; t is time; υ is the fluid kinematic viscosity; P is the
pressure; and ρ is the fluid density.
To simplify programming and ease algorithm implementation, the
Navier–Stokes equations are commonly written in vector form. To do so, all
governing equations are regrouped into a single equation for a vector con-
taining all flow variables (i.e., density, velocity components, pressure, etc.).
For instance, the incompressible Newtonian Navier–Stokes equations in vec-
tor form are given by

∂Q ∂E ∂F ∂G ∂Eυ ∂Fυ ∂Gυ

Γ + + + − − − =0 (10.40)
∂t ∂x ∂y ∂z ∂x ∂y ∂z

where Γ is a diagonal matrix; Q is the vector containing the flow variables;

E, F, and G are the convective flux vectors; and Ev , Fv and Gv are the viscous
flux vectors. These terms are specifically defined as

Γ = diag(0, 1, 1, 1) (10.41)

P
 
u
Q=  (10.42)
v
w 
 
7328_C010.fm Page 393 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 393

and

 u   v   w 
     
u2 + P   uv   uw 
E =  ρ F=

 
 v 2 + P  G =  vw 
 (10.43)
 uv   ρ  P
    w 2 + 
 uw   vw   ρ 

 0   0   0 
     
∂
 u
 ∂u   ∂y   ∂u 
 ∂x     ∂z 
Eυ =  ∂v  Fυ =  ∂v  Gυ =  ∂v  (10.44)
 ∂x   ∂y   ∂z 
     
 ∂w   ∂w   ∂w 
 ∂x     ∂z 
∂
 
y

10.9.2 Grid Generation

Most flow fields in the cardiovascular system involve flows though complex
channels (i.e., flows through heart valves, blood flow in the heart, flow in
arteries, etc.). To compute such flows, the geometries must be described in
detail and computational grids must be generated. This constitutes the first
step toward computing the flows and is generally referred to as the grid
generation problem. This step consists of choosing an appropriate coordinate
system that facilitates the description of the geometry under consideration.
For instance, to simulate flow through a straight, circular blood vessel, a
simple choice to generate a grid would be to use a Cartesian coordinate
system (Figure 10.38a). However, it is evident that the use of such a system

θ
y

x
(a) (b)

FIGURE 10.38
Grid generation for flow through a model of a straight blood vessel.
7328_C010.fm Page 394 Monday, October 16, 2006 2:27 PM

394 Biofluid Mechanics: The Human Circulation

Geometry

FIGURE 10.39
Generalized coordinate grid for an aortic arch model.

will not guarantee the presence of grid nodes located near the surface of the
blood vessel. This will make the enforcement of the physical boundary
conditions challenging. For this rather simple problem, a more suitable coor-
dinate system is the cylindrical polar system shown in Figure 10.38b. The
coordinate system consists of two coordinate lines: (1) lines of constant θ
and perpendicular to the surface body, and (2) lines of constant r and parallel
to the surface body. As shown in Figure 10.38b, a coordinate line coincides
with the surface body and, therefore, the boundary conditions can be easily
applied to all nodes located along that line. Such a coordinate system is
called a body-fitted coordinate system.
For complex body shapes, the choice of a body-fitted coordinate system
is not always straightforward. For instance, consider the aortic arch geometry
depicted in Figure 10.39. The body-fitted coordinates do not resemble any
commonly used coordinate systems (i.e., cylindrical, spherical, etc.). Such a
system is classified as a generalized nonorthogonal coordinate system. With this
coordinate system, body-fitted grids can be generated for any geometrical
body. However, it is important to emphasize that the handiness of the gen-
eralized body-fitted coordinate system is thwarted by the complexity of the
governing differential equations once they are transformed to generalized
coordinates.
With the use of body-fitted coordinates, flows around and through arbi-
trarily complex geometries can be treated as easily as flows in simple geom-
etries. Transforming the complex physical domain, defined by the generalized
coordinates, into a simple computational domain does this. This computational
domain, also called the transformed domain, is essentially an equally spaced
Cartesian grid — it is a rectangular domain for 2-D problems or a parallel-
epiped for 3-D problems. The transformation from the physical flow domain
to the computational domain is depicted in Figure 10.40. In a 2-D case, the
7328_C010.fm Page 395 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 395

j = jmax

Transformation
η (x, y, z) → (ξ, η, ζ)
ξ ∆η

η ∆ξ
y j=1
i=1 i = imax
x ξ

FIGURE 10.40
Transformation from the physical flow domain to the computational domain.

physical domain is described using Cartesian coordinates (x,y) and the com-
putational domain using generalized coordinates (ξ,η). The transformation
(x,h,z)
(x,y,z) can be described as

ξ = ξ( x , y , z), η = η( x , y , z), ς = ς( x , y , z) (10.45)

and the inverse transformation, (x,y,z)
(ξ,η,ζ) can also be defined as

x = x(ξ, η, ς ), y = y(ξ, η, ς ), z = z(ξ, η, ς ) (10.46)

Using these coordinate transformations and formulas for transforming the

various differential operators, the two dimensional Navier–Stokes equations
can be formulated in generalized curvilinear coordinates as

1 ∂Q ∂E ∂ F ∂Eυ ∂ Fυ
Γ + + − − =0 (10.47)
J ∂t ∂ξ ∂η ∂ξ ∂η

where J is the Jacobian of the geometric transformation defined as

 ξ x ηx 
J = det   = ξ x η y − ξ y ηx (10.48)
 ξ y ηy 

1 1
E = ( ξ x E + ξ y F ), F = ( η x E + η y F ) (10.49)
J J

1 1
Eυ = (ξ x Eυ + ξ y Fυ ) , Fυ = ( ηx Eυ + ηy Fυ ) (10.50)
J J
7328_C010.fm Page 396 Monday, October 16, 2006 2:27 PM

396 Biofluid Mechanics: The Human Circulation

and E, F, Eυ, Fυ are defined as previously in Equation 10.43 and Equation

10.44. Furthermore, ξix j = ∂ξi ∂x j with xj the jth direction coordinate in the
physical domain and ξi the ith direction coordinate in the computational
domain.
As seen from Equation 10.47 through Equation 10.50, the use of a gener-
alized curvilinear coordinate system simplifies the shape of the computa-
tional domain at the expense of complicating the governing equations.

10.9.3 Discretization Techniques

As mentioned in the previous section, the first step in simulating flows is to
transform the continuous physical flow domain into a discretized computa-
tional domain consisting of a finite number of points, called nodes. In the
computational domain, a continuous function can then be represented by a
set of discrete values at each node and every continuous derivative of such
a function can be approximated by algebraic expressions involving only
values of the function at the nodes. The second step toward simulating flows
is, therefore, to reduce the continuous governing equations to a system of
algebraic equations that represents discrete approximations of the original
equations at nodes of the computational domain. This step relies on spatial
discretization techniques. A successful spatial discretization scheme should
(1) have good spatial resolution, (2) ensure stability of the numerical algo-
rithm, and (3) be simple to implement. The accuracy of these discretization
techniques depend on two main criteria: (1) the size of the grid spacing
(which defines how well the discretized domain approximates the continu-
ous domain) and (2) the order of accuracy of the discretization formulas
used to approximate the function derivatives (which corresponds to the rate
at which discretization-related errors approach zero as the grid spacing tends
to zero). Several discretization methods have been developed and the most
widely used, the Finite Difference method, will be briefly introduced in
the following section.

10.9.3.1 Finite Difference Method

The general concept of the finite difference method is simple and based on
the calculus definition of the derivative. Consider the first order derivative
of the function u = u(x). By definition, this derivative is:

∂u lim u( x + ∆x) − u( x)
= (10.51)
∂x ∆x → 0 ∆x

Assuming ∆x to be small but finite, the derivative can be approximated by

∂u u( x + ∆x) − u( x)
≈ (10.52)
∂x ∆x
7328_C010.fm Page 397 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 397

and the accuracy of this approximation can be estimated by expanding the

function u in a Taylor series as follows

∂u ∆x 2 ∂2 u ∆x 3 ∂3u ∆x 4 ∂ 4u
u( x + ∆x) = u( x) + ∆x + + + +… (10.53)
∂x 2 ∂x 2 6 ∂x 3 24 ∂x 4

Substituting Equation 10.53 into Equation 10.52, the first order derivative
can be expressed as

u( x + ∆x) − u( x) ∂u ∆x ∂2 u ∆x 2 ∂3u ∆x 3 ∂ 4u
= + T with T = + + +…
∆x ∂x 2 ∂x 2 6 ∂x 3 24 ∂x 4
(10.54)

where T is the truncation error of the approximation.

The leading term of the truncation error, T, indicates the order of accuracy
of the expression. In the present case, the discrete approximation equation
is first order accurate in space. Using the notation θ for the order of accuracy,
the discretization scheme can be written as

u( x + ∆x) − u( x) ∂u
= + θ( ∆x) (10.55)
∆x ∂x

It is important to note that as the grid spacing approaches zero all the terms
in the truncation error series approach zero.
Using Taylor series expansions or polynomial fitting, one can derive dis-
crete approximations of any order of accuracy for continuous derivatives. As
an example, several discretization schemes for the first order derivative of
the function u using the notation given in Figure 10.41 are listed in Table 10.1.
The very simple implementation of the finite difference method explains
its popularity. Nonetheless, one of the main disadvantages of this method
is that the computational grid has to be sufficiently smooth in order to
maintain high order of accuracy on nonuniform meshes. Alternative discret-
ization techniques, such as Finite Volume and Finite Elements, have been
developed to overcome the drawbacks of the Finite Difference method.
Despite its limitations, the finite difference method is still the most popular
one currently used.

∆x

xi-2 xi-1 i xi+1 xi+2

FIGURE 10.41
Discretized domain.
7328_C010.fm Page 398 Monday, October 16, 2006 2:27 PM

398 Biofluid Mechanics: The Human Circulation

TABLE 10.1
Examples of Different Discretization Scheme of a First-Order Derivative
Schemes Formulas
Central second-order accurate formula
∂u ui +1 − ui −1
= + ϑ( ∆x 2 )
∂x 2∆x
First-order accurate, one-sided Forward
∂u ui +1 − ui
formula = + ϑ( ∆x)
∂x ∆x
Backward
∂u ui − ui −1
= + ϑ( ∆x)
∂x ∆x
Second-order one-sided Forward
∂u −3ui + 4ui +1 − ui + 2
formula = + ϑ( ∆x 2 )
∂x 2 ∆x
Backward
∂u 3ui − 4ui −1 + ui − 2
= + ϑ( ∆x 2 )
∂x 2 ∆x

10.9.3.2 Temporal Integration

The governing equation of fluid motion contains both spatial and temporal
derivatives and similar finite difference schemes are commonly used to
discretize the temporal derivatives. The system of governing equations is
integrated in time using suitable time-marching techniques. Several of these
time-marching techniques have been developed and can be classified into
two main categories: (1) explicit and (2) implicit methods. The general
concept of these two categories can be explained using, for instance, the
1-D linear advection-diffusion equation that governs the propagation of
waves subjected to dissipation

∂u ∂u ∂2 u
+c −υ 2 =0 (10.56)
∂t ∂x ∂x

where c represents the wavespeed and ν is the kinematic viscosity. This

equation can be discretized in space using finite difference schemes such as

∂u i u −u u −2 u i + u i −1
+ c i +1 i −1 − υ i +1 =0 (10.57)
∂t 2 ∆x ∆x 2

Using a first-order, backward-differencing scheme, the equation can also

be discretized in time

uni +1 − uni uni +1 − uni −1 uni +1 − 2 uni + uni −1

+c −υ =0 (10.58)
∆t 2 ∆x ∆x 2

where n denotes the time level and ∆t the time increment.

7328_C010.fm Page 399 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 399

Equation 10.58 can be rewritten as

 uni +1 − uni uni +1 − uni −1 uni +1 − 2 uni + uni −1 

uni +1 = uni − ∆t  +c −υ  (10.59)
 ∆t 2 ∆x ∆x 2 

It can be seen that if the solution is known at the time level n, the solution
at the time level n+1 can be simply evaluated using Equation 10.59. Such a
method is called explicit time-marching scheme. However, if one wants to
evaluate the spatial derivative at the time level n + 1, then Equation 10.57
becomes

uni +1 − uni uni ++11 − uni −+11 uni ++11 − 2 uni +1 + uni −+11
+c −υ =0 (10.60)
∆t 2 ∆x ∆x 2

It is clear that the solution for un+1

i cannot be calculated without knowing
the solutions of the surrounding nodes (in this case, uni ++11 and uni −+11 ). This par-
ticular time integration scheme therefore requires the solution of a coupled
set of equations. Such schemes are called implicit time-marching schemes and
several algorithms have been proposed to solve the coupled set of equations.
Once discretization schemes have been selected and all terms of the
Navier–Stokes equations have been discretized, the system of governing
equations can be solved to numerically obtain the flow fields in the geometry
of interest.

10.9.4 Computational Modeling of Blood Flows

Developing computational models of blood flow though the cardiovascular
system is challenging for the current generation of CFD software and com-
puter hardware. This difficulty stems from the fact that blood flow includes
several phenomena that are difficult problems in their own right. These
phenomena include the following.

10.9.4.1 Body Motion

Structure and fluid interactions are important physical phenomena that need
to be accurately computed to fully understand the fluid dynamics in the
cardiovascular system. For instance, the opening and closing motion of the
leaflets of the heart valves depends on the flow motion and, reciprocally,
the fluid motion through the heart valves depends on the position and
motion of the leaflets. Fluid–structure interactions are present throughout
the cardiovascular system, from the interaction of valve leaflets with the
blood to that of compliant vascular walls that distend with the blood pul-
satility. In some cases, veins may even collapse in normal function due to
their thin walls, creating a severe obstruction to the flow of blood.
7328_C010.fm Page 400 Monday, October 16, 2006 2:27 PM

400 Biofluid Mechanics: The Human Circulation

TABLE 10.2
Examples of Blood Vessel Parameters in Dogs and Humans
Canine Parameters [6] Human Parameters [25]
Mean Wall Mean Mean Wall
Mean Diameter Thickness Peak Reynolds Diameter Thickness
(mm) (mm) Number (mm) (mm)
Artery 4 (F) 0.4(F) 1000(F) 25 (AA) 1
5 (C) 0.3(C) 3000 (MP)
17 (MP) 0.2 (MP) 4500 (AA)
Arteriole 0.05 0.02 0.09 0.03 0.006
Capillary 0.006 0.001 0.001 0.008 0.0005
Venule 0.004 0.002 0.035 0.02 0.001
Vein 10 (IVC) 0.2 (IVC) 700 (IVC) 5 0.5
Note: MP – main pulmonary, F – femoral, IVC – inferior vena cava, C – carotid, AA –
ascending aorta.

10.9.4.2 Scale Disparity

There is wide disparity in length scales in the cardiovascular system. This
is highlighted in Table 10.2, where a few key blood vessel dimensions are
listed. It is clear that the length scales range from the order of 25 to 30 mm
(e.g., heart valve orifice, aorta diameter, etc.) for large-scale flows to length
scales of a few microns (e.g., hinge regions of mechanical heart valves,
capillaries, etc.) for small-scale flows. This disparity raises several issues
regarding model resolution, and the use of stretched computational grids is
often required to fully capture all the critical fluid structures. Additionally,
in large-scale vessels, the uneven vessel wall surface may be negligible,
whereas in the microcirculation, smooth muscles in the vessel wall constrict
to regulate the flow locally and, thus, may induce relatively large changes
in surface geometry. Computational issues may also arise from variation in
the flow’s time-scale. Flow time scales vary widely from that of the heart
cycle period (or greater) for the effects of fluid stresses on blood cells to a
few milliseconds during mechanical heart valve closure when cavitation may
occur and damage blood cells.

10.9.4.3 Blood Flow Properties

One of the most commonly used modeling simplifications is the assumption
that blood is Newtonian. While this assumption may hold for simulations
of flow through large vessels, the presence of individual blood cells and the
non-Newtonian character of the blood must be recognized in the microcir-
culation.

10.9.4.4 Pulsatility and Reynolds Number

Most numerical studies have focused on the simulation of steady flow fields.
While this may be a reasonably valid assumption in blood vessels, such as
7328_C010.fm Page 401 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 401

the veins, the pulsatility induced by the beating of the heart cannot be
neglected in most cases as it greatly affects the flow field. For instance,
experimental studies clearly indicate that the flow field through a fully open,
bileaflet mechanical heart valve under steady conditions is considerably
different from the flow field during the peak forward flow phase. Addition-
ally, as can be seen from Table 10.2, blood flow through the cardiovascular
system is subjected to a wide range of Reynolds numbers, indicating that
laminar, transitional, and even turbulent flows may be encountered. For
instance, blood flow through mechanical prosthetic heart valves can become
turbulent at peak forward flow rate and such turbulence is thought to play
an important role in clinical phenomena, such as platelet activation and
thrombus formation and embolization. Appropriate turbulence models are,
therefore, required to numerically evaluate such flow fields. On the other
hand, flow through capillaries is nonhomogeneous, unsteady, and is catego-
rized by a very small Reynolds number (<< 1). One of the main challenges
encountered in simulating the flow in the capillary circulation is the contin-
uous variation of blood flow in the capillaries. Laminarization and turbu-
lence are important flow phenomena, but have not yet been modeled
together. Numerical methods capable of allowing laminar and turbulent flow
to coexist in the same domain are needed. At present, variants of large-eddy
simulation (LES) techniques appear to be the most promising for predicting
the turbulence and relaminarization that occur during the cardiac cycle.

10.9.4.5 Symmetry
True geometries of blood vessels or of heart valves often show one or more
axes of symmetry. For simplicity, one may want to assume symmetry of the
flow and simulate only parts of the flow fields. However, previous studies
have shown that symmetry of the geometry does not always imply symmetry
of the flow fields. For instance, simulated flow fields through a bileaflet
mechanical heart valve showed that, despite the symmetry of the geometry,
symmetry of the flow could not be assumed and the entire cross section had
to be modeled, even at a low Reynolds number.

10.9.4.6 Grid Resolution

Studies have additionally shown that grid resolution is a chief parameter in
numerical studies. Adequate grid resolution is necessary to fully capture the
complexity of the flow structures and to obtain grid-independent solutions.
It is important to realize that many of the phenomena that researchers,
engineers, and clinicians want to predict using numerical models, namely
thrombosis formation and the final destination of emboli, require highly
detailed grid resolution. For instance, CFD-based predictions of wall shear
stress in the human right coronary artery are particularly sensitive to mesh
refinement and require many more grid nodes to achieve grid independence
than the velocity field alone.
7328_C010.fm Page 402 Monday, October 16, 2006 2:27 PM

402 Biofluid Mechanics: The Human Circulation

10.9.4.7 Validation
In vitro and in vivo studies have shown the complexity of flow in the cardio-
vascular system. Therefore, the capability of CFD codes to simulate such
complex flow needs to be tested rigorously by comparing their results with
detailed experimental measurements. There are several experimental studies
reported in the literature but, so far, only a few have been carried out with
CFD model development and validation in mind. For example, no experi-
ments have been carried out focusing on the details of the flow in realistic
mechanical heart valve geometries for Reynolds numbers over which lami-
nar flow conditions prevail. Even though such Reynolds numbers would not
be relevant from a physiological standpoint, the availability of a comprehen-
sive data set for laminar flow would greatly facilitate the validation of the
numerical method free of any turbulence modeling-related uncertainties.
Detailed experiments are also needed for Reynolds numbers in the physio-
logical range, which, unlike most previous work, should be driven by specific
needs for fine-tuning and validating advanced turbulence models. Ideally,
all numerical methods developed should be extensively validated, both qual-
itatively and quantitatively, using sophisticated experimental measurement
techniques.

10.9.5 Potential for Future Numerical Techniques

To date, cardiovascular hemodynamics has been studied largely by increas-
ingly complex experimental techniques, both in vitro and in vivo, which have
highlighted the complexity of the blood flow field throughout the cardio-
vascular system. However, experiments provide only limited information
about the flow field, such as three-dimensional velocity fields at very local-
ized sites. Accurate numerical studies could potentially provide the entire
three-dimensional picture of the flow fields. As seen previously, the cardio-
vascular system poses unique challenges to even the most advanced CFD
tool available today. However, growing computational resources, the advent
of massively parallel clusters (i.e., an assembly of several conventional CPUs
into a network to perform large-scale computer simulations), and the devel-
opment of sophisticated CFD algorithms have made it possible to attempt
to model such complex flows.
One of the major difficulties in applying numerical tools to cardiovascular
flows originates from the lack of understanding of the exact geometry of the
blood vessels. Today, advances in imaging techniques, such as magnetic
resonance imaging (MRI), computed tomography (CT), echocardiography,
and transesophageal echocardiography (TEE) techniques, have considerably
improved our ability to obtain accurate three-dimensional geometries of
blood vessels, heart valves, and cardiac chambers. For example, Figure 10.42
shows the generation of both numerical and experimental models from raw
MRI images. The raw images are first stacked (Figure 10.42a) and then
7328_C010.fm Page 403 Monday, October 16, 2006 2:27 PM

Fluid Dynamic Measurement Techiques 403

Stack of raw MR images Interpolated MR images Segmented geometry

(a) (b) (c)

10mm

Numerical model Experimental model

(d) (e)

FIGURE 10.42
Example of generation of computational and experimental models from raw MR images.

interpolated (Figure 10.42b). The blood vessel of interest, in this case, the
Total Cavo-Pulmonary Connection, is then segmented from the rest of the
image (Figure 10.42c). The obtained geometry can then be used to generate
both a computational mesh (Figure 10.42d) and an experimental Rapid Pro-
totype (Figure 10.42e) of the connection. Because this procedure can provide
identical experimental and computational geometries, it allows for an accu-
rate validation of the numerical tool to be made.
Previous blood flow studies have shown that extended exposure of blood
cells to high levels of shear stress leads to hemolysis, platelet activation,
and initiation of the coagulation cascade. Furthermore, aggregation of lysed
or activated blood cells and thrombus formation occur in regions of low flow
where the blood cell residence time is sufficiently elevated. Additionally,
data indicate that platelet activation depends on both shear stress levels
and exposure time while thrombus formation is a function of the residence
time. It is, therefore, essential to also assess these parameters in order to
7328_C010.fm Page 404 Monday, October 16, 2006 2:27 PM

404 Biofluid Mechanics: The Human Circulation

understand blood cell damage and thrombus formation. Such quantities

are, however, difficult to obtain using current experimental techniques and,
therefore, validated CFD models constitute the only alternative for accessing
such valuable information. For example, a Lagrangian particle tracking code
could be used to estimate the red blood cell trajectories and then provide
added insight into the influence of various flow structures to induce cell
damage, platelet activation, and thrombus formation.
7328_C011.fm Page 405 Monday, October 16, 2006 2:27 PM

Bibliography

Asakura, T., and Karino, T. (1990) Flow patterns and spatial distributions of
atherosclerotic lesions in human coronary arteries. Circulation Research
66:1054–1066.
Atkinson, P., and Woodcock, J. (1982) Doppler Ultrasound and Its Use in Clinical
Measurement, Academic Press, London.
Ballyk, P.D., Walsh, C., Butany, J., and Ojha, M. Compliance mismatch may promote
graft-artery intimal hyperplasia by altering suture-line stresses. J. Biomechan-
ics 1998. 31: 229–37.
Becker, G.J. Intravascular stents: General principles and status of lower-extremity
arterial applications. Circulation 1991. 83(Suppl I): I–122–I–136.
Bellhouse, B.J., and Bellhouse, F.H. Mechanism of closure of the aortic valve. Nature,
1968. 217(123): 86–7.
Bergel, D.H. (1961a) The static elastic properties of the arterial wall, J. Physiol., 156:
445–457.
Bergel, D.H. (1961b) The dynamic elastic properties of the arterial wall. J. Physiol.,
156: 458–469.
Berger, S.A., and Jou, L.A., Flows in stenotic vessels. Annual Reviews of Fluid
Mechanics, 2000. 32: 347–382.
Brown, M.M. Balloon angioplasty for extracranial carotid disease in Advances in
Vascular Surgery Vol. 4, 1996, A.D. Whittemore, Ed., Elsevier, Philadelphia.
Burton, A.C. (1954) Yearbook Publishers, Elsevier, New York.
Burton, A.C. (1971) Physiology and Biophysics of the Circulation. Year Book Medical
Publishers, Chicago.
Caro, C.G., Fitz-Gerald, J. M., and Schroter, R. C. (1971) Atheroma and arterial wall
shear. Observation, correlation and proposal of a shear dependent mass trans-
fer mechanism for atherogenesis. Proc. Royal Soc. London B. 177: 109–159.
Caro, C.G., Pedley, T.J., Schroter, R.C., and Seed, W.A. (1978) The Mechanics of the
Circulation, Oxford Medical Publications, Oxford.
Chandran, K.B. (1993) Flow dynamics in the human aorta. J. Biomech. Eng. 115: 611–615.
Cho, K.J., and Williams, D.M., (1991) Current Therapy in Vascular Surgery, 2nd ed.,
Ernst, C.B., and Stanley, J.C. Eds., B.C. Decker, Mosby Yearbook, Elsevier,
Philadelphia
Cooney, D.O. (1976) Biomedical Engineering Principles: An Introduction to Fluid, Heat
and Mass Transport Processes, Marcel Dekker, New York.
Copley, A.L., and Stainsby, G. (Eds.) (1960) Flow Properties of Blood, Pergamon Press,
Oxford
Dean, W.R. (1927) Note on the motion of fluid in a curved pipe. Phil. Mag. 20:
208–223.
DeArmond, S.J., Fusco, M.M., and Dewey, M.M. (1974) Structure of the Human Brain,
Oxford University Press, London.

405
7328_C011.fm Page 406 Monday, October 16, 2006 2:27 PM

406 Biofluid Mechanics: The Human Circulation

DeBakey, M.E., Lawrie, G.M., and Glaeser, D.H. Patterns of atherosclerosis and
their surgical significance. Ann. Surg., 1985. 201(2): 115–31.
Didisheim, P. and Watson, J.T. Cardiovascular Applications in Biomaterials Science:
An Introduction to Materials in Medicine, Ratner, B.D., Ed., 1996, Elsevier,
Philadelphia
Dinsdale, A., and Moore, F. (1962) Viscosity and Its Measurement, Reinhold Publishing,
New York.
Engeler, C.E., Edlicky, J., Letourneau, J.W., Castaneda-Auniga, J.G., Hunger, W.R.,
and Amplat, D. W. Intravascular sonography in the detection of arterioscle-
rosis and evaluation vascular interventional procedures. AJR, 1991, 156:
1087–1090.
Ernst, C.B., and Stanley, J.C. (1991) Current Therapy in Vascular Surgery, 2nd ed. B.C.
Decker, Philadelphia.
Farthing, S., and Peronneau, P. Flow in the thoracic aorta. Cardiovasc. Res., 1979.
13(11): 607–20.
Fields C, Cassano A, Allen C., Meyer A.C., Rittgers, S.E., Makhoul, R.G., and Bowlin
G.L. The persistence of electrostatically seeded endothelial cells lining a small
diameter expanded polytetrafluoroethylene vascular graft. J. Biomaterials
Applications 16: 157–173, 2001
Frazin, L.J., Lanza, G., Vonesh, M., Khasho, F., Spitzzeri, C., McGee, S., Mehlman,
D., Chandran, K.B., Talano, J., and McPherson, D.D. (1990) Functional chiral
asymmetry in descending thoracic aorta. Circulation 82: 1985–1994.
Fry, D.L. (1968) Acute vascular endothelial changes associated with increased blood
velocity gradients. Circulation Research 22: 165–197.
Fung, Y.C. (1984) Biodynamics: Circulation, Springer-Verlag, New York.
Geddes, L.A. (1984) Cardiovascular Devices and Applications, John Wiley & Sons,
New York.
Ghista, D.N., and Rao, A.P. Structural mechanics of the mitral valve: stresses sus-
tained by the valve; non-traumatic determination of the stiffness of the in
vivo valve. J. Biomech., 1972. 5(3): 295–307.
Gotto, A.M. Jr., Robertson, A.L. Jr., Epstein, S.E., DeBakey, M.E., and McCollum,
C.H. III. (1977) Atherosclerosis, Upjohn, Pfizer, Inc., New York.
Gupta, S.K., (1993), Vascular Graft Monitoring, R.G. Landes Co., Austin, Texas
Guyton, A.C., and Hall, J.E. (2000) A Textbook of Medical Physiology, 10th ed., W.B.
Saunders, Philadelphia.
Gwertzman, G. Clinically tested intravascular stents and endovascular stent grafts
in Endovascular Stented Grafts for the Treatment of Vascular Diseases, Marin, M.L.,
Veith F.J., and Levine, B.A. Eds., 1995, R.G. Landes Co., Austin, Texas
Hatle, L., and Angelson, B. (1985) Doppler Ultrasound in Cardiology: Physical Principles
and Clinical Applications, Lea and Febiger, Philadelphia.
He, X., and Ku, D.N. Pulsatile flow in the human left coronary artery bifurcation:
average conditions. J. Biomech. Eng., 1996. 118(1): 74–82.
Helmlinger, G., Geiger, R.V., Schreck, S., and Nerem, R.M. Effects of pulsatile
flow in cultured vascular endothelial cell morphology. J. Biomech. Engr., 1991.
113: 123–131.
Hibbeler, R.C. (2003) Mechanical of Materials, 5th ed., Pearson Prentice Hall, Upper
Saddle River, NJ.
Holzapfel, G.A., Gasser, T.C., and Ogden, R. W. (2000) A new constitutive frame-
work for arterial wall mechanics and a comparative study of material models.
J. Elasticity 61: 1–48.
7328_C011.fm Page 407 Monday, October 16, 2006 2:27 PM

Bibliography 407

How, T.V., Rowe, C.S., Gilling-Smith, G.L., and Harris, P. Interposition vein
cuff anastomosis alters wall shear stress distribution in the recipient
artery, J. Vascular Surgery 2000. 31 : 1008–17.
Ionescu, M.I., and Ross, D.N. Heart-valve replacement with autologous fascia lata.
Lancet, 1969. 2(7616): 335–8.
Jacob, S.W., and Francone, C.A. (1974) Structure and Function in Man, 3rd ed., W.B.
Saunders Company, Philadelphia.
Kaiser, G.A., Hancock, W.D. Lukban, S.B., and Litwak, R.S. Clinical use of a new
design stented xenograft heart valve prosthesis. Surg Forum, 1969. 20: 137–8.
Keynton, R.S., Rittgers, S.E., and Shu, M.C.S. The effect of angle and flow rate upon
hemodynamics in distal vascular graft anastomosis: An in vitro model study.
J. Biomech. Eng,. 1991. 113: 458–463.
Keynton, R.S., Evancho, M.M., Sims, R.L., and Rittgers, S.E. The effect of graft
caliber upon wall shear within in vivo distal vascular anastomosis. J. Biomech.
Engr., 1999. 121: 79–88.
Keynton, R.S., Evancho, M.M., Sims, R.L., Rodway, N.V., Gobin, A., and Rittgers,
S.E. Intimal hyperplasia and wall shear in arterial bypass graft distal anas-
tomoses: An in vivo model study, J. Biomech. Engr. 2001; 123: 464–73.
Kim, W.Y., Bisgaard, T., Nielsen, S.L., Poulsen, J.K., Pedersen, E.M., Hasenkam,
J.M., and Yoganathan, A.P. Two-dimensional mitral flow velocity profiles in
pig models using epicardial Doppler echocardiography. J. Am. Coll. Cardiol.,
1994. 24(2): 532–45.
Kleinstreuer, C., Buchanan, Jr., J.R., Lei, M., and Truskey, G.A. Computational
analysis of particle-hemodynamics and prediction of the onset of arterial
diseases, in Cardiovascular Techniques, Leondes, C. Ed. 2001, CRC Press: Boca
Raton. I–1 to I–69.
Ku, D.N., Giddens, D.P., Zarins, C.K., and Glagov, S. Pulsatile flow and atheroscle-
rosis in the human carotid bifurcation. Positive correlation between plaque
location and low oscillating shear stress. Arteriosclerosis, 1985. 5(3): 293–302.
Ku, D. N. (1997) Blood flow in arteries, Ann. Rev. Fluid Mech. 29: 399–434.
Kunzelman, K.S., Cochran, R.P., Chuong, C., Ring, W.S., Verrier, E.D., and Eberhart, R.D.
Finite element analysis of the mitral valve. J. Heart Valve Dis. 1993. 2(3): 326–40.
Langille, B.L., and O’Donnell, F. (1986) Reductions in arterial diameter produced by
chronic decreases in blood flow are endothelium-dependent, Science 231: 405–407.
Leo, H.L., He, Z., Ellis, J.T., and Yoganathan, A.P. Microflow fields in the hinge
region of the CarboMedics bileaflet mechanical heart valve design. J. Thorac.
Cardiovasc Surg., 2002. 124(3): 561–74.
Li, J. K-J. (1987) Arterial System Dynamics, New York University Press, New York.
Li X.M, Evaluation of hemodynamic factors at the distal end-to-side anastomosis
of a bypass graft with different POS:DOS ratios, Ph.D. Dissertation, Dept. of
Biomedical Engineering, The University of Akron, 1998
Lim, K.O., and Boughner, D.R. Mechanical properties of human mitral valve
chordae tendineae: variation with size and strain rate. Can. J. Physiol Pharma-
col., 1975. 53(3): 330–9.
LoGerfo, F.W., Nowak, M.D., and Quist, W.C. Structural details of boundary layer
separation in a model human carotid bifurcation under steady and pulsatile
flow conditions. J. Vasc. Surg. 1985. 2: 263–269.
Lutz, R. J., Hsu, L., Menawat, A., Zrubek, J., and Edwards, K. Comparison of steady
and pulsatile flow in a double branching arterial model. J. Biomech. 1983. 16:
753–766.
7328_C011.fm Page 408 Monday, October 16, 2006 2:27 PM

408 Biofluid Mechanics: The Human Circulation

Malek, A.M., Alper, S.L., and Izumo, S. (1999) Hemodynamic shear stress and its
role in atherosclerosis. JAMA. 282(21): 2035–42.
Marcus, M.L. (1983) The Coronary Circulation in Health and Disease, McGraw-Hill,
New York.
McDonald, D.A. (1974) Blood Flow in Arteries. Williams and Wilkins, Baltimore.
Merrill, E.W., Margetts, W.G., Cokelet, G.R., and Gilliland, E.R. (1965) The Casson
equation and rheology of blood near zero shear. In Proceedings of the Fourth
International Congress on Rheology (Copley, A.L. Ed.), Wiley-Interscience, NY,
135–143.
Milnor, W.R. (1989) Hemodynamics, 2nd ed., Williams and Wilkins, Baltimore Morgan,
G.W., and Kiely, J.P. (1954) Wave propagation in a viscous liquid contained in
a flexible fluid, J. Acoust. Soc. Am., 26: 323–328.
Moore, J.E., Jr., and Berry, J.L. (2002) Fluid and solid mechanical implications of
vascular stenting. Ann. Biomed. Eng. 30: 498–508.
Nanda, N.C. (1985) Doppler Echocardiography, Igaku-Shoin, New York.
Nichols, W.W., and O’Rourke, M.F. (1990) McDonald’s Blood Flow in Arteries, 3rd Ed.
Lea and Febiger, Philadelphia.
Noordergraaf, A. (1978) Circulatory Systems Dynamics, Academic Press, NY.
Ojha, M., Ethier, C.R., Johnson, K.W., and Cobbold, R.S.C. Influence of angle on
wall shear stress distribution for an end-to-side anastomosis model, J. Vasc.
Surg. 1990. 12: 747–53.
O'Rourke, M.F., Vascular impedance in studies of arterial and cardiac function.
Physiol Rev, 1982. 62(2): 570–623.
Papadaki, M., and Eskin, S.G. Effects of fluid shear stress on gene regulation of
vascular cells. Biotechnol Prog. 1997. 13(3): 209–21.
Paulsen, P.K., and Hasenkam, J.M. Three-dimensional visualization of velocity
profiles in the ascending aorta in dogs, measured with a hot-film anemometer.
J. Biomech., 1983. 16(3): 201–10.
Peterson, L.H., Jensen, R.E., and Parnell, J. (1960) Mechanical properties of arteries
in vivo. Circulation Research 8: 622–639.
Poots, K., Cobbold, R.S.C., Johnson, K.W., Appugliese, R., Kassam, M., Zuech, P.E.,
and Hummel, R.L. A new pulsatile flow visualization method using a pho-
tochromic dye with application to Doppler ultrasound. Ann. Biomedical En-
gineering, 1986. 14: 203–218.
Reul, H., Talukder, N., and Muller, E.W. Fluid mechanics of the natural mitral valve.
J. Biomech., 1981. 14(5): 361–72.
Riley, W.F., Sturges, L.D., and Morris, D.H. (1998) Mechanics of Materials, 5th ed., John
Wiley & Sons, New York.
Rittgers, S.E., and Shu, M.C.S. Doppler color-flow images from a stenosed arterial
model: Interpretation of flow patterns. J. Vasc. Surg. 1990. 12: 511–22,
Rushmer, R.F. (1976) Cardiovascular Dynamics, W. B. Saunders, Philadelphia.
Salisbury, P.F., Cross, C.E., and Rieben, P.A. Chorda Tendinea Tension. Am. J. Phys-
iol., 1963. 205: 385–92.
Sawyer, P.N., (1987) Modern Vascular Grafts, McGraw-Hill Book Co., New York.
Scott, M.J., and Vesely, I. Morphology of porcine aortic valve cusp elastin. J. Heart
Valve Dis., 1996. 5(5): 464–71.
Schlant, R C., and Alexander, R.W., Eds. (1994) Hurst’s The Heart, Arteries and Veins,
8th ed., McGraw-Hill, Health Professions Division, New York.
Seed, W. A., and Wood, N. B. (1971) Velocity patterns in the aorta. Cardiovascular
Research 5: 319–330.
7328_C011.fm Page 409 Monday, October 16, 2006 2:27 PM

Bibliography 409

Senning, A., Fascia lata replacement of aortic valves. J. Thorac. Cardiovasc Surg.,
1967. 54(4): 465–70.
Senning, A., Results following aortic valve replacement with autologous Fascia lata.
Thoraxchir. Vask. Chir., 1971. 19(4): 304–8.
Silverthorn, D.U. (2001) Human Physiology. An Integrated Approach, 2nd ed., Prentice
Hall, Upper Saddle River, NJ.
Sloth, E., Houlind, K.C., Oyre, S., Kim, W.Y., Pedersen, E.M., Jorgensen, H.S., and
Hasenkam, J.M. Three-dimensional visualization of velocity profiles in the
human main pulmonary artery with magnetic resonance phase-velocity map-
ping. Am. Heart. J., 1994. 128: 1130–8.
Texon, M. (1957) A hemodynamic concept of atherosclerosis, with particular refer-
ence to coronary occlusion. Arch. Int. Med. 99: 418–427.
Thubrikar, M., Piepgrass, W.C., Shaner, T.W., and Nolan, S.P. The design of the
normal aortic valve. Am. J. Physiol., 1981. 241(6): H795–801.
Travis, B.R., Leo, H.L., Shah, P.A., Frakes, D.H., and Yoganathan, A.P. An analysis
of turbulent shear stresses in leakage flow through a bileaflet mechanical
prostheses. J. Biomech. Eng., 2002. 124(2): 155–65.
Tsakiris, A.G., Von Bernuth, G., Rastelli, G.C., Bourgeois, M.J., Titus, J.L., and Wood,
E.H. Size and motion of the mitral valve annulus in anesthetized intact dogs.
J. Appl. Physiol., 1971. 30(5): 611–8.
Vesely, I., and Noseworthy, R. Micromechanics of the fibrosa and the ventricularis
in aortic valve leaflets. J. Biomech., 1992. 25(1): 101–13.
Vesely, I., Macris, N., Dunmore, P.J., and Boughner, D. The distribution and mor-
phology of aortic valve cusp lipids. J. Heart Valve Dis., 1994. 3(4): 451–6.
Villemarette, P.A., Kornick, A.L., Rosenberg, D.M., Holmes, J., and Hower, J.F. Jr.
Use of color flow Doppler to evaluate vascular access graft function, J. Vasc.
Tech., 1989. 13: 164–170.
White, R.A., and Cavaye, D.M. (1993), in A Text and Atlas of Arterial Imaging —
Modern and Developing Technology, Cavaye, D.M., and White, R.A., Eds., W.
W. Lippincott, Philadelphia.
Whitmore, R.L. (1968) Rheology of Circulation, Pergamon Press, New York.
Womersley, J.R. (1955a) Method for the calculation of velocity, rate of flow and
viscous drag in arteries when the pressure gradient is known, J. Physiol. 127:
553–563.
Womersley, J.R. (1955b) Oscillatory motion of a viscous liquid in a thin-walled elastic
tube. I. The linear approximation for long waves. Phil. Mag. 46: 199–221.
Womersley, J.R. (1957a) The mathematical analysis of arterial circulation in a state of
oscillatory motion. Wright Air Development Center, Technical Report WADC-
TR-56-614.
Womersley, J.R. (1957b) Oscillatory flow in arteries: the constrained elastic tube as a
model of arterial flow and pulse transmission, Phys. Med. Biol. 2: 178–187.
Yellin, E.L., Peskin, C., Yoran, C., Koenigsberg, M., Matsumoto, M., Laniado,
S., McQueen, D., Shore, D., and Frater, R.W. Mechanisms of mitral valve
motion during diastole. Am. J. Physiol, 1981. 241(3): H389–400.
Yoganathan, A.P., Cardiac Valve Prostheses, in The Biomedical Engineering Handbook,
Bronzino, J.D. Ed. 2000, CRC Press: Boca Raton, FL.
Yoganathan, A.P., Eberhardt, C.E., and Walker, P.G. Hydrodynamic performance of
the Medtronic Freestyle Aortic Root Bioprosthesis. J. Heart Valve Dis. 1994.
3(5): 571–80.
Young, D.F. (1979) Fluid mechanics of arterial stenosis. J. Biomech. Eng. 101: 157–173.
7328_C011.fm Page 410 Monday, October 16, 2006 2:27 PM
7328_Index Page 411 Tuesday, October 10, 2006 11:57 AM

Index

A Basement membrane, 87
Bernoulli equation, 26–29, 169–178
Abdominal aortic aneurysm, 172, 251, 319 aneurysms, 170–172
Advanced lesions, 107 arterial stenoses, 170–172
Adventitia, 87, 152 cardiac valve stenoses, 172–178
Adverse pressure gradients, 44 fluid dynamics, 26–29
Aliasing, 371 total vs. hydrostatic pressure measurement,
Alveolar epithelium, 97 169–170
Anastomoses, 317 Bifurcations and branches, flow through,
Anastomotic intimal hyperplasia, 324–333 236–241
Aneurysms, 170–172 Bileaflet valve, prosthetic heart valve, 299–302
flow through, 241–253 Bingham plastics, 6
formation of, 251 Björk–Shiley, 282
Angina pectoris, 93 Blood damage, prosthetic heart valve, in vitro
Angiogenin, 104 studies, 308–309
Angioplasty, 333–338 Blood vessel mechanics, 152–165
Angiotensin II, 105 cardiac muscle, material characterization
Aortic valve, 257, 259–267 of, 164–165
mechanical properties, 261–263 material behavior, 155–161
valve dynamics, 263–267 residual stress, 162–164
Aortoiliac bifurcation, 236 wall of blood vessel, structural components,
Apparent viscosity, 6 152–155
Arterial stenoses, 170–172 Boundary layer separation, 39, 44–45,
flow through, 241–253 183
Arteriovenous fistulas, 317–319 fluid dynamics, 44–45
Atherogenesis, hemodynamic theories, Bragg cell, 375
222–228 Bulk modulus, 7, 56–57, 193
low pressure, low, high wall shear stress Bundle of His, 74
theories, 223–227
oscillatory shear index, 227–228
time varying wall shear stress, 227–228 C
wall shear stress gradients, 227–228
Atherogenic, defined, 225 Caged ball valve, prosthetic heart valve,
Atheroprotective, defined, 225 298–299
Atherosclerosis, 94, 105–110 Capillary endothelial cell layer, 97
lesion growth, sequence, 107–110 Capillary viscometer, 116–123
morphology of, 106–107 Cardiac conduction, 74–77
Atherosclerotic plaque, 105, 107 Cardiac cycle, 76
Cardiac function, 77–82
Cardiac muscle, material characterization,
B 164–165
Cardiac structure, 73–74
Back-scattering mode, 375 Cardiac valves, 82–84
Baroreceptors, 104 stenoses, 172–178

411
7328_Index Page 412 Tuesday, October 10, 2006 11:57 AM

412 Biofluid Mechanics: The Human Circulation

Cardiovascular physiology, 69–112 symmetry, 401

atherosclerosis, 105–110 validation, 402
lesion growth, sequence, 107–110 discretization techniques, 396
morphology of, 106–107 equations, 390–393
cardiac valves, 82–84 finite difference method, 396
cerebral circulation, 98–99 future numerical techniques, 402–404
coronary circulation, 91–94 grid generation, 393–396
heart, 70–82 temporal integration, 398–399
cardiac conduction, 74–77 Cone and plate viscometer, 116–123
cardiac function, 77–82 Conservation of mass, 11–13, 15–18, 241,
cardiac structure, 73–74 251–252
lungs, gas exchange in, 94–98 Conservation of momentum, 13–14, 18–22
microcirculation, 99–103 Continuous-wave, defined, 364
pulmonary circulation, 94–98 Continuous-wave Doppler ultrasound
regulation of circulation, 103–105 velocimeter, 290
renal circulation, 99 Continuum models, pulsatile flow dynamics,
systemic circulation, 84–91 195–222
Carotid sinus, 239 elastic tube, wave propagation in, 213–222
Carpentier–Edwards bioprosthesis, 286 entrance effects, 221
Carpentier–Edwards pericardial valve, 287 homogeneous, isotropic, and Hookean
Casson fluids, 7 material, 222
Casson’s relationship, 179 laminar flow, 220
Cavitation, prosthetic heart valve, 306–307 linearization of equations, 221
Cavitation bubbles, 306 Newtonian fluid, 220
CDFM. See Color Doppler flow mapping reflected waves, 221
Celia, 95 thin-walled tube, 222
Central venous pressure, 90 uniform cylindrical tube, 220–221
Cerebral circulation, 98–99 Moens–Korteweg relationship, 196–205
CFD. See Computational fluid dynamics wave propagation, arterial system, 195–222
Characteristic impedance, 353 Womersley model, 205–213
Chordae tendinae, 84, 267 Contraction coefficient, 173
Chronic obstructive pulmonary disease, 94 Control volume, 11, 13
Circle of Willis, 98 Convection, 18
Circulation Convective acceleration, 21
hydrostatics in, 168–169 COPD. See Chronic obstructive pulmonary
regulation of, 103–105 disease
Closing volume, 294 Coronary circulation, 91–94
Coagulation potential, prosthetic heart valve, Couette, 120
in vitro studies, 308–309 Creep, 65
Coaxial cylindrical viscometer, 116–123 Cubical dilatation, 56
Collapsible vessels, flow in, 185–189 Curved vessels, flow through, 231–236
Color Doppler flow mapping, 372 CVP. See Central venous pressure
Commisures, 268 Cylindrical shape, rigid tube flow models,
Compatibility condition, thick-walled 179–180
cylindrical tube analysis, 61–63
Compressibility, as intrinsic fluid property, 7
Computational fluid dynamics, 26, 390–396 D
blood flow, computational modeling,
399–402 Dean number, 234
blood flow properties, 400 Deep vein thrombosis, 90
bodily motion, 399 Density, as intrinsic fluid property, 4
grid resolution, 401 Design trends, prosthetic heart valves, 312
pulsatility, 400–401 Diastole, 77
Reynolds number, 400–401 Dichrotic notch, 88
scale disparity, 400 Dimensional analysis, 30–32
7328_Index Page 413 Tuesday, October 10, 2006 11:57 AM

Index 413

fluid dynamics, 30–32 ESV. See End systolic volume

Dimensionless continuity equation, 31 Euler’s equation of motion, 26
Dimensionless form, 31 Excitation, 383
Discharge coefficient, 174
Doppler burst, 377
Drag coefficients, 14 F
Drug-eluting, 342
Ductile, 49 Fahraeus–Lindqvist effect, 135
Duplex, 372 Fast Fourier transformation, 363
Durability, prosthetic heart valve, 309–312 Fatty streak, 107
fatigue, 310–311 FFT. See Fast Fourier transformation
mineralization, 311–312 FGF. See Fibroblast growth factor
wear, 310 Fibrils, 164
DVT. See Deep vein thrombosis Fibroblast growth factor, 104
Dynamic similarity, 31 Fibrosa, 259
Dynamic viscosity, 5 Filling phase, defined, 80
Fisk’s law, 97
Flow divider, 236
E Flow limitation, 186
Flow visualization, 356–359
EDRF. See Endo-thelial-derived relaxation Fluid dynamic measurement, 343–408
factor blood flow, computational modeling
EDV. See End diastolic volume blood flow properties, 400
Edwards-Duromedics, 310 bodily motion, 399
Edwards prima, 288 grid resolution, 401
Effective orifice area, 174 pulsatility, 400–401
Ejection phase, 80 Reynolds number, 400–401
Elastase, 171, 251 scale disparity, 400
Elastic behavior, 47–50 symmetry, 401
Elastic modulus, 49 validation, 402
Elastin, 251 blood flow measurement, 347–352
Electromagnetic flow meter, 347 computational fluid dynamics, 390–396
Emboli, 105 blood flow, computational modeling,
Embolizes, 241 399–402
EMF meter. See Electromagnetic flow discretization techniques, 396
End diastolic volume, 80 equations, 390–393
End systolic volume, 80 finite difference method, 396
End-to-side, 317 future numerical techniques,
Endo-thelial-derived relaxation factor, 103 402–404
Endocardium, 164 grid generation, 393–396
Endoluminal graft implants, 333–338 temporal integration, 398–399
Endoluminal stent graft, 336 flow visualization, 356–359
Endothelial cells, 40 impedance measurements, 352–356
seeding, 320 laser Doppler velocimetry, 374–381
Engineering, 50–51 dimensions, calculation of, 376–377
Entrance length fringe patterns, 377–379
defined, 39 phase window averaging, LDV data,
effect on arterial flow development, 182–185 380–381
EOA. See Effective orifice area photodetectors, 379
Epicardium, 92, 164 probe volume specifications, 376–379
Epithelial basement membrane, 97 signal processing, 379–380
EPTFE, 319 magnetic resonance imaging/velocity
Equations of motion, 22 mapping, 381–390
Equilibrium equation, thick-walled gradient-echo, 387–388
cylindrical tube analysis, 60–61 imaging procedure, 385–388
7328_Index Page 414 Tuesday, October 10, 2006 11:57 AM

414 Biofluid Mechanics: The Human Circulation

magnetic resonance phase velocity Hydrostatics, 9–10, 168–169

mapping, 388–390 Hypertension, 105
pulse sequences, 385–388 Hypertrophy, 161
slice excitation, 383–384
spatial encoding, 384–385
spin-echo, 386–387 I
pressure measurement, 344–347
ultrasound Doppler velocimetry, IH. See Intimal hyperplasia
359–374 IMA. See Internal mammary artery
Fluid-filled catheter manometer, 169 Image reconstruction, 383
Frank–Starling, 104 Impedance measurements, 352–356
Frequency encoding, 384 Improved durability, 313
Fully developed flow, rigid tube flow models, Improved hemodynamic characteristics, 313
180 Incremental elastic modulus, 51
Fully developed flow region, 38 Incremental modulus, 51, 63
Internal mammary artery, 319
Intima, 87, 152
G Intimal hyperplasia, 107, 323
anastomotic, 324–333
Gas exchange in lungs, 94–98 Intrinsic fluid properties, 4–8
Generalized Hooke’s law, 53–56 compressibility, 7
Gorlin equation, 175 density, 4
Gradient-echo, 385 surface tension, 8
Growth factors, 229 viscosity, 4–7
Ionescu–Shiley pericardial xenograft, 287
Ischemia, 249
H Isotropic, defined, 49
Isovolumic contraction phase, 80
Hagen–Poiseuille law, 36, 45–46, 84, 86, 178
Hancock modified orifice, 287
Hancock Porcine xenograft, 286 K
Haynes 25 Stellite Björk-Shiley, 311
HDL. See High density lipoproteins Kelvin model, 65
Heart, 70–82 Korotkov sounds, 171
conduction, 74–77
function, 77–82
structure, 73–74 L
valves, 82–84
Hemodynamic assessment, prosthetic heart Laminar flow, rigid tube flow models, 179
valve, 289–308 LaPlace’s equation, 251
Hemolysis, 40 Larmor frequency, 382
with fluid dynamically induced stresses, Laser Doppler anemometry, 374
150–152 Laser Doppler velocimetry, 374–381
High density lipoproteins, 107 phase window averaging, LDV data,
High intensity transient signals, 307 380–381
High shear stress, 223 photodetectors, 379
Hinge flow, 304 probe volume specifications, 376–379
History of prosthetic heart valve, 279–289 dimensions, calculation of, 376–377
mechanical valves, 279–284, 288 fringe patterns, 377–379
tissue valves, 284–288 signal processing, 379–380
HITS. See High intensity transient signals Leakage flow, 304
Homografts, 284 prosthetic heart valve, 304–306
Hooke’s law, 53–56 Leakage jets, 305
Hoop stress, 58 Leakage volume, 294
Hydrogen bubble, 359 Linear momentum equation, 14
Hydrostatic pressure measurement, 169–170 Local acceleration, 21
7328_Index Page 415 Tuesday, October 10, 2006 11:57 AM

Index 415

Longitudinal stress, 58 Myocardial infarction, 94

Low density lipoproteins, 106 Myocardium, 71
Low pressure, 223
high wall shear stress theory, atherogenesis,
223–227 N
Lungs, gas exchange in, 94–98
Lunula, 260 Native heart valves, 257–274
aortic valve, 259–267
mechanical properties, 261–263
valve dynamics, 263–267
M mitral valve, 267–273
mechanical properties, 269–270
Macroscopic balances, mass/momentum,
valve dynamics, 270–273
10–14
pulmonic valve, 259–267
conservation of mass, 11–13
mechanical properties, 261–263
conservation of momentum, 13–14
valve dynamics, 263–267
Magnetic resonance imaging/velocity
tricuspid valve, 267–273
mapping, 381–390
mechanical properties,
imaging procedure, 385–388
269–270
gradient-echo, 387–388
valve dynamics, 270–273
spin-echo, 386–387
Navier–Stokes equation, 14, 22, 26, 31, 116,
magnetic resonance phase velocity
178, 390
mapping, 388–390
Newtonian fluid, rigid tube flow models,
pulse sequences, 385–388
178–179
slice excitation, 383–384
Nitinol, 335
spatial encoding, 384–385
Nodes, 25
Magovern prosthesis, 279
Non-Newtonian fluids, pressure-flow
Mass, 10–11
relationship, 139–150
conservation of, 11–13, 15–18
Bingham plastic, 141–144
macroscopic balance, 10–14
Casson’s fluid, 144–150
microscopic balance, 15–26
power law fluid, 140–141
Mass conservation principle, 38
Nonuniform geometric models, 191–256
Materials mechanics. See Solid mechanics
Normal stresses, 21
Maxwell model, 64
Nyquist sampling limit, 371
Medtronic freestyle, 288
Medtronic Hall, 282
Medtronic Mosaic, 287
MI. See Myocardial infarction O
Microcirculation, 99–103
Omniscience, 282
Microscopic balance, 15–26
On-X, 284
mass/momentum, 15–26
OPCAB, 317
conservation of mass, 15–18
Oscillatory flow, 33
conservation of momentum, 18–22
Oscillatory shear index, 227–228
mathematical solutions, 22–26
OSI. See Oscillatory shear index
MIDCAB, 317
Oxy-hemoglobin, 97
Mitral valve, 257, 267–273
mechanical properties, 269–270
valve dynamics, 270–273 P
Modulus of compression, 57
Modulus of rigidity, 53 P wave, 77
Moens–Korteweg relationship, pulsatile flow Papillary muscles, 84, 267
dynamics, 196–205 Paracrines, 107
Momentum, 10 Parasympathetic, 104
conservation of, 13–14, 18–22 Particle image velocimetry, 356
macroscopic balance, 10–14 Patency, 322
microscopic balance, 15–26 Pathlines, 356
7328_Index Page 416 Tuesday, October 10, 2006 11:57 AM

416 Biofluid Mechanics: The Human Circulation

Percutaneous transluminal angioplasty, 334 pressure drop, 290–291

Pericardial valve, prosthetic heart valve, 303 regurgitation, 294–296
Peripheral resistance units, 85 proximal flow convergence, 295–296
Perivascular leaks, 337 turbulent jet theory, 295
PET. See Polyethylene terephthalate shear stresses, 296–303
Phagocytosis, 107 thrombus deposition, 308–309
Phase domain demodulation, 363 tilting disc valve, 299
Phase encoding, 384 Proximal flow convergence, 295
Phase quadrature demodulation, 362 PRUs. See Peripheral resistance units
Phase velocity mapping, 390 PTA. See Percutaneous transluminal
Photochromic dye, 359 angioplasty
Photomultipliers, 379 PTFE. See Polytetrafluoroethylene
Physical properties of blood, 123–124 Pulmonary circulation, 94–98
PIV. See Particle image velocimetry Pulmonic, defined, 257
Plane stress analysis, 55 Pulmonic valve, 259–267
Plaque cap rupture, 249 mechanical properties, 261–263
Plasma, 123 valve dynamics, 263–267
Platelet activation, with fluid dynamically Pulsatile flow dynamics, continuum models,
induced stresses, 150–152 195–222
Poiseuille flow, 116, 178 elastic tube, wave propagation in, 213–222
Poisson’s ratio, 51–52 entrance effects, 221
Polyethylene terephthalate, 319 homogeneous, isotropic, and Hookean
Polytetrafluoroethylene, 319 material, 222
Porcine valve, prosthetic heart valve, 302–303 laminar flow, 220
Power law, 6 linearization of equations, 221
Pressure diuresis, 105 Newtonian fluid, 220
Pressure drop, prosthetic heart valve, reflected waves, 221
290–291 thin-walled tube, 222
Pressure measurement, 344–347 uniform cylindrical tube, 220–221
Pressure-strain modulus, 63 Moens–Korteweg relationship, 196–205
PRF. See Pulse repetition frequency wave propagation, arterial system, 195–222
Principle of mass conservation, 38 Womersley model, 205–213
Principle of superposition, 54 Pulse Doppler, 367
Prosthetic heart valves, 277–314 Pulse pressure, 88
bileaflet valve, 299–302 Pulse repetition frequency, 369
blood damage, in vitro studies, 308–309 Purkinje fibers, 74
caged ball valve, 298–299 PVM. See Phase velocity mapping
cavitation/HITS, 306–307
coagulation potential, in vitro studies,
308–309 Q
current types of prostheses, 288–289
design trends, 312 QRS complex, 77
durability, 309–312 Quasi-steady, 43
fatigue, 310–311
mineralization, 311–312
R
wear, 310
flow patterns, 296–303 Range resolution, 367
hemodynamic assessment, 289–308 Rate of shear, 5
history of, 279–289 Reactive hyperemia, 93, 103
mechanical valves, 279–284, 288 Regulation of circulation, 103–105
tissue valves, 284–288 Regurgitant volume, 294
leakage flow, 304–306 Regurgitation, prosthetic heart valve,
orifice area, effective, 292–294 294–296
pericardial valve, 303 proximal flow convergence, 295–296
porcine valve, 302–303 turbulent jet theory, 295
7328_Index Page 417 Tuesday, October 10, 2006 11:57 AM

Index 417

Renal circulation, 99 Side scattering mode, 375–376

Resistance, vascular, 180–182 Signal read-out, 383
Restenosis, 334 Simplified Bernoulli equation, 28
Reynolds number, 34 Sinus of valsalva, 82, 260
Reynolds transport theorem, 10, 13 Sluice effect, 186
Rheology of blood, 116–152 Smeloff–Cutter valve, 279
capillary viscometer, 116–123 Solid mechanics, 47–68
coaxial cylindrical viscometer, 116–123 bulk modulus, 56–57
cone and plate viscometer, 116–123 elastic behavior, 47–50
hemolysis, with fluid dynamically induced engineering, 50–51
stresses, 150–152 generalized Hooke’s law, 53–56
non-Newtonian fluids, pressure-flow incremental elastic modulus, 51
relationship, 139–150 Poisson’s ratio, 51–52
Bingham plastic, 141–144 shearing stresses, strains, 53
Casson’s fluid, 144–150 thick-walled cylindrical tube analysis,
power law fluid, 140–141 60–63
physical properties of blood, 123–124 compatibility condition, 61–63
platelet activation, with fluid dynamically equilibrium equation, 60–61
induced stresses, 150–152 thin-walled cylindrical tube analysis, 57–60
viscometry, 116–123 true strain, 50–51
viscous behavior of blood, 124–139 viscoelasticity, 63–67
hematocrit, effect of, 130 Sorin bicarbon, 283
plasma, 124–126 Spatial encoding, 383
protein content in plasma, effect of, 131 Spatial wall shear stress gradient, 228
sigma effect, 137–139 Specific gravity, 4
temperature, effect of, 130–131 Spin-echo, 385
tube diameter, effect of, 135–139 Spongiosa, 260
whole blood, 126–130 Squeeze flow, 304
yield stress for blood, 131–135 St. Jude Medical Regent, 283
Rigid tube flow models, 178–182 Starling resistor phenomenon, 186–187
cylindrical shape, 179–180 Starr–Edwards ball-and-cage valve, 279
fully developed flow, 180 Static/steady flow models, 167–190
laminar flow, 179 Bernoulli equation, 169–178
Newtonian fluid, 178–179 aneurysms, 170–172
no slip at vascular walls, 179 arterial stenoses, 170–172
rigid wall, 180 cardiac valve stenoses, 172–178
steady flow, 179 total vs. hydrostatic pressure
Rigid wall, rigid tube flow models, 180 measurement, 169–170
collapsible vessels, flow in, 185–189
entrance length, effect on arterial flow
S development, 182–185
hydrostatics in circulation, 168–169
Saphenous vein graft, 319 rigid tube flow models, 178–182
Sarcolemma, 164 cylindrical shape, 179–180
Sarcomeres, 164 fully developed flow, 180
Scallops, 269 laminar flow, 179
Scanner, 382 Newtonian fluid, 178–179
Secondary flow, 233 no slip at vascular walls, 179
Separated flow region, 243 rigid wall, 180
Separation point, 44 steady flow, 179
Shear-dependent mass transport, 225 vascular resistance, 180–182
Shear modulus, 53 Steady flow, rigid tube flow models, 179
Shear stresses, 4, 21 Stenosis, 29, 108, 170, 241, 278
prosthetic heart valve, 296–303 Stent implants, 333–338
Shearing, 5, 53 biomechanics of, 338–342
7328_Index Page 418 Tuesday, October 10, 2006 11:57 AM

418 Biofluid Mechanics: The Human Circulation

Straight tube fluid mechanics, 32–44 Turbulent jet theory, 295

flow development, 38–40 TWSSG. See Temporal wall shear stress
flow pulsatility, 41–44 gradient
flow stability, 33–41
steady laminar flow, 33–36
turbulent flow, 36–38 U
viscous, turbulent shear stress, 40–41
Streaklines, 356 Ultrasound Doppler velocimetry, 359–374
Streamlines, 26, 356 Unsteady flow, nonuniform geometric
Stress relaxation, 65 models, 191–256
Stroke volume, 80
Superposition principle, 54
Surface tension, as intrinsic fluid property, 8 V
Surfactants, 97
Valve annulus, 267
SVG. See Saphenous vein graft
Valve leaflets, 267
SWSSG. See Spatial wall shear stress gradient
Varicose veins, 90
Syncytium, 73
Vascular compliance, 86
Systemic circulation, 84–91
Vascular endothelial growth factor, 104
Systole, 77
Vascular graft implants, 315–317
Vascular grafts, 315
T clinical experience, 322–324
materials, 319–322
T wave, 77 Vascular resistance, 180–182
Temporal wall shear stress gradient, 227 Vascular stent, 334
Terminal impedance, 353 Vascular therapeutic techniques, 315–342
Thick-walled cylindrical tube analysis, 60–63 anastomotic intimal hyperplasia, 324–333
compatibility condition, 61–63 angioplasty, 333–338
equilibrium equation, 60–61 arteriovenous fistulas, 317–319
Thick-walled vessel, 60 endoluminal graft implants, 333–338
Thin-walled cylindrical tube analysis, 57–60 stent implants, 333–338
Thromboresistance, 313 biomechanics of, 338–342
Thrombosis, 322 vascular graft implants, 315–317
Thrombus, 105 vascular graft materials, 319–322
prosthetic heart valve, 308–309 vascular grafts, clinical experience with,
Tilting disc valve, 282, 299 322–324
Time rate of change of momentum, 13 Vascular walls, no slip, rigid tube flow
Time varying wall shear stress, models, 179
hemodynamic theory, 227–228 VEGF. See Vascular endothelial growth factor
Transit time flow meter, 350 Velocity coefficient, 174
Transitional flow, 36 Velocity profile, 369
Transmural pressure, 58 Vena contracta, 172
Tricuspid valve, 257, 267–273 Venous return, 90
mechanical properties, 269–270 Ventricularis, 260
valve dynamics, 270–273 View boxes, 359
True strain, 50–51 Viscoelasticity, 63–67
Tube, straight, fluid mechanics, 32–44 Viscometry, 116–123
flow development, 38–40 Viscosity, 4
flow pulsatility, 41–44 as intrinsic fluid property, 4–7
flow stability, 33–41 Viscous behavior of blood, 124–139
steady laminar flow, 33–36 hematocrit, effect of, 130
turbulent flow, 36–38 plasma, 124–126
viscous, turbulent shear stress, 40–41 protein content in plasma, effect of, 131
Turbulence intensity, 38 sigma effect, 137–139
Turbulent flow, 33, 36 temperature, effect of, 130–131
7328_Index Page 419 Tuesday, October 10, 2006 11:57 AM

Index 419

tube diameter, effect of, 135–139 X

whole blood, 126–130
yield stress for blood, 131–135 Xenograft, Hancock Porcine, 286
Voigt model, 65
Volumetric strain, 56
Y
W Yield stress, 6
Wall of blood vessel, structural components, Young’s modulus, 49
152–155
Wall shear stress gradients, 227
Waterfall effect, 186 Z
Windkessel models, human circulation,
192–195 ZCC. See Zero-cross counting
Windkessel theory, 192 ZCD. See Zero-crossing detection
Womersley number, 43 Zero-cross counting, 363
WSSG. See Wall shear stress gradients Zero-crossing detection, 363
7328_Index Page 420 Tuesday, October 10, 2006 11:57 AM
7328_Index Page 421 Tuesday, October 10, 2006 11:57 AM
7328_Index Page 422 Tuesday, October 10, 2006 11:57 AM