## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Introduction to Continuum Biomechanics

Copyright © 2008 by Morgan & Claypool All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means—electronic, mechanical, photocopy, recording, or any other except for brief quotations in printed reviews, without the prior permission of the publisher. Introduction to Continuum Biomechanics Kyriacos A. Athanasiou and Roman M. Natoli www.morganclaypool.com ISBN: 9781598296174 paperback ISBN: 9781598296181 ebook DOI: 10.2200/S00121ED1V01Y200805BME019 A Publication in the Morgan & Claypool Publishers series SYNTHESIS LECTURES ON BIOMEDICAL ENGINEERING #19 Lecture #19 Series Editor: John D. Enderle, University of Connecticut Series ISSN ISSN 1930-0328 ISSN 1930-0336 print electronic

**Introduction to Continuum Biomechanics
**

Kyriacos A. Athanasiou and Roman M. Natoli

Rice University

SYNTHESIS LECTURES ON BIOMEDICAL ENGINEERING #19

elasticity. thermoelasticity.iv ABSTRACT This book is concerned with the study of continuum mechanics applied to biological systems. linear viscoelasticity. Casson equation. continuum biomechanics.e.. but individuals working at biotechnology companies that deal in biomaterials or biomechanics should also find the information presented relevant and easily accessible. This vast and exciting subject allows description of when a bone may fracture due to excessive loading. We have written for senior undergraduate students and first year graduate students in mechanical or biomedical engineering. poroelasticity. a process known as mechanotransduction. Newtonian fluids. biomechanics. blood flow. KeywoRdS Continuum mechanics. biphasic theory . down to how cells respond to mechanical forces that lead to changes in their behavior. i. how blood behaves as both a solid and fluid.

To all of my graduate students for sharing the passion for the unrelenting pursuit of excellence. and Thasos for being the center and providing the apogee of my life. and my earthly father. for the gift of life. -RMN . -KA2 To God the father. for instilling in me a love of science and mathematics. the late Joseph D. Natoli.v dedication To Kiley. Aristos. these two are together now. Hopefully.

.

and differential equations is helpful. a basic understanding of elementary mechanics. However. Chapters 5 and 6 deal exclusively with fluids. and is easily covered in a standard. Furthermore. three credit hour course. how blood behaves as both a solid and fluid. as we present everything needed to follow this book.e. We then address linear elasticity. We assume no knowledge of tensor algebra or calculus. In this book. spring-dashpot) descriptions are presented. semester-long. we briefly look at poroelasticity and thermoelasticity. We first cover general isotropic elasticity. which are examples of coupled theories. the correspondence principle and methods for analyzing dynamic loading of viscoelastic materials are discussed. a process known as mechanotransduction. While non-Newtonian fluids are briefly discussed. vector algebra and calculus. using bone as an example. we begin with elasticity. as it is the archetypical constitutive theory. Here. We depart from linearity in Chapter 6. we first develop the necessary mathematical. Both integral and differential (i. kinematic. we return to linear theory. continuum biomechanics. devoting some time to hyperelastic descriptions of finite deformation. discussing linear viscoelasticity. We have written for senior undergraduate students and first year graduate students in mechanical or biomedical engineering. In Chapter 7. detailing the experimental observations supporting this model and solving some of the classical problems. This vast and exciting subject allows description of when a bone may fracture due to excessive loading.. Chapter 9 deals with ..vii Foreword This book is concerned with the study of continuum mechanics applied to biological systems. The material is presented in the same order that is usually taught. multivariable calculus. but individuals working at biotechnology companies that deal in biomaterials or biomechanics should also find the information presented relevant and easily accessible. In Chapter 4. as applied to tendons or ligaments. where the Casson equation describing blood flow is studied in its different regimes. our biological examples are tendons/ligaments and a single cell. and the equations of motion are solved for flow within a large cylindrical vessel. i. which are described by a linear constitutive equation. This book has evolved from notes that have been developed for a first-year graduate course in biomechanics at Rice University. and stress analysis concepts essential to continuum theory (Chapters 1–3). Finally. down to how cells respond to mechanical forces that lead to changes in their behavior. In Chapter 8. the main focus of Chapter 5 is Newtonian fluids.e.

a solutions manual is available upon request. We devote the majority of the chapter to biphasic theory. Within each chapter. We hope you find the material comprehensive enough. For the more interested reader. In addition. . we are excited by the subject of continuum biomechanics and hope that you share our enthusiasm by the time you finish this book! A note to instructors: sections and chapters preceded by an asterisk in the table of contents can be omitted from a one semester course without loss of continuity. as applied to articular cartilage and fibrocartilage biomechanics. solving both creep and stress relaxation problems in confined compression and deriving the governing equation for unconfined compression. but not overly bogged down with mathematics.viii INTRoduCTIoN To CoNTINuuM BIoMeChANICS the subject of mixture theory. Of course. each chapter is followed by several problems that reinforce or expand upon the material presented. of important concepts. which we call demonstrations. we give examples. references to classical papers and more advanced textbooks on both mechanical and biological aspects of the subject matter are provided. Also.

...................................................5 Problems ...............................2.................................... 21 1...............................3 Tensor-Valued Functions of a Scalar .. 9 1.........1..1.............................. and Principal Directions . Spatial Description ............4....2.............................4............................... 29 2................................................................ Vector.........................................................................................................................................................................1........3 Tensor Symmetry.............................................. 22 1....1 Scalar Invariants of a Tensor and the Cayley-Hamilton Theorem...................................................................5 Transpose of a Tensor ......... and Tensor Fields ......... 21 1................... 12 1.............................. ..4 Other Useful Tensor Relationships ..... 30 2........2............................................. 15 1...... 29 2.2 Eigenvalues and Eigenvectors......... 19 1...............................6 Orthogonal Tensor .................................................................................. Tensor Calculus .....................3............................... 9 1................................ 9 1.............. 8 1..............2 Material vs........ Principle Values.........................................1......................4....................................2 Multiple Equations (Free Indices) ............................................................................1 Description of the Motion of a Continuum .............................................................................1 Tensor Definition ...1 Symmetric vs.......... Antisymmetric Tensors ................................. 15 1.....3 Principal Values and Principal Directions ..........................................................................................................2.........2 Tensors .................3...4 Identity Tensor ..........1 1........................................................................ 20 1.........................................2.. 6 1............................................3 Kronecker Delta ............................... 32 2.........2 Trace of a Tensor ............ix Contents Introduction ....... 7 1...........4 Gradient and Divergence of Scalar.............. 20 1............ 11 1...............3 Material Derivative ............ 16 1.................................. 23 Kinematics of a Continuum ..........................2 Components of a Tensor............1 Summation (Dummy Indices) ..... 5 1.........4....................................................... 12 1.................................................... 5 1..4 Manipulations.........................................................3 Sum and Product of Tensors .....................1 Indicial Notation ...... 12 1..................3...........5 1......2........

.........10 Planar Approximations (2D Simplification) ...............................3 Principle of Moment of Momentum (Proof of Stress Tensor Symmetry) .....................5 Maximum Shear Stress ....................... 77 4...... 80 4.......................................................................................................................1 Summary Up to Now ................................9 Classical Problems in Elasticity .....3 Experimental Observations of Infinitesimal Linear Elasticity............................................ 60 3. 3............................................................................................6 Equations of Motion (Conservation of Linear Momentum).................................1 Stress Vector (“Traction”) ....................................... 58 3..7 Boundary Condition for the Stress Tensor ................7 2......................................6 Material Properties of Elastic Materials ................................................. 79 4..............9..................... 67 * 4..........................................................x INTRoduCTIoN To CoNTINuuM BIoMeChANICS 2............................................ 63 elasticity.... 34 Principal Strains ..........................................4 2....................................................... 51 3.. 87 4............... 68 4......................... 75 4............... 52 3.........................8 Compatibility Conditions for Infinitesimal Strain Conditions....... 41 Problems ....................... 73 4......5 Isotropic Linearly Elastic Solid ...........6 2.........................................................................................................................1 Simple Infinitesimal Extension of a Linear Elastic Solid .. 76 4..........................2 Pure Bending of a Beam ............ 57 3..................................................................................10 Problems .........7 Equations of the Infinitesimal Theory of Elasticity ..4 Principal Stresses ......................9............ 67 4.............................................. 39 Rate of Deformation ....................................... 69 4....................................................... 92 4..........................................2 General Elasticity ...........................2 Approximations Leading to Linear Elasticity .... 40 Continuity Equation (Conservation Of Mass) .......... 81 4.... Deformation-Induced Strain ..2 Stress Tensor and its Components................................................. 81 4......................................... 54 3........................ 45 Stress ..............................9 Demonstrations ..................................2...................................................... 52 3.........................................................................9...................................8 2....................................................................................5 2................... 84 4.................................................................3 Torsion of a Circular Cylinder ....... 72 4...............................1 Hyperelasticity ....................8 Alternative Stress Definitions ............................................................9............................................ ........... 53 3..............................................................................1 Plane Strain .................................................4 Linearly Elastic Solid ... 51 3.....................2...............................................................10................................... 39 Dilatation ................................................ 58 * 3............................................... 92 4.........................

............................................ 115 * 5............................................................................................................ 97 5........................................................ 107 5......................5 Casson Equation .....................6 Blood Rheology ..........9...13 Problems ...........................7 Boundary Condition .....1 Introduction................................................................................................................................................................................................................. 108 5......... 128 6......... 117 5.11 Vorticity Vector .........................................................................................12.................................................4 Non-Newtonian Behavior of Blood .................................9...............1 Plane Couette Flow ...................................................................... 139 7....................................................................1 Introduction to Fluids .......... 134 Viscoelasticity ................................................... 102 5.............................................. 7..................................... 109 5.......................................................2 Basics and Material Properties of Blood .....................................9 Classical Flows ........8 Laminar Flow of Blood in a Tube .......................... 112 * 5....2 Plane Poiseuille Flow........................................................... 106 5.................. 113 * 5................................................4 Meaning of l and m ...............9 Problems ................... 94 * 4...............5 Incompressible Newtonian Fluid................................................7 Summary ............12 Irrotational Flow............2 Hydrostatics...................................................... 105 5................. 129 6............................................. 120 Blood and Circulation ................................ 123 6.............................................................................................................................. Fluids ......8 Important Definitions ......................................... 107 5........................1 Irrotational Flow of an Inviscid Incompressible Fluid . 126 6......... 124 6...............6 Navier–Stokes Equations.....1 Introduction........3 Reynolds Numbers for Blood .................................................................. 139 6...... 123 6...............................................................................................................................................CoNTeNTS xi 4........10 Non-Newtonian Fluids .................................................. 101 5...................... 126 6.. 116 5............................11 Anisotropic Linear Elasticity ............................................................ ......................................................................................... 94 4......... 108 5................................................................ 124 6................................................................................................9.............................................................................................................12 Problems ...........................3 Newtonian Viscous Fluid ................ 106 5................................................... 101 5.........................................................................................2 Plane Stress ..................................................... 123 6...................... 104 5..................................3 Extensions of Plane Poiseuille Flow .............................10............................

....................... .................................2.......1 Introduction................................................................................7.........................1 Maxwell Fluid ...................................................................................6 7....................................... 179 9.................... 163 8............... 165 8................................... 144 7... 183 9......................... 172 8.................................................................2 Constitutive and Governing equations (“u–q ” Formulation) ...................... 179 9............................................3............................... 172 8..................... 163 8........... 167 8................ 158 Poroelasticity and Thermoelasticity........................................4 Conservation of Momentum ............... 150 Boundary Value Problems and the Correspondence Principle.......................................................................................... 139 1-D Linear Viscoelasticity (Differential Form Based on Mechanical Circuit Models) ................................3 Thermal Prestress/Prestrain.......................3..2 Poroelasticity ........................................... 156 Problems ............ 180 9..............4 u–p Formulation of Poroelastic Governing Equations .......5 Consolidation of a Finite Layer (Terzaghi’s Problem) ..........................3 Thermoelasticity ............. 153 7............................................. 141 7. 145 7....................2 Definitions ....................................................................................................1 Introduction and Fourier’s Law ...................................................................................................................................... 163 8................8 7................................2.............................1 Terzaghi’s Principle of Effective Stress ...............3 Conservation of Mass ........................................................................................ 176 Biphasic Theory ............... 164 8........................3... 185 9........3 Constitutive Equations and Material Constants ...............2 Kelvin–Voigt Solid ............................ 146 1-D Linear Viscoelasticity (Integral Formulation) ..................................... 163 8..........................................................................................2...3 7..............................................................................................................................4 7...... 151 Dynamic Behavior of Viscoelastic Materials ............................2... 156 Limiting Cases of Linear Viscoelasticity are the Hookean Solid and Newtonian Viscous Fluid .......................................................................xii INTRoduCTIoN To CoNTINuuM BIoMeChANICS 7..................................2.............2 7..... 173 8.....................................................4 Beyond the Canonical Models .3......7 7............5 Constitutive Equations .....................3............................... 140 7.......... 175 8............................... 148 3-D Linear Viscoelasticity ............1 Introduction........................ 169 8.... 181 9........5 *7.....................................................3....................9 * 8.............................4 Problems ...3 Standard Linear Solid.......1 Dynamic Maxwell Fluid .......................................... Definition of Viscoelasticity .......................................................3......2 Darcy’s Law ......................

................................................................................................................................................................................ 199 Afterword ............................................... 188 9.9 Summary and Equations of Motion ...........................................................7 9......8 9....................................................................................................................... 205 .............................................................................6 9................7........................................................... 186 Confined Compression . 189 9.............. 196 Acknowledgments ....................................................................7.......... 194 Problems ......................................2 Stress Relaxation......... 203 Author Biography ...................................................................CoNTeNTS xiii 9........................................ 192 Unconfined Compression .... 201 Bibliography ............................................................1 Creep .....................................

.

Constitutive relationships are equations defining idealized materials. at the macroscopic level when subject to different loading conditions. matter is not continuous. However. . constitutive equations vary from material to material and. bone deformation. which consist of atoms and subatomic particles. we could describe the macroscopic behavior of materials based on detailed knowledge of the microscopic behavior using statistical mechanics. etc.g. both solids and fluids (including liquids and gases).e. principle of entropy. In this text.1 Introduction CoNTINuuM TheoRy We know that matter is composed of discrete units called molecules.. Continuum mechanics studies the response of materials. General principles are common to all materials (e.. Continuum theory describes relationships between phenomena by paying no attention to the molecular structure of materials.g. Some examples are an elastic body (deformation 0 as load 0). One is an integral formulation over a finite volume. though even the simplest of problems can be mathematically intractable. and linearly viscous fluid (stress linearly related to strain rate). In general. principle of linear momentum and angular momentum. for a given material. governing equations) and constitutive relationships. and fortunately. The second is field equations for a particle or differential material volume at every point.) These governing equations can take two different. we consider only the latter. may vary depending on loading conditions. Thus. While rigid body motion is included in this theory as a special reduction. linearly elastic solid (stress linearly related to strain). It regards matter as indefinitely divisible.. In principle. yet equivalent. an assumption that is justified only so far as it can adequately capture experimental observations. blood flow). The subject of continuum mechanics can be subdivided into two primary parts: general principles (i. deformation of materials under applied loads is its most common application. many aspects of daily life can be described or predicted with theories that pay no attention to the molecular structure of materials (e. CoNTINuuM MeChANICS Mechanics is the branch of physics relating applied forces to motion. forms. conservation of mass and energy.

This book concentrates on the mechanics of empirically observed relationships between stress and strain or strain rate. in combination with the constitutive equations. The elegance of continuum mechanics is its mathematical framework that yields concise treatment of the governing equations and constitutive relationships. It is important to keep in mind that your study of continuum biomechanics is rich in the sense that the majority of equations and problem solving techniques you will learn are applicable to a large range of engineering materials and not just to human locomotion or to tissues and cells within the body. We will investigate aspects such as these. as well as design and analysis. but acts as a linear biphasic material under quasi-static loading conditions common to confined. under circumstances of interest to scientists and engineers. Biomechanics is suitable for research and development. In this text. We do not concentrate on the anatomy. Continuum biomechanics is part of a larger field of mechanobiology. In continuum mechanics two aspects of investigation can be considered. determinism. The second is solving the governing equations. with examples taken from biological systems. of problems having biological or medical interest. which is the study of the response of biological systems to mechanical forces. As a framework for mechanobiology. and failure theories. physiology. The first response is the mechanical response of the system. Explicit rules for constitutive equation formulation (e. One is the formulation of constitutive equations appropriate for describing behaviors of particular classes of materials under certain loading conditions. we present continuum mechanics applied to biological systems. articular cartilage exhibits behavior reminiscent of an incompressible hyperelastic solid under quick-loading conditions.2 INTRoduCTIoN To CoNTINuuM BIoMeChANICS For example.)[1] and advanced methods for solving particular problems are left for an intermediate text focused on more specialized applications. we seek to develop the necessary tools to work on both aspects at an introductory level. one can think of two distinct. which is determined by governing equations. responses.g. over the course of this text. leading to protein production. or structure–function relation- . constitutive relationships. there is an inherent treatment of biology in addition to the description of mechanics principles. and creep indentation testing. yet linked. In this text. coordinate invariance. and ultimately affecting a tissue’s structure–function relationship. to name a few. CoNTINuuM BIoMeChANICS Biomechanics is the application of engineering mechanics to biological systems. This response is determined by the history of mechanotransduction. etc. and many other constitutive equations. The second response is a biological one. As the term biomechanics implies. unconfined. the process by which mechanical signals initiate gene transcription. or continuum biomechanics. such as may occur during impact.. and material objectivity.

With this said. as we expect this knowledge to have been obtained through other means. No where may this be truer than in the role of a biomechanician.INTRoduCTIoN 3 ships of the tissues of interest. It has been said that a bioengineer is one who takes a biological problem and reduces it. to a problem in one of the engineering disciplines. let us get started! . under appropriate assumptions.

.

(1.1 INdICIAL NoTATIoN Continuum mechanics is a mathematically elegant theory. 2...1. + an xn or (1. k in Eqs. 3.3) are called dummy indices. For example.2) and (1. 1. By convention. j. we must be able to concisely express the equations governing and describing the behavior of materials. if an index is repeated once within a simple term.1 Summation (dummy Indices) Consider the sum S = a1 x1 + a2 x2 + a3 x3 + . .3) The indices i. We will see shortly that indicial notation simplifies the bookkeeping process of the many mathematical expressions needed in continuum mechanics by allowing us to write them in a compact form.1) S= i =1 ∑ ai xi n (1. it is a dummy index indicating a summation. each dummy index has values 1. As such.2) or S= j=1 ∑ aj x j n = k=1 ∑ ak xk n (1. Thus.5 CHAPTER 1 Tensor Calculus 1.

we can reduce Eqs. expressions such as S = ai bi ci (where i is repeated more than two times) are meaningless. 2. 3. by introducing the concept of a free index. 3.6) Using dummy indices. it is understood that the summation occurs over repeated indices without having to write the summation symbol.7) are of the same form.4) become and S = ai x i (1. 2. Eqs.7) Notice Eqs.1.2 Multiple equations (Free Indices) Consider y1 = a11 x1 + a12 x2 + a13 x3 y2 = a21 x1 + a22 x2 + a23 x3 y3 = a31 x1 + a32 x2 + a33 x3 (1. With this convention. .6) to y1 = a1n xn y2 = a2n xn y3 = a3n xn (1.7) can be written as yi = ain xn (1. Eqs.6 INTRoduCTIoN To CoNTINuuM BIoMeChANICS S= and i =1 3 ∑ ai xi ∑ ∑ aij xi xj 3 3 (Sum of 3 terms) (1. indicating multiple equations. Free indices appear only once in each term of an equation. They typically take on the values 1. 1.1).8) where i is the free index and n is a dummy index (as described in Section 1. (1.5) S = aij xi xj According to this convention. differing only by the index 1. (1. Thus. we will employ Einstein’s summation convention. Thus.1. (1.4) S= (Sum of 32 terms) i =1 j =1 To alleviate writing the summation symbol continuously. (1.

then the expression represents nine equations.11) i..14) or (1.16) 3.TeNSoR CALCuLuS 7 For example. d11 = d22 = d33 = 1 and d12 = d13 = d21 = … = 0. In general.e. Tij = Aim Ajm (1. δim Tmj = Tij Remember. T12 = A11 A21 + A12 A22 + A13 A23 There are eight other equations for the other combinations of i and j. ⎤ ⎡ ⎤ δ11 δ12 δ13 1 0 0 ⎢ ⎥ ⎢ ⎥ ⎣ δ21 δ22 δ23 ⎦ = ⎣ 0 1 0 ⎦ δ31 δ32 δ33 0 0 1 (1. .15) (1.13) 2.16) represents nine equations.10) 1.3 Kronecker delta The Kronecker delta is defined as δij = � 1 i=j 0 i �= j (1. Now let us consider some examples: 1. (1. (1.9) represents nine equations. For example. The matrix ⎡ is the identity matrix.1. (1. each with three terms on the right-hand side. each with three terms on the right-hand side. Taking the free indices i = 1. yi = aimxm represents three equations. The free index appearing in every term or equation must be the same. j = 2.12) δii = δ11 + δ22 + δ33 = 3 δ1m am = δ11 a1 + δ12 a2 + δ13 a3 = a1 δ2m am = a2 δ3m am = a3 δim am = ai (1. If there are two free indices in an equation. Eq.

1. V. 4. accomplished by changing m to n.22) (1. Substitution. (1. as the dummy index is repeated more than twice. should we quit there. ® are unit vectors perpendicular to each other.20) into Eq. Eq. the new expression. ®. ai = Uimbm and bi = Vimcm (1. 1. (1. strain).20) to m. (1. stress. If p = ambm and q = cmdm. (1. Multiplication. dij .4 Manipulations 1. If ®. The Kronecker delta.19) write Eq. and c.21) Expression (1.22) represents three equations (one free index) with nine terms on each right-hand side (two dummy indices).19) in terms of U.g.20) becomes bm = Vmncn which. ai are components of a vector (e. Thus.17) � 1 1 2 2 3 3 1 2 1 e i · e j = δij 3 2 1 � (1. Tij are components of a tensor (e. 2.19). we need a new dummy index. (1. (1. (1.15).16)..20) (1. and in Eq. displacement.8 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Note: In Eq. (1. 1 2 3 (1. ® × ® = ® × ® = ® × ® = 1 and ® × ® = ® × ® = ® × ® = … = 0.19). traction). then changing the dummy index m to n in the expression for q results in pq = ambmcndn (1. Hence. bm = Vmmcm.21) into Eq. we need to change i in Eq.23) . To substitute Eq.. However. yields ai = UimVmncn (1. upon substituting Eq. (1. can be used to contract indices as follows: δim δmj = δij δim δmj δjn = δin e e e 5. Given two expressions.g.18) e e e e e e e e e e e e or. would be meaningless.

Furthermore. ® ® ® ® ® ® ® ® ® a × b = ai bi ® ® (1. e be unit vectors in a rectangular Cartesian coordinate system. the dot product is commutative and distributive such that ® × b = a ® ® ® b × ® and ® × (b + ®) = ® × b + ® × ®.2. The Cartesian components ® ® ® ® ® of a vector a are given by a1 = a × ®. So.15). Factoring. It transforms any vector into another vector. Thus.28) T has the following properties ˜ where a is a scalar. a · b = ai bj δij = ai bi . denoted by T . e .24) ® The dot product of ® and b is also given by ® × b = |a ||b |cos q. where q is the interior angle between a a ® the two vectors.2 TeNSoRS 1. from Eq. a2 = a × ® .2. Given two vectors a = ai ei and b = bi ei .29) 1. (1. (1. ni = dij nj . and a3 = a × e 3. using Eq.18).25) Tij nj − λ ni = 0 then.TeNSoR CALCuLuS 9 3. a a c a a c 4. � � � � � � T a + b = Ta + Tb ˜ � ˜ ˜ � � � T α a = αT a ˜ ˜ (1. we can also define the magnitude of a vector as � �� � � � √ � a � = a · a = ai ai ® ® ® (1.27) 1.2 1 2 Components of a Tensor 3 1 2 Let e . Eq. respectively.26) becomes (1.1 Tensor definition A tensor is a linear transformation. � � Then. or e e ® ® ® . a × b = (ai ei ) × (bj ej ) = ai bj ( ei × ej ). If Using the dot product. A ˜ second-order tensor represents nine scalar quantities Ta=b ˜ � � (1.26) � � Tij nj − λ δij nj = Tij − λ δij nj = 0 (1. Vector dot product. (1.

30) a = a1 e 1 + a2 e 2 + a3 e 3 = ai e i ® � � � � (1..36) . Note that T ®i are not perpendicular unit vectors. What is the vector T ®j ? e ˜ Solution. b = T a in matrix format is ˜ ® (1.31) ® ® Now consider a tensor T .30) e e e e ˜ ˜ ˜ ® ˜ and (1. For any vector a .33) and (1.33) ⎡ from which we identify ⎤ ⎡ ⎤⎡ ⎤ b1 a1 T11 T12 T13 ⎢ ⎥ ⎢ ⎥⎢ ⎥ b2 ⎦ = ⎣ T21 T22 T23 ⎦ ⎣ a2 ⎦ ⎣ b3 T31 T32 T33 a3 bi = Tij aj (1. b = a1T ®1 + a2T ®2 + a3T ®3.35) T e j = Tkj e k ˜ � (1. the components of b are � � � (1.e. (1.32) b1 = e 1 · b = e 1 · (ai T e i ) = ai e 1 · T e i = a1 e 1 · T e 1 + a2 e 1 · T e 2 + a3 e 2 · T e 3 b 2 = ai e 2 · T e i b 3 = ai e 3 · T e i ˜ � � � � ˜ � � � ˜ � � ˜ � � ˜ � � ˜ � � ˜ � or e e bi = aj®i × T ®j ˜ ® Now. � � � (1.34). we find that the components of T are ˜ Tij = e i · T e j ˜ Demonstration. Using Eqs.10 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ai = a · e i Equivalently. � � � (1. b = T a is a vector given by ˜ ˜ � ® b = T a = T (ai e i ) = ai T e i ˜ ˜ ˜ i.32).34) Comparing Eqs. (1.

T32 ⎡ ⎤ 0 � � ⎢ ⎥ � � = e 3 · T e 2 = ⎣ 0 ⎦ · T12 T22 T32 = (0)(T12 ) + (0)(T22 ) + (1)(T32 ) = T32 ˜ 1 (1. 1. e e ˜ Now. the i in ®i in Eq.24). let us consider T e 2 as an example. Then.35) determines the row. (1. refer to Eq. Eq. By definition.35) with i = 3 and j = 2.3 Sum and Product of Tensors Let T and S be two tensors. (1.37) and (1.37) Thus.41) or.2. (1. Note: The expression T ®j = Tkj ®k will become useful when trying to determine the compoe e ˜ nents of a tensor when the resulting transformation of original unit vectors is known.40) (T + S ) a = T a + S a ˜ ˜ ˜ ˜ Examining the components of the sum. From Eq. in matrix notation. (1.37) and the definition of the dot product. ˜ ˜ (1. [T + S ] = [T ] + [S ] ˜ ˜ ˜ ˜ (1. ⎡ ⎤ T12 ⎢ ⎥ � � � � (1. the same way that i denotes the row in the exe pression Tij . ˜ ⎤⎡ ⎤ ⎡ ⎤ ⎡ 0 T11 T12 T13 T12 ⎥⎢ ⎥ ⎢ ⎥ ⎢ � T e 2 = ⎣ T21 T22 T23 ⎦ ⎣ 1 ⎦ = ⎣ T22 ⎦ ˜ T31 T32 T33 T32 0 ® (1. � � � (T + S )ij = Tij + Sij ˜ ˜ (1. the same way that j denotes the column in the expression e ˜ Tij .39) Thus. the j in T ®j determines the column.TeNSoR CALCuLuS 11 To understand this.38).42) .38) ⎣ T22 ⎦ = T12 e1 + T22 e2 + T32 e3 = Tk2 ek T32 Comparing Eqs. (1. we see T ®2 = Tk2 ®k .

˜ .44) (1.12 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where [T ] is the matrix representation of the tensor T .49) Note: The transpose of the product of two tensors is the product of the two tensors transposed in reverse order.48) must also hold. I .6 orthogonal Tensor An orthogonal tensor.45) (T S )ij = Tim Smj ˜ ˜ [T S ] = [T ][S ] Note: In general.4 Identity Tensor The identity tensor.47) 1.50) 1.46) from which it is evident that I can be represented as ˜ I = dij ˜ (1.2. Q . is the linear transformation that transforms any vector into itself. ˜ ˜ For the product of two tensors. (T S)T = S T T T ˜ ˜ ˜ ˜ (1.2. is a linear transformation for which transformed vectors preserve their ˜ lengths and angles. Thus. T S ¹ S T .2. (T S ) a = T (S a) ˜ ˜ ˜ ˜ � � (1.5 Transpose of a Tensor The transpose of a tensor T is defined as the tensor T T that satisfies ˜ ˜ a · (T b) = b · (T T a) � ˜ � � ˜ � (1.48) � � � � Considering the unit vectors. for which Eq. e i · (T e j) = e j · (T T e i ) which ˜ ˜ shows that Tij = (T T )ji (1. given an orthogonal tensor Q .43) (1. (1. ˜˜ ˜˜ ˜ ˜ ˜ ˜ 1. ˜ ® ® I a= a ˜ (1. Thus.

Q b ˜ ˜ (1. for arbitrary (nonzero) vectors a and b . (1.54).TeNSoR CALCuLuS 13 | Q a | = |a | ˜ � � (1. (1. a ˜ ® and.53) Substituting Eq.52) Note: In general. T ¹ T T T ¹ I. From ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ � �� � �� � � Qa · Qb = a · b ˜ ˜ (1.56) (1. (1.57) (1.58) . Q® ¹ a . � � �� � � �� cos a. where � +1 indicates a rotation det(Q) = −1 indicates a reflection ˜ (1. Only their lengths are equal. Furthermore. Why? Let us take a look. for an orthogonal tensor Q .53) into Eq.51) Note: In general. � � �� � �� � � T � � Q a · Q b = b · Q (Q a) ˜ ˜ ˜ ˜ (1. so it follows.48). but Q Q T = Q T Q = I . ® � � � � � a · b = b · Q T (Q a) ˜ � ˜� � � � � b · a − b · QT Q a = 0 ˜ ˜ �� � � b · I −QT Q a = 0 ˜ ˜ ˜ ® I = QT Q = QQT ˜ ˜ ˜ ˜ ˜ Hence. it is also true that ˜ Q = Q -1 ˜ ˜ In indicial notation Qim Qjm = Qmi Qmj = dij Note: The determinant of an orthogonal tensor is ±1.54) Therefore. T T ˜ ˜ Eq.55) (1. b = cos Q a.

e . Examine whether R is orthogonal.59) � � R e2 = − e1 Recall Eq. Find the matrix representation of the tensor R describing this rotation. e 1. Find the matrix representation of the tensor S describing this ˜ rotation. Find the final position of a point. From Eqs. (1.36). R e2. ˜ 4.60) ⎤ 0 −1 0 ⎥ ⎢ [R ] = ⎣ 1 0 0 ⎦ ˜ 0 0 1 ⎡ FIguRe 1. ⎤ ⎡ R11 R12 R13 ® ® ® ⎥ ⎢ R = ⎣ R21 R22 R23 ⎦ with R e1. R is the transformation ˜ � � R e1 = e2 ˜ (1. 0) after these two rotations. 5.1: Standard coordinate system rotated 90° about the ®3 axis. Suppose this rigid body experiences a 90° right-hand rotation about the original ®1 axis e by the right-hand rule.14 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Demonstration. Find the determinant of R . (1. p. � � � � � � � � � � � � � � � (1.36) and (1.1). ˜ 2. Solution. 1. originally at (1.59) ˜ ˜ ˜ ˜ R31 R32 R33 R e3 = e3 ˜ ˜ � � R e1 = R11 e1 + R21 e2 + R31 e3 = e2 ˜ ˜ ˜ R e2 = R12 e1 + R22 e2 + R32 e3 = − e1 R e3 = R13 e1 + R23 e2 + R33 e3 = e3 So. A rigid body is rotated 90° by the right-hand rule about the ®3 axis (see Figure 1. ˜ 3. and R e3 as columns. 1.

0)T.3 TeNSoR SyMMeTRy. R is orthogonal. This is different from orthogonal ˜ ˜ ˜ tensors. Hence. T symm. � � S describes only the second rotation. R S r yields p’ = (-1. � ˜ � � 0 −1 0 � � � � � 3.g. p = (-1.TeNSoR CALCuLuS 15 2. of point p after both rotations is ⎤⎡ ⎤⎡ ⎤ ⎡ ⎤ −1 0 −1 0 1 1 0 0 ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ �� r = S R r = ⎣ 0 0 −1 ⎦ ⎣ 1 0 0 ⎦ ⎣ 1 ⎦ = ⎣ 0 ⎦ ∼∼ 0 1 0 0 0 1 1 0 ⎡ So. Following Eq. The vector describing the location of point p is r = (1. Therefore. R is a rotation. PRINCIPAL VALueS. (1. ˜ ˜ ˜˜ ˜ ˜ 1. 1. ® Note: The order of rotations matters. The components of a symmetric tensor satisfy . 0)..60). RRT = ⎣ 1 0 0 ⎦ ⎣ −1 0 0 ⎦ = ⎣ 0 1 0 ⎦ = ∼ I ∼ ∼∼ 0 0 1 0 0 1 0 0 1 0 0 1 Therefore. ANd PRINCIPAL dIReCTIoNS 1. the new location.3. A symmetric tensor does not necessarily preserve lengths and angles. not the combined total transformation. R is orthogonal iff (“if and only if ”) R RT = I . 1). p . ∼ � � ˜ �0 0 1� S e1 = e1 ∼ S e2 = e3 ∼ � � � 4. ˜ ⎤ ˜⎡ ⎤˜ ⎡ ⎤⎡ ⎡ ⎤ 0 1 0 0 1 0 0 −1 0 1 0 0 � � ⎥ ⎥ ⎢ ⎢ ⎥⎢ ⎢ ⎥ RT = ⎣ −1 0 0 ⎦. 1. T S ¹ S T .55). So. e. 0. ˜ S e3 = − e2 ∼ � é1 0 0 ù [S ] = ê0 0 -1ú ú ê ê0 1 0 ú ë û ® 5.1 Symmetric vs. det(R ) = � 1 0 0 � = −(−1)(1) = 1 . is defined as symmetric iff T symm = (T symm)T. Antisymmetric Tensor A tensor. Recall Eq. This exemplifies the fact that ˜˜ given two tensors T and S . (1.

T asymm.66) (T .63) T ∼ T ∼ symm = = T + TT ∼ ∼ 2 T − TT ∼ ∼ 2 (1. ˜ A tensor. T31 = . a is defined as an eigenvector of T if it transforms under T into a ® ˜ ˜ ˜ a vector parallel to itself.61) This means that T12 = T21. From Eqs. 1.62) This means that T12 = .46). Any tensor T can always be decomposed into the sum of a symmetric and antisymmetric ˜ tensor. Antisymmetric tensors are also known as asymmetric or skew tensors.65) and (1. as I a = a and l = 1. ® ® ® T n = ln = l I n ˜ ˜ ® ® with n · n = 1. This means ® ® Ta = l a ˜ (1.16 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Tij = Tji (1. Also. T = T symm + T asymm ˜ ˜ ˜ where (1. The components of antisymmetric ˜ ˜ ˜ tensors satisfy Tij = -Tji (1. ˜ ˜ ® Now let n be a unit eigenvector. and T23 = .3.T13. T31 = T13. the diagonal elements of T asymm must ˜ be zero.T21. Also.(T asymm)T .64) asymm Another relevant decomposition of tensors is into spherical and deviatoric tensors (see Problem 14). ® ® Note: Any vector is an eigenvector of I .T32 .l I )n = 0 ® ˜ ˜ (1.2 eigenvalues and eigenvectors Consider a tensor T and a vector ®. is antisymmetric iff T asymm = . (1. the diagonal elements of T symm and ˜ (T symm)T must be equal. For definiteness. This implies (1. and T23 = T32.65) where l is called the eigenvalue.67) . all eigenvectors will be of unit length.

For the given T .69) (1.69) is of the form A® = 0. we must have det ( A ) = 0. l’s. in long form. Thus. (1.68) and (T11 − λ )a1 + T12 a2 + T13 a3 = 0 T21 a1 + (T22 − λ ) a2 + T23 a3 = 0 T31 a1 + T32 a2 + (T33 − λ ) a3 = 0 a12 + a22 + a32 = 1 (1. ˜ |T . show that ®1 is an eigenvector of T with T11 as the corresponding e ˜ ˜ eigenvalue. in order to find nonzero ˜ ˜ eigenvectors.36).72) . in component form (let n = ai ®i). Eq. (1. and corresponding eigenvectors. A -1 does ˜ ˜ ˜ x not exist and det ( A ) = 0.68) or (1. Then. recall Eq. n ’s. and a3. (1.e. 0. Also.TeNSoR CALCuLuS ® or. Solution.65). (1. To find the eigenvalues. (1. ∼ 0 4 −3 Solution.71) or. a2. (1. (1.27) e 17 (T ij . In long form. ˜ Demonstrations.l dij)aj = 0 with aj aj = 1.65) that says T ® = l ®. 1. Eq. Recall Eq. find the eigenvalues.70) are used to solve for eigenvectors of a tensor. Given T with T21 = T31 = 0. (1. Eqs. (1. Given T = ⎣ 0 3 4 ⎦. (1. It is known as the characteristic equation of T . (1. Eq. which satisfies Eq.l ˜ | = 0 I ˜ (1.72) is a cubic equation in l. using Eq.69).69) are a system of linear homogeneous equations in a1. Recall from linear algebra that the homogeneous ® ® equation Ax = 0 has only the trivial solution x = (0.70) Eqs. ∼ ∼ ⎡ ⎤ 2 0 0 � � ⎢ ⎥ ® 2. a a � � � � � ˜� T e1 = T11 e 1 + T21 e 2 + T31 e3 ⇒ T e1 = T11 e 1 . Eq.68). and (1. 0) unless A is a singular matrix. (1. i. and..72).72) is ˜ � � �T − λ T T13 � 12 � 11 � � � � T21 T22 − λ T23 � = 0 � � � T31 T32 T33 − λ � (1. we will make use of the characteristic equation.

Eq. Using Eq.75) become These are two equations with two unknowns. when solved. Eqs. (1. n3 = ± √ (− e 2 + 2 e 3 ). eigenvectors are of unit length. corresponding to l2 = 5 is �2 = ± √ (2 �2 + �3 ) Similarly. we need to solve Eq. a12 + a22 + a32 = 1. for n e e 1 5 � � � λ3 = −5. (1. (2a3 )2 + a2 = 1 3 ⇒ 1 a3 = ± √ 2 a2 = ± √ 5 5 1 ® Therefore.68) for each l.70). (1. a2 = a3 = 0 and a1 unspecified. give a1 = 0 and a2 = 2a3. (1. Eqs. the eigenvector. To find the eigenvectors. corresponding to l1 = 2 is n 1 = ± ®1 . l2 = 5.68) for the given T is ˜ (2 − λ ) a1 + 0a2 + 0a3 = 0 (1. e For l2 = 5. n 2 . the ® ® eigenvector. Using Eq. (1. (1. which when solved give.75) become −3a1 + 0a2 + 0a3 = 0 0a1 + −2a2 + 4a3 = 0 0a1 + 4a2 + −8a3 = 0 which.74) � � �2 − λ � 0 0 � � � � 4 3−λ � 0 �=0 � � � 0 4 −3 − λ � (1.75) 0a1 + (3 − λ ) a2 + 4a3 = 0 0a1 + 4a2 + (−3 − λ ) a3 = 0 0a1 + 0a2 + 0a3 = 0 0a1 + 1a2 + 4a3 = 0 0a1 + 4a2 + −5a3 = 0 So. n 1. and l3 = -5 . for l1 = 2. 5 .73) yielding l1 = 2.18 INTRoduCTIoN To CoNTINuuM BIoMeChANICS which becomes or (2 − λ ) [(3 − λ ) (−3 − λ ) − 16] = 0 (2 − λ ) (λ 2 − 25) = 0 (1.70). But. so a1 = ± 1. Thus.

2) e e e ˜ � � � � � � � � � � � � � � � � FIguRe 1. the original matrix [T ] in the ®1 . T11 = n1 · T n1 = n1 · (λ1 n1 ) = λ1 ∼ T22 = n2 · T n2 = n2 · (λ2 n2 ) = λ2 ∼ T33 = n3 · T n3 = n3 · (λ3 n3 ) = λ3 ∼ T12 = n12 · T n1 = n1 · (λ3 n2 ) = 0 ∼ Similarly. ˜ . ®2 . there exist three principal directions that are mutually perpendicular. Theorem: The eigenvalues of any real symmetric tensor are real and called principal values. From Eqs. using n 1 as the basis.2: Reference coordinate system for T . 1.35) and (1. ® ® ® Let n 1 . they are used to determine the directions and magnitudes of maximum and minimum stresses and strains. T13 = T21 = T23 = T31 = T32 = 0.65).TeNSoR CALCuLuS 19 Why have we reviewed the “Eigen-stuff ”? As we will see in Chapters 2 and 3. n 3 be the principal directions (eigenvectors) of a real symmetric tensor T . and rate of deformation tensor.3. ˜ ® (1. strain tensor. The real eigenvectors (at least three) of any real symmetric tensor are called principal directions. such as the stress tensor. ®3 coordinate system (Figure 1.3 Principal Values and Principal directions Continuum biomechanics deals with real symmetric tensors. n 2 . Thus. Theorem: For a real symmetric tensor.

is a cubic equation in l.80) . I2. I3 are invariants given by I 1 = T11 + T 22 + T33 = Tii The scalar characteristic equation of a tensor T . ˜ 1.I1 l2 + I 2l .3) as � � T � ∼ n 1.79) I 3 = det(T ) ˜ (1.3: Rotated coordinate system in which T is diagonal.l dij | = 0.20 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 1. where the diagonal elements are the principal values (eigenvalues) of [T ]. ˜ ® ® ® can be written in a rotated n 1.77) (1.76) which is a diagonal matrix. n 3 coordinate system (Figure 1.4. n 3 � � ⎤ λ1 0 0 ⎥ ⎢ = ⎣ 0 λ2 0 ⎦ 0 0 λ3 ⎡ (1.1 Scalar Invariants of a Tensor and the Cayley-hamilton Theorem l3 . ˜ (1. n 2.I3 = 0 where I1.78) � � � � � � �T � � � � � � 11 T12 � � T22 T23 � � T11 T13 � I2 = � �+� �+� � � T21 T22 � � T32 T33 � � T31 T33 � and = (T11 T22 − T21 T12 ) + (T22 T33 − T32 T23 ) + (T11 T33 − T31 T13 ) 1 = (Tii Tjj − Tij Tij ) 2 (1. n 2 . |Tij .4 oTheR uSeFuL TeNSoR ReLATIoNShIPS 1.

1 can all be written in terms of tr (T ) ˜ (see Problem 17).83) (1. ˜ ˜ Also. (1.88) (1. t (e.85) (1. denoted as tr (T ). for the 3x3 tensors we have been considering. a is a scalar and a is a vector.g. ˜ ˜ ˜ ˜ d dS dT (TS ) = ∼ ∼ + T ∼ S ∼ ∼∼ dt dt dt d dT dS (T + S ) = ∼ + ∼ ∼ ∼ dt dt dt � dα d� dT α (t)T = T+α ∼ ∼ ∼ dt dt dt d � da dT � (T a) = ∼ a + T ∼ ∼ dt dt dt d T (T ) = dt ∼ ® In these expressions.86) (1. T 3 − I1T 2 + I2 T − I3 ∼ = 0 I ∼ ∼ ∼ (1.82) 1.TeNSoR CALCuLuS 21 Alternatively.89) � dT ∼ dt �T .. tr (T ) = Tii = T11 + T22 + T33 ˜ tr (T ) = tr (T T ) ˜ ˜ (1.2 Trace of a Tensor The trace of a tensor T . I1 = λ1 + λ2 + λ3 I2 = λ1 λ2 + λ2 λ3 + λ3 λ1 I3 = λ1 λ2 λ3 (1. Then.84) It is notable that the invariants given in section 1.3 Tensor-Valued Functions of a Scalar Let T = T (t) and S = S (t) be tensor-valued functions of a scalar. It says that a square matrix satisfies its own characteristic equation. 1. Thus.4. time).4.81) The Cayley–Hamilton theorem is a useful result from linear algebra.4.87) � (1. is the sum of the diagonal terms.

e.22 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 1. div n (r ). is defined to be a scalar field given by the trace of the gradient ® of n . density. (1. electric potential) at that point. displacement or velocity fields). is a vector field whose form in Cartesian coordinates is ∂ φ �i .g.94) In Cartesian coordinates. be a scalar-valued function of the position vector ®.95) .91) ® Let n (r ) be a vector-valued function of position (e. x2.92) ∂ ν1 ⎢ ∂x ⎢ 1 � � ⎢ ∂ν � ⎢ ∇ν = ⎢ 2 ⎢ ∂ x1 ⎢ ⎣ ∂ ν3 ∂ x1 ® ⎡ ∂ ν1 ∂ x2 ∂ ν2 ∂ x2 ∂ ν3 ∂ x2 ⎤ ∂ ν1 ∂ x3 ⎥ ⎥ ∂ ν2 ⎥ ⎥ ⎥ ∂ x3 ⎥ ⎥ ∂ ν3 ⎦ ∂ x3 (1. In Cartesian coordinates ® � � ∂ vi � ∇ν = ij ∂ xj or in matrix notation..e.4. i. f (r ) ® describes a scalar field. Ñf(r ). � div(ν ) = ∂ v1 ∂ ν 2 ∂ ν 3 ∂ νi + + = ∂ x1 ∂ x2 ∂ x3 ∂ xi (1.. f gives the value of a scalar (e. temperature.93) ® ® The divergence of n (r ). f(x1. is a second-order tensor. e ∂ xi ⎤ ⎡ ∂ /∂ x 1 ∂ � ⎥ ⎢ (1. i.4 gradient and divergence of Scalar. Vector. ⎤ ∂ φ /∂ x1 ∂φ � ∂φ � ∂φ � ∂φ � ⎥ ⎢ ∇φ = e1 + e2 + e2 = e i = ⎣ ∂ φ /∂ x2 ⎦ ∂ x1 ∂ x2 ∂ x3 ∂ xi ∂ φ /∂ x3 ⎡ ® (1. For each position r ® r .90) ∇≡ e i = ⎣ ∂ /∂ x 2 ⎦ ∂ xi ∂ /∂ x 3 So. and Tensor Fields ® ® Let f (r ). the ® gradient of n . Ñn .. the gradient of f..g. ® div (n ) tr (Ñn ) = Ñ× n ® ® ® (1. x3).

� � ∂ Tim � (1. PRoBLeMS ⎡ .5 ⎤ 7 0 11 � � 1. How many equations are described? How many terms are on the right-hand side? Which are the free indices? Which are the dummy indices? Consider the expression Tij = Aijmn xm xn + BipCjp. Given S = ⎢ 4 2 1 ⎥ ⎣ ⎦ ∼ 0 6 2 Evaluate (a) Sii (b) SijSij (c) SjkSjk (d) SmnSnm 2. In Cartesian ˜ coordinates.96) ∇·T= ei ∼ ∂ xm ® Note that the expressions for the gradient and divergence have different forms in cylindrical and spherical coordinates from those given for Cartesian coordinates. How many equations are described? How many terms are on the right-hand side? Which are the free indices? Which are the dummy indices? Now answer a)–d) again for Tij = Aijmn xm yn + Biprt Ejt cpdr. let T (r ) be a tensor field. The divergence of a tensor field is a vector field. (a) (b) (c) (d) 3.TeNSoR CALCuLuS 23 Finally. Summary: Gradient increases rank • scalar vector • vector tensor Divergence reduces rank • tensor vector • vector scalar 1. (a) (b) (c) (d) (e) Consider the expression ja = fa + Dab eb.

⎣ ⎦ ∼ ˜ (Hint: You will need three orthogonal eigenvectors. ∼ 0 0 −10 Find its scalar invariants and then evaluate the eigenvalues using l 3 . Given . Evaluate dim dmj djn dnx. ⎤ ⎡ 5 16 4 � � ⎥ ⎢ T = ⎣0 8 11. and find its correspond√ 0 67 5 −1/ 2 ing eigenvalue. (You can use MatlabÓ’s “roots” command to solve the cubic equation. How many equations are described? How many terms are on the right-hand side? What are the free indices? What are the dummy indices? ® e ae 5. Given that Tij = 2mEij + lEkk dij .) 0 0 6 ⎡ ⎤ 7 1 3 � � ⎢ ⎥ 13. Find the eigenvalues and eigenvectors for the tensor T .I3 = 0. Consider 2 ⎦.) ⎡ ⎤ 8 0 0 � � 12. In this case. Show that⎣ 0 ⎦is an eigenvector of the matrix⎣ 6 −17 6 ⎦. What is this value? 7. (a) (b) (c) (d) Consider the expression Tijk = Aijkmn xm xn. j. show that (a) W = 1 Tij Eij = μ Eij Eij + 1 λ (Ekk )2 2 2 2 (b) N = Tij Tij = 4μ Eij Eij + (4μλ + 3λ 2 ) (Ekk )2 � � � � � � � � 9. why does this not violate basic algebraic rules? 6. T = ⎣8 3 2⎦ ∼ 2 8 1 . 8.I1l 2 + I2l . Show that gijk ci cj ck = g111 + g121 + g211 + g221 + g112 + g122 + g212 + g222 for a particular value of c1 and c2.24 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 4. Why can we say ai = a × ®i and a = ®i ®i? In other words. where T = ⎢ 0 8 0 ⎥. i. Show that a · Tb = b · T T a = Tij ai bj ∼ ∼ ⎡ ⎤ ⎡ √ ⎤ 5 13 0 1/ 2 ⎥ ⎢ ⎢ ⎥ 10. and k take on the values 1 and 2.

Eqs.97) T = sph(T ) + dev(T ) ˜ ˜ ˜ sph(dev (T ))=0 ˜ Is sph(T ) symmetric or antisymmetric? Describe why dev(T ) is symmetric only if T is ˜ ˜ ˜ symmetric. (0. (a) Find the matrix representation of T . 14. 0). The invariants of a second-order tensor can all be expressed as a function of the trace of the tensor and the trace of the tensor squared or cubed. ˜ (d) Show that T is a reflection by calculating its determinant. (b) Based on your answer to (a).64) present the decomposition of a tensor into its symmetric and antisymmetric parts. Another important tensor decomposition is into spherical (or hydrostatic) and deviatoric parts. (c) Why is A not an orthogonal tensor? ˜ 17. and (0. ˜ (b) A plane intersects the original coordinate axes at (1.4 helpful. ˜ show that I2 and I3 can be written as . Find the vector. let ® be ˜ perpendicular to the plane of reflection). ˜ 16. A tensor. 0. The unit ® vector normal to this plane is transformed by A . 0. e3 15.e. n. 1). You will find Figure 1. 1.TeNSoR CALCuLuS 25 (a) Decompose the tensor into its symmetric and antisymmetric parts. Given the following definitions for the spherical and deviatoric components. 0). provide a general statement about Tii if a tensor is antisymmetric. ˜ (c) Prove that T is an orthogonal tensor. transforms a Cartesian coordinate system by rotating it 30° counterclockwise ˜ about the ®2 axis and extending vectors along the ®2 axis to three times their original e e length. What is I1 in terms of tr (T )? Also. sph(T ) = ∼ verify (a) (b) (c) 1 tr(T )I 3 ∼ ∼ and dev(T ) = T − sph(T ) ∼ ∼ ∼ (1. resulting from this ˜ transformation. (1. ˜ Note: This decomposition is helpful when modeling the behavior of incompressible materials.. A . T transforms every vector into its mirror image with respect to the 1–2 plane (i. (a) Find A . ˜ (b) Verify that T is indeed a tensor (or linear transformation). (d) Decompose T from problem 13 into its spherical and deviatoric parts.

132. I2 = �2 � �� 1 �� tr(T ) − tr T 2 ∼ ∼ 2 � � �3 1 � 3� 3 � 2� 1� I3 = tr T − tr T tr(T ) + tr(T ) ∼ ∼ ∼ 3 2 ∼ 2 (1.. In this problem.26 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 1. 213) ⎪ ⎩ if any two of ijk are equal (i. 113. Show that the trace satisfies tr (aS + bT ) = a tr (S ) + btr(T ). e e e . we described the dot product of two vectors in indicial notation.. 123. Show that the cross product of two vectors is given by ® e a ´ b = (eijk aj b k ) ®i ® (1.100) (1.a2 b1) ®3.. 232. 312.4: Relationships among the lengths of the legs of a 30°-60°-90° triangle. ⎧ ⎪ 1 when ijk is an even permutation of 123 (i.e.99) where eijk = ®i×(®j ´ ®k).101) Recall. ˜ ˜ ˜ ˜ 19.e. 231) ⎨ εijk = −1 when ijk is an odd permutation of 123 (i. e e e or equivalently.e.(a1 b3 . 333) 0 (1.98) (Hint: Use the Cayley-Hamilton theorem for I3) 18. we will examine the cross product of two vectors. 321.a3 b1) ®2 + (a1 b2 . the cross product of two vectors is (a2 b3 .a3 b2) ®1 . In this chapter.

q is the angle between the vectors ® and b . Tensors can be formed from the outer product of two vectors. ® ® ® 20. 21. or tensor) and 2) express the operation with its result in indicial notation with respect to Cartesian coordinates. The outer product. vector. ˜ ˜ • • • • .. What type ˜ of field (scalar. and n is a unit vector perpendicular to a ® ® ® ® a both ® and b chosen such that the set (a . tensor) is the left-hand side? The right-hand side? It is important to remember that both sides of an equation must match in terms of units and type of field. Show that any tensor T can be expressed as T = Tij (ei Ä ej ). Show Ñ×(a f I ) = aÑf .TeNSoR CALCuLuS 27 Note: An alternate definition of the cross product is a ´ b = (ab sin q )n . or dy® adic product. It also conforms to the fola ® ® ® lowing rules when forming inner products: (® Ä b )× ® = ®(b × ®) and a (b Ä ®) a c a c c ® ® ® ® ® = (a × b ) c . scalar. of two vectors is formed as [® Ä b ]ij = ai bj . b . (a) Gradient of a scalar (b) Gradient of a vector (c) Divergence of a vector (d) Divergence of a tensor 22. For each the following. where a is a constant and f is a scalar valued function. vector.e. 1) indicate what the operation produces (i. where a and b are ® ® magnitudes. n ) make a right-handed triad.

.

body (configuration). If we describe tensor quantities with respect to the origin. with respect to the origin or its reference position.a2 .a2 . we want to know the instantaneous velocity field (and other properties) and its evolution in time with respect to a current position of interest. Thus. The position of a particle at t = t0 is described by its position vector a . In this chapter. velocity.29 CHAPTER 2 Kinematics of a Continuum 2. we use an Eulerian (spatial) description. When we describe the flow of water in a river. we are using the Lagrangian (material) description. let the position vector of the particle be x (i.. or other quantities. x = x (a . the particle moves). Thus.a3 . having followed the individual particles during the motion.a3 . and time. whereas the Lagrangian method follows the trajectory of each individual particle. t) ® ® ® ® ® ® ® ® ® ® ® ® ® ® (2. x2. we will develop the basic kinematic concepts essential to continuum mechanics. as time goes on. t0). Also. Let O(x1. and where a = a1e 1 + a2 e 2 + a3 e 3 and x = x1e 1 + x2 e 2 + x3e 3. ® measured from a fixed point.t) (2.2) .t) x2 = x2 (a1 . A body at time t = t0 occupies a cer® tain region of physical space. If we use a particular position of the object at some time.a3 . x3) be a fixed frame as shown in Figure 2. The Eulerian method analyzes what happens at every fixed point in space.a3 . and acceleration.a2 . Instead. xi = xi (a1 .a2 . a2.t) or with ai = xi (a1.1. (2. a3.1 deSCRIPTIoN oF The MoTIoN oF A CoNTINuuM Kinematics is the study of motion dealing with position. O. t0 ) = a . Eq.t) x3 = x3 (a1 .1) with x (a .1) becomes x1 = x1 (a1 . we do not desire to identify the locations from which each particle of water originates. we are not interested in describing the water’s velocity. At a later time t.e. Kinematic analyses are based on relationships among position.

a2. We can depict these changes using Lagrangian (material) or Eulerian (spatial) descriptions. x2. a2. SPATIAL deSCRIPTIoN When a continuum is in motion. n as functions of the particles). x3. (2. a3 are the material coordinates (i. they define the path line or trajectory followed by that particle. we ® follow the particles and express q. a particle is located at material coordinates a1. a3) identifies the different particles of the body and are said to be the material coordinates. t) n = n (x1. Eq. For a specific particle. At t = t0. motion has moved this particle to coordinates x1.e.. a2. t) n = n (a1. velocity n . x3 ® defining the position vector x .1) or (2. At a later time. (2. x2. t) ® ® Every particle is described by its coordinates at time t0. etc. quantities that are associated with specific particles (temperature ® q.2) is said to define a motion for a continuum. a1. position. 2. Eulerian (Spatial) description—spatial coordinates are the independent variables q = q (x1. t) ® ® . x3.30 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 2. a2. 1.) change with time and. t. a2. x2. a3.2). the triple set (a1. a3. a3 ® defining the position vector a .1: Motion of a continuum. In Eq. 2. Lagrangian (Material) description—material coordinates are the independent variables q = q (a1.2 MATeRIAL VS. perhaps.

a2.2: Unit cube subjected to the motion described by x = a + kta2 ®1. At t = t ¢.0). Demonstration: ® ® ® Consider the motion x = a + kta2 e 1.KINeMATICS oF A CoNTINuuM 31 x1. If the temperature is given by the spatial description. we observe the changes at fixed locations). x3 are known as spatial coordinates (i. Note: The material and spatial descriptions are related by the motion. at t = 0.0). Sketch the configuration at time t ¢ for a body which. (b) Obtain the velocity and rate of change of temperature for particular material particles.e.. shown in Figure 2.0).0. Spatial positions are occupied by different particles at different times. x2. x3)O = (0. 1. x2. i.2 where the material coordinates (a1. the coordinates of O are (x1. Solution: 1.0. point O remains at (0.. a2. 2. the coordinates of point O are (a1. Express the answer in both material and spatial coordinates. e . has the shape of a cube of unit sides. q = x1 + x2 . Break down the equations of motion into x1 = a1 + kta2 x2 = a2 x3 = a3 At t = 0. (a) Find the material description of q.e.0. a3)O = (0. a3) give the position of a particle at t = 0. ® ® FIguRe 2.

Any particle on material line OC has position (a1. point A is at (a1. any particle on material line CB has position (a1.1. n1 = ka2. looking at the different lines.0) at t = t ¢.0. 2. The material de® rivative is a derivative taken with respect to a coordinate system moving with velocity n .0) at t = t ¢. a2. namely. as x2 = a2 . a3)OC = (0. Line OA is stationary. the velocity of that particle is given by So. Now. every particle on line OC moves horizontally to the right through a distance linearly proportional to its height. change of field quantity independent of time due to “flow” from one spatial coordinate to another. material line BA moves as line OC. So. (a) q = x1 + x2 and x = a + kta2 e 1. n2 = n3 = 0. a2. x3)A = (1.0. � ∂ xi � � νi = ∂ t �ai fixed � � � ∂θ � ∂ � (x1 + x2 )� = = ν1 + ν2 = ka2 = kx2 � � ∂ t ai fixed ∂t ai fixed Note: As q = x1 + x2 . allows us to introduce the material derivative. x3)OC = (a2k t ¢. n2 = n3 = 0 is the material description. a2. Furthermore.0. x3)OA = (a1. So.0. x2.0). a2. So. point A is at (x1.0) at t = 0 and (x1. a3)CB = (a1. To calculate the rate of temperature change for a material particle. Thus. the material description of q is ® ® ® q = (a1 + kta2) + a2 = a1 + (kt + 1) a2 (b) Because a particular material particle is designated by a specific ai . it is independent of time in the spatial description. x2. (a1. The material derivative (also called the Lagrangian or convective derivative) is defined as the time rate of change of a quantity (such as temperature or velocity) associated with material particle. a3)A = (1. So. a2.32 INTRoduCTIoN To CoNTINuuM BIoMeChANICS At t = 0. and it is .0) at t = 0 and (x1.0) at t = 0 and (x1. Similarly. a2.0) at t = t ¢.3 MATeRIAL deRIVATIVe The concept just presented.0).1. for any particle on OA. 2. x2. However. each particle experiences temperature changes as it flows from one spatial coordinate to another. a3)OA = (a1. x3)CB = (a1+k t ¢. the spatial description is n1 = kx2 . This motion is known as simple shearing motion. At t = t ¢. line CB is displaced to the right by distance kt ¢. A also remains stationary. x2.

where the are obtained with fixed values of ai . Dt ® ® ® Solution. Consider a particle at position xi at time t that has with it an associated value (e. The velocity was found ® ® q to be n = kx2 e 1. t + dt. q = q (a1.. Find the material derivative... x2. So. xi + ni dt. It is denoted by D . ∂t ∂ t �ai fixed Thus.3) � = Dt ∂ t �ai fixed However. = + ν · ∇θ . and the temperature field was given as q = x1 + x2. Demonstration.g. In Cartesian coordinates. then � Dθ ∂θ � (2.KINeMATICS oF A CoNTINuuM 33 often used in fluid and classical mechanics. Consider the previous example with motion x = a + kta2 e 1. D . a3. then � ∂ xi ∂ xi � � = νi .5) � � Dθ ∂θ � ∂ θ ∂ x1 ∂ θ ∂ x2 ∂ θ ∂ x3 ∂ θ � � � = + + + = Dt ∂ t �ai fixed ∂ x1 ∂ t ∂ x2 ∂ t ∂ x3 ∂ t ∂ t �xi fixed (2. a2 .4) where the last equality has made use of Eqs. x3.24) and (1. At a later time. where ∇θ = ∂ xi Dt ∂t Dθ � � � = 0 + (kx2 e 1 ) · ( e 1 + e 2 ) = kx2 Dt . and n is a vector valued function of stationary spatial coordinates. ∂θ � Dθ ∂θ � � � e i = e 1 + e 2 . q = q (x1. i. t ).t) due to both change in ti me and spatial location. The material derivative captures the nature of this change in the infinitesimal limit. (2. t + dt). in these equations. i. t ). Dθ ∂θ ∂θ ∂θ ∂θ ∂θ ∂θ ∂θ � + ν2 + ν3 = = = + ν1 + νi + ν · ∇θ Dt ∂t ∂ x1 ∂ x2 ∂ x3 ∂t ∂ xi ∂t (2. Remember. a2. t ) with xi = xi (a1. (1.e. t + dt) –q(xi. and has associated with it the value q(xi + ni dt. a3. q is a ® scalar valued function of stationary spatial coordinates. scalar or vector) q(xi .e.90).t). the particle has moved to a new location. The associated value q has changed by an amount q(xi + ni dt.5) means. When the material description of Dt the quantity is used. when the spatial description of the quantity is used. Let us pause for a moment to consider what Eq.

P and Q.g. The spatial. frame is a way of looking at motion that focuses on specific points in the space through which the material moves. u ( a + d a . point Q undergoes a displacement. Plotting the position of an individual particle through time gives the pathline of the particle.6) Similarly.g.. We are interested in how material points close to each other change their relative positions during a deformation. 2. . | | x = a + u ( a . or Eulerian reference. and arrives at its new position. As an illustration. d® small. changing to another configuration at time t as shown in Figure 2. ® ® ® FIguRe 2. consider sitting on the bank of a river and watching the water pass by your location.. u . reference frame is a way of looking at motion where the observer follows individual material (e.34 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Summary. and arrives at its new position. fluid) particles as they move through space and time.3: Deformation of a Continuum. They are related by the material derivative Dθ ∂θ � = + ν · ∇θ Dt ∂t which describes the rate of change of q while moving with the material (e. or Lagrangian. An example of this approach is if one is on a boat drifting down a river.4 deFoRMATIoN-INduCed STRAIN Consider a body of particular configuration at some reference time t0. initially close to each other. fluid flow) are determined as vectors at discrete locations. fluid) at velocity n .3. u = x . The motion of a material particle is described by its dis® placement. Values about the motion (e.a .. a ® ® ® Point P undergoes a displacement. t ) ® ® ® ® (2. The material.g.e. i. Consider two points on this body.t )..

u1 ( a . a2 + da2 . a3 .a2 + da2 . t) − u1 ( a.12) .y.11) dx3 = da3 + u3 (a + da. u1 (a + da . t) − u2 ( a.y. a2 .t)] + [u1 (a1 + da1 .t) − u1 (a1 .7). t) = u1 (a1 + da1 . t ) ® ® ® (2.t ) ® ® ® ® ® ® ® (2.a2 .a2 .a3 .t)] + [u1 (a1 + da1 . � � � � � ∂ u1 ∂ u1 ∂ u1 da1 + da2 + da3 ∂ a1 ∂ a2 ∂ a3 � (2.d a + u ( a + d a . t ) = da1 + Similarly.a3 . a3 . t ) . a2 .7) Subtracting Eq.t) − u1 (a1 + da1 . t) − u1 (a1 .z) − F (x. a3 + da3 .KINeMATICS oF A CoNTINuuM 35 x + d x . a3 + da3 . t) can be rewritten as � � � u1 (a1 + da1 .a2 . t ) − u1 (a1 .t) − u1 (a1 + da1 . t ) = da2 + � � � � � � ∂ u2 ∂ u2 ∂ u2 da1 + da2 + da3 ∂ a1 ∂ a2 ∂ a3 ∂ u3 ∂ u3 ∂ u3 da1 + da2 + da3 ∂ a1 ∂ a2 ∂ a3 � � (2. a2 + da2 . (2.10) dx2 = da2 + u2 (a + da. = lim dx→0 ∂x dx � u1 ( a + da.a . t ) ® ® ® ® ® ® (2.t) = [u1 (a1 + da1 .a2 . t ) . t ) − u1 ( a. t ) = ∂ u1 ∂ u1 ∂ u1 da1 + da2 + da3 ∂ a1 ∂ a2 ∂ a3 which leads to dx1 = da1 + u1 (a + da. Then. let us examine the first of the three equations it represents. t ) = da3 + � � � � (2. d x1 = d a1 + u1 ( a + d a .6) from (2. t) = u1 (a1 + da1 .a2 + da2 .a3 + da3 .a3 .a3 .8) To understand Eq.a2 + da2 .8). t) − u1 (a1 . t) − u3 ( a.a3 .a2 + da2 . a3 . (2.a3 + da3 .u ( a .9) Now. t) − u1 (a . we find that ® d x = d a + u ( a + d a .z) .t)] Recall from multivariable calculus that � � ∂F F (x + dx.

14) Ñ u is the displacement gradient. which is the definition of the gradient of a vector.91) as ( ∇u )ij = → ® ∂ ui ∂aj (2. The Eulerian strain tensor. (2. For completeness.13). we see that u (a + ® ® ® ® ® d a . (1. is identified as a tensor describing the deformation.13). using F .12) can be written more compactly as dx = da + ( ∇u )da � � � � (2. The components of ® the gradient of u are given by Eq. Ñ u . Then. t ) - u ( a .11).18) We will return to these equations when we discuss general elasticity.17) (2.4). The left Cauchy–Green deformation tensor. by Eq.36 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Eqs. Note that Ñ u is ® ® ® ® a tensor. which become through the motion. ˜ B = FF T ˜ ˜˜ 3. Now.10). d a 1 and d a 2. These tensors are 1. (1. t ) = (Ñ u ) d a .13) ® ® ® where the gradient of u . d x 1 = da 1 + (∇u )da 1 d x 2 = da 2 + (∇u )da 2 � � � � � � � � (2. d x 1 and d x 2 (see Figure 2. (2. we can define several finite defor˜ ˜ mation and finite strain tensors. Comparing Eqs. (2. consider two material vectors ® ® ® ® issuing from P.8) and (2. The right Cauchy–Green deformation tensor. X. The components of the deformation gradient tensor are given as Fij = ∂ xi ∂ aj (2.19) ® ® A measure of the deformation is given by the dot product of d x 1 and d x 2. but (Ñ u )d a is a vector (Eq. ˜ C = F TF ˜ ˜ ˜ 2. There is a closely related tensor known as the deformation gradient (see Problem 3). C . ˜ (2.28)). and (2. B . (2.16) Ξ= ∼ 1 (I − B−1 ) 2 ∼ ∼ (2.15) F is clearly not a symmetric tensor. .

21) d x 1 · d x 2 = da 1 · da 2 + 2(da 1 ) · (E *da 2 ) ∼ ® ® ® ® (2. ˜ ˜ ® ® ® ® dx 1 · d x 2 = da 1 · da 2 + da 1 · (∇u )da 2 + da 1 · [(∇u )T da 2 ] + da 1 · (∇u )T [(∇u )da 2 ] � � � � � � � � � � � � � � � � � � � � � � � � � = da 1 · da 2 + da 1 · {(∇u ) + (∇u )T + (∇u )T (∇u )}da 2 = da 1 · da 2 + da 1 · (2E *)da 2 ∼ � where we have defined E* = ∼ Thus. a (T b )= b ·(T T a ). displacements are small. (2.KINeMATICS oF A CoNTINuuM 37 FIguRe 2.22). for the transpose of a tensor. (1. which characterizes deformation in the neighbor˜ hood of particle P based on a material description. then dx 1 · dx 2 = da 1 · da 2. if E * = 0. E * is the finite Lagrangian strain tensor. the higher order product (Ñ u )T (Ñ u ) can be neglected and ® ® .. Thus.22) From Eq.48).20) Recalling Eq.23) E∗ = + + ij 2 ∂ aj ∂ ai ∂ ai ∂ aj For infinitesimal strain. lengths of and angles between material ˜ lines remain unchanged. i. The components of E * with respect to Cartesian ˜ coordinates are � � ∂ uj ∂ uk ∂ uk 1 ∂ ui (2. � � 1� � � � � (∇u ) + (∇u )T + (∇u )T (∇u ) 2 � � � � � (2. d x 1 · d x 2 = da 1 · da 2 + da 1 · (∇u )aa 2 + da 2 · (∇u )da 1 + [(∇u )da 2 ] · [(∇u )da 1 ] � � � � � � � � � � � � � � (2.4: Relative motion of two material vectors during the deformation of a continuum.e.

g. ˜ ® 2E12 gives the decrease in angle between two elements initially in the x1 and x2 directions).26) in matrix form is � � � 1 ∂ u1 + ∂ u2 1 ∂ u1 + 2 ∂ x2 ∂ x1 2 ∂ x3 ⎢ ⎢ � � � ⎢ ∂ u2 1 ∂ u2 + [E] = ⎢ 1 ∂ u1 + ∂ u2 ∼ ⎢ 2 ∂ x2 2 ∂ x3 ∂ x1 ∂ x2 ⎢ ⎣ � � � � ∂ u3 1 ∂ u1 + ∂ u3 1 ∂ u2 + ∂ u3 2 ∂ x3 ∂ x1 2 ∂ x3 ∂ x2 ∂ x3 ⎡ ∂ u1 ∂ x1 �⎤ ∂ u3 ∂ x1 ⎥ �⎥ ∂ u3 ⎥ ⎥ ∂ x2 ⎥ ⎥ ⎦ (2. Eq. in the x1. ˜ The off-diagonal elements of E give the decrease in angle between elements.. Thus. and x3 directions.e. . shear strain (e.28) The diagonal elements of E give unit elongations. ∼ 2 Note: Under the assumption of infinitesimal deformation.24) So. then E = 0. (1. normal strains. Eq. � � ˜ ˜ � � The infinitesimal rotation tensor is given by Ω = 1 (∇u ) − (∇u )T .e.26) E is the infinitesimal strain tensor (note the absence of the *). (2.26). referred to a Cartesian ˜ system.27) Eij = + + = 2 ∂ aj ∂ ai 2 ∂ xj ∂ xi where the last equality results from ai » xi . (2. also reduces to Eq.22) is d x 1 · d x 2 = da 1 · da 2 + 2(da 1 ) · (E da 2 ) ∼ � � � � � � (2. i.62).25) where E≡ ∼ � 1� � � � (∇u ) + (∇u )T = (∇u )symm 2 (2. W . E is linear in the ˜ displacement gradients. Its components. (2. i. for small deformations. (2. Eq.. are � � � � ∂ uj ∼ 1 ∂ ui ∂ uj 1 ∂ ui (2.. recall Eq. as material and spatial coordinates can be assumed identical because particles do not move much during small deformations. If Ñ u is antisymmetric.38 INTRoduCTIoN To CoNTINuuM BIoMeChANICS E* ≈ ∼ � 1� � � � (∇u ) + (∇u )T = (∇u )symm 2 (2. the Eulerian strain tensor. and we have an infinitesimal rigid body rotation. x2.18).

1: Simple deformations TyPe oF deFoRMATIoN x1 or r x2 or q x3 or z Uniform extension or compression Uniform dilatation Simple shear Pure torsion Pure bending Plane strain l1a1 l1a1 a1 + a2tang R r (a1) x1(a1. These are the eigenvalues of E . ˜ This concludes our discussion of deformation and displacement. i. (1.n 1 2 3 0 0 E3 E1.77). e is the first invariant of E . Thus. They are obtained from the characteristic equation of E .n .a2) l2a2 l1a2 a2 Q + cZ l3a3 l1a3 a3 Z Z a3 q(a2) x2(a1.5 PRINCIPAL STRAIN Because the strain tensor. there exist at least three mutually perpendicular direc˜ ® ® ® tions. Table 2.29) [E]� � � = ⎣ 0 E2 0 ⎦ ∼ n .30) It can be shown (see problem 11) that where dV represents a small material volume. and E3 are the principal strains of E .e. n 1. E2. n 3 with respect to which the matrix of E is diagonal (recall Eq. n 2. ˜ recall Eq. E . e = Eii . e is equal to the change in volume per unit volume.6 dILATATIoN e = Eii = Δ(d V ) dV (2. Furthermore.a2) . is symmetric. 2.KINeMATICS oF A CoNTINuuM 39 2.1 represents idealized states of deformation commonly discussed in problems or used as testing regimens. ˜ ⎤ ⎡ E1 0 0 ⎥ ⎢ (2.. and they include the ˜ ˜ maximum and minimum normal strains. (1.76)). TABLe 2.

t Dt ∂t So. We will see this in action several times later in this text. li’s are constants known as stretch ratios. θ = tan x2 . t = (∇ν )d a Dt (2. (2. t = ν a + d a. the velocity of a particle should be the same whether a material or spatial description is used. Eq.3).31) Recall that the material derivative is taken with respect to ai fixed.g.1 gives idealized deformations without reference to a particular material model. t) − u (a . t − ν a. We wish to compute the rate of change of its length and direction. � D � �� D �� � D � �� � � � dx = da + u (a + da . the stress generated in the material.7 RATe oF deFoRMATIoN ® ® Consider a material element. we know that the time rate of change of the displacement of a particle is its velocity.8). the Cartesian matrix representation of the velocity gradient tensor. we must have a constitutive equation describing the relationship between stress and strain for that particular material. There® fore. and z(Z) refer to cylindrical coordinates. 2. e. In this table. q(Q). Hence. t) for which Eq. can be written as . t) Dt Dt Dt (2. located at x at time t. Eq. x1 = l1a1 indicates that the body is stretched (or compressed) l1 times its original length in the x1 direction. Thus. Ñn . from basic physics. P. dx . Table 2. �� � � � r(R). we get (2.33) is true when a is replaced by x on the right-hand side. taking the material derivative of Eq. t − ν a. constitutive equation.g...33) Now. i. With this table.40 INTRoduCTIoN To CoNTINuuM BIoMeChANICS In Table 2. t − u a. (2. issuing from material point. where r = x 2 + x 2 .. F can be determined for a deformation. For simple shear. (2.31) becomes � � � � � �� � D � �� ∂ �� �� � � � � � � � � dx = u a + d a. Principal strains and their corresponding directions can then be found. Finally. and 1 2 x1 z = x3. To determine the behavior of a material in response to any of these deformations. e.32) � � � D � �� � � � � � � � � d x = ν a + d a. Thus.e. comparison with the demonstration on simple shearing motion should convince you that kt = tang. there is a velocity ® ® ® ® function n ( x . (2.1. pure torsion and bending are given in terms of cylindrical coordinates. Also. though the functional form of the velocity equation may be different. along with the sev˜ eral deformation and strain tensors presented.

x2.KINeMATICS oF A CoNTINuuM 41 Ñn can be decomposed into the sum of its symmetric and antisymmetric parts. (2. or strain-rate tensor. ® ∂ ν1 ⎢ ∂ x1 ⎢ � ⎢ ∂ν [∇ ν ] = ⎢ 2 ⎢ ∂ x1 ⎣ ∂ ν3 ∂ x1 � ⎡ ∂ ν1 ∂ x2 ∂ ν2 ∂ x2 ∂ ν3 ∂ x2 ∂ ν1 ∂ x3 ⎥ ⎥ ∂ ν2 ⎥ ∂ x3 ⎥ ⎥ ⎦ ∂ ν3 ∂ x3 ⎤ (2. the continuity equation says.35) and (2. Then. its volume may change. which simplifies the solution of many problems in fluid mechanics. 2.g.37). extension rates in the ˜ ˜ x1. and. unlike for E .. but its total mass will remain unchanged. Let r denote the density of a particle and dV its volume. and W = (Ñn )asymm is the spin ˜ ˜ tensor.37) is an exact measure of the deformation rate.36) W= ∼ ® D = (Ñn )symm is the rate of deformation tensor. We will study fluids in some detail in Chapter 5.38) . Dij are linear in the velocity field. ˜ (2. and x3 directions. Eq.. shear rate (e. therefore.8 CoNTINuITy eQuATIoN (CoNSeRVATIoN oF MASS) If we follow a particle through its motion. The off-diagonal elements of D give the rate of decrease in angle between ˜ elements. the diagonal elements of D give the rates of elongation. there are no approximations made in arriving at Eq. However. i.. i. D (ρ dV) = 0 Dt (2.37) Similar to E . In matrix form.e. 2D12 gives the decrease in angle between two elements initially in the x1 and x2 directions).e.34) with ∇ν = D + W ∼ ∼ D= ∼ � 1� � � (∇ν ) + (∇ν )T 2 � 1� � � (∇ν ) − (∇ν )T 2 ® (2. D is ˜ � � � �⎤ ⎡ ∂ ν1 1 ∂ ν1 + ∂ ν2 1 ∂ ν1 + ∂ ν3 2 ∂ x2 ∂ x1 ∂ x1 2 ∂ x3 ∂ x1 ⎥ ⎢ ⎢ � � �⎥ � ⎢1 ∂ν ∂ ν2 1 ∂ ν2 + ∂ ν3 ⎥ [D] = ⎢ 2 ∂ ν1 + 2 ⎥ ∼ 2 ∂ x3 ∂ x2 ∂ x1 ∂ x2 ∂ x2 ⎥ ⎢ ⎣ � ⎦ � � � ∂ ν3 1 ∂ ν1 + ∂ ν3 1 ∂ ν2 + ∂ ν3 2 ∂ x3 ∂ x1 2 ∂ x3 ∂ x2 ∂ x3 (2.

Using E . u1 = ka22 and u2 = u3 = 0. Thus. dV Dt ∂ xi D .42 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Using the product rule. the infinitesimal strain tensor. 3. ˜ ∂ νi D ρ (2.c2 plane.e. Given the following displacement components. ρ where � � Dρ � div ν + =0 Dt (2. ® ® 2. Determine the ratio of the deformed to undeformed lengths of the differential elements (“stretch”) of part 2 and the change in angle between these elements. 4. ρ It can be shown that D Dρ (dV ) + dV =0 Dt Dt 1 D ∂ νi (dV ) = . Consider the unit square OABC shown in Figure 2.. Let k = 1 ´ 10 -4 and obtain E .39) ρ + =0 ∂ xi Dt In invariant form. FIguRe 2. Sketch the deformed shape of the unit square OABC. the first scalar invariant of the rate of deformation tensor.5. which is the rate of change of volume per unit volume. . d x 1 and d x2) of the material elements ® ® ® ® ® ® d a 1 = d a 1 e 1 and d a 1 = d a 2 e 2 initially at point C. ˜ ® ® 5. find the unit elongation of material elements d a 1 and d a 2. 1.40) Dρ ∂ρ � = + ν · ∇ρ Dt ∂t Demonstration. Find the deformed vectors (i. and find the decrease ˜ in angle between these two elements.5: Undeformed unit square in the c1 .

Note: d a 1= ⎣ 0 ⎦ and d a 2 = ⎣ da2 �2 ⎦ e . e . d �2 = d �2 + ∇ u d �2 . Material line OA: a2 = 0 Þ u1 = u2 = u3 = 0. CB is displaced by k to the right.6). = u2 = u3 = 0.25). lines become parabolic in shape (see Figure 2.2.KINeMATICS oF A CoNTINuuM 43 Solution: 1. So.0) is ® � � � ∇ u |@C ∂ u1 ⎢ ∂ a1 ⎢ ⎢ ∂u = ⎢ 2 ⎢ ∂ a1 ⎣ ∂ u3 ∂ a1 � � ⎡ ∂ u1 ∂ a2 ∂ u2 ∂ a2 ∂ u3 ∂ a2 ∂ u1 ∂ a3 ∂ u2 ∂ a3 ∂ u3 ∂ a3 � ⎤ ⎤ ⎡ � 0 da1 e 1 ⎢ ⎥ ⎥ ⎢ � � � Using Eq. As in the demonstration of Section 2.6: Deformed shape of unit square when subjected to displacement u = ka 2 ®1. The displacement gradient. Ñu . (1.1. (2. d �2 = 2kda � + da � . Material line CB: a2 = 1 Þ u1 = k. � � = 1 = 1 and � �1� da1 �d a � ® 2 FIguRe 2. x a a 0 0 d x 1= d a 1 + ∇ u d a 1 � ⎤� � � ⎥� ⎥� ⎥� ⎥� ⎥� ⎦� � � @C ⎤� 0 2k a2 0 � � ⎥� ⎢ = ⎣ 0 0 0 ⎦� � 0 0 0 �a ⎡ ⎡ 2 =1 ⎤ 0 2k 0 ⎢ ⎥ = ⎣0 0 0⎦ 0 0 0 ⎡ Therefore. So. So. Using Eq. 2. at point C = (0.19). x 2e1 2 e2 d x 1 = da1 e 1 + 0 � � � � � �1� �d x � da 3. OA is not displaced. Material lines OC and AB: u1 = ka22.

(2. � �1� � �2� �d x � �d x � � �π 2kda1 da2 d x1 · dx2 2k � �� � = � cos (θ ) = � � � � =� = cos − γ = sin(γ ) �1 �2 2 da1 da2 4k2 + 1 4k2 + 1 �d x � �d x � � � γ = sin −1 � � 4k2 + 1 2k � ≈ 1.44 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ® ® FIguRe 2. Essentially. � � � � �2� � �d x � (2kda2 )2 + (da2 )2 � �= = 4k2 + 1 = 1. So. � �2� da2 �d a � π To find the angle change. γ = − θ 2 � � d x1 · dx 2 (see Figure 2. From Eq.7: Definition of angles used to calculate the change in angle between da 1 and da 2 rotated clockwise during the motion.9999999 × 10−4. ∂ aj ∂ ai ∂ aj ∂ a2 4. Eij = ⎤ 0 k a2 0 ⎥ ⎢ [E] = ⎣ ka2 0 0 ⎦. recall the alternative definition of the dot product. there is very little stretch. Because 1 = 2k a2 .27). Note that this has been determined in radians. g. and all other + = 0.7) and cos(θ ) = � � � � .00000002 . Now. 1 2 � � ∂ ui ∂ uj ∂ ui ∂u . with k =1 ´ 10-4 ∼ 0 0 0 ⎡ .

that if the value of k was “large” in this problem. Consider the following description of motion. (Hint: write out the components and obtain a (x )) (c) Which description of motion. This angle decrease agrees with that calculated in part 3 as follows: γ = sin −1 � where the first approximation follows from small k and the second from sin-1( x ) » x for x 1.KINeMATICS oF A CoNTINuuM 45 � � 5. ® ® ® ® ® 2. a2 = 1. (a) Is this a material or spatial description of motion? ® ® (b) Find the inverse description. Fluid is flowing in a pipe. It should be emphasized. (a) Find the material description of pg. ® ® ® ® . and you would like to know the pressures and flow rates. Based on the displacement gradient alone. So. An automobile crashes. using infinitesimal strain calculations would lead to gross error (see problem 17). there is no elongation of material elements da 1 and da 2. � 2k 4k2 + 1 � ≈ sin−1 � 2k 1 � ≈ 2k 2. [E]� ∼ @ C 0 k 0 ⎢ ⎥ = ⎣k 0 0⎦ 0 0 0 ⎡ ⎤ ® ® Since E11 = 0 and E22 = 0. First. Second. Recall that a is the initial configuration of the body of interest. we can calculate changes in unit length (normal strain) as well as changes in angles (shear strain). we can use basic principles. At point C. The angle decrease is 2E12 = 2k = 2 × 10−4 .5ka 2 t e 2 with a pressure profile described by pg = 2 1 - rgx2.9 PRoBLeMS 1. by simply applying the definition of strain (in this case infinitesimal strain) we can obtain the same information. Consider the motion x = a + ka 3 t e 1 . x = a + kta2 e 1 + kt 3a3 e 2. The problem solved above demonstrates two ways of quantifying local deformations around a point in a continuum. ii. and you are describing the motion of the head of a crash test dummy. material or spatial. would work best for the following situations and why? i.

while the Eulerian strain tensor is a function of B -1.000. in terms of F . ˜ ˜ ˜˜ ® ® ® ® Hint: The following two expressions will be useful for solving the above problem. Chapter 1 problems ˜ ˜ 17 and 18 may help. Take the case of simple shear from Table 2. where the gradient is taken with respect ˜ ˜ to material coordinates.1.46 INTRoduCTIoN To CoNTINuuM BIoMeChANICS (b) Obtain the material description of the velocity vector. (a) Knowing that u = x . (b) Use (a) to write the Lagrangian strain tensor. ˜ ˜ (e) Finally. ( S + T )T = S T + T T and ( S T )-1 = T -1 S -1. Given two tensors. (d) Use MatlabÓ to sketch the configuration at time t = 0 of a rectangular prism with edges at the coordinate axes and at a1 = 1. E . in terms of F . In words. and infinitesimal strain tensors. explain why the names of these tensors make sense. S and T . 3. What is it in terms of ˜ ˜ C? ˜ (c) Write the infinitesimal strain tensor. in terms if F . E *. 6. In this problem. Lagrangian. Based on what you know about ˜ ˜ the components of F and F -1. The components of F -1 are F −1 = ij ∂ xj ˜ of C . ˜ ˜ ˜ ˜ ˜ ˜˜ ˜ ˜ ˜ � � ∂ ai .a . show that Ñ u = F- I. X . . show E *= F T X F . how does this ˜ ˜ expression demonstrate that E is symmetric? ˜ (d) Write the Eulerian strain tensor. (c) Obtain the rate of pressure change. a2 = 2 and its resulting deformation given k = 1 ´ 10-4 at time t = 1. (a) I1C = 3 + 2I1E * ˜ ˜ (b) I2C = 3 + 4I1E * + 4I2E * ˜ ˜ ˜ (c) I3C = 1 + 2I1E * + 4I2E * + 8I3E * ˜ ˜ ˜ ˜ Note: The tensor to which the invariant refers has been superscripted. Prove the following relationships between the invariants of C and E *. we will examine various relationships among the deformation and displacement gradients and the Eulerian. ˜ ˜ 5. (b)The Lagrangian strain tensor is a function 4.

10. g small). ∂t Dρ ∂ νi derive the material derivative form of the continuity equation. respectively. e ˜ 2 ® (b) Based on the above result. the rate of deformation and spin tensors. This problem looks at the infinitesimal rotation vector. where W is the infinitesimal rotation tensor. You will need the fact ˜ ˜ 3 γ5 that the Taylor expansion of tan γ = γ + γ3 + 215 + . Given a displacement field ® u.KINeMATICS oF A CoNTINuuM 47 (a) Calculate E *. 1 � (a) Show that ω = − εijk Ωjk �i . ˜ ˜ (b) Calculate E . The infinitesimal rotation vector is defined as � � 1 � � ω = curl u (2. The continuity equation can be derived from a simple mass balance (i.e. rate of accumu∂ρ � lation = rate in – rate out) in the form + ∇ · (ρ ν ) = 0 . (b) Decompose this tensor into symmetric and antisymmetric parts. what is the formula for the curl of ® ® asymm n in terms of (Ñn ) ? ® ® 9. expand the material derivative and write out the dot product and the summation over i. ® (a) Calculate the gradient of u . (Hint: Start by calculating F ). Write +ρ Dt ∂ xi this equation out in its full form..e. given a general vector n . In this problem. Comment on why this is different from E *. = 0 . It can be shown that the relationship between the initial (dV ) and deformed (dV ) vol0 umes is .. The velocity gradient tensor can be decomposed into its symmetric and antisymmetric parts. for |γ | < π . Starting from this equation. i.. ˜ ˜ (c) Show that E * reduces to E for small deformation (i.e.41) 2 Recall that curl (u ) º Ñ ´ u . you will investigate the decomposition of the displacement gradient tensor. (c) What does the symmetric part represent? Note: The antisymmetric part is the infinitesimal rotation tensor.. 2 (d) Calculate the principal strains and principal directions for E given a simple shear de˜ formation.. 7. 8.

. A rod with circular cross-section is twisted such that the following displacement field is developed: u1 = -y x2x3 u2 = y x1x3 u3 = 0. 1+t Δ(dV ) . . with edges parallel to the coordinate axes. For the above velocity field. ∼ (b) For an incompressible material (i. 15. Find the principal strains. For an element at the point (1. a compression in this case) and an initial volume of 2 cm3. 1. show that Eq. ˜ (c) Given a uniform dilatation with l = ½ (i. with k = 1 ´ 10-5.. Consider the displacement field u1 = k1/2a1a3 u2 = k (a2 + 2a1). 2) with sides parallel to the coordinate axes. Calculate the infinitesimal strain and infinitesimal rotation tensors.e.. 1. with y constant.e.48 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ρ0 (a) Based on the principle of conservation of mass. 14. i. is subjected to the following displacement field: u1 = -ka12 u2 = k1/2a2 u3 = 0. what is the density as a function of time? 13. with k = 1 ´ 10-5. A unit cube.e. where A is originally at (0. 0) and B is originally at (1. i. Show that the dilatation is given by.42) implies ρ = detF .. and u3 = ka12. volume remains constant throughout the deformation). what does dilatation dV (a) An incompressible material has constant density. e = Eii = measure? � � x1 � � 12.e. (2.42) Note: det F is commonly referred to simply as J. What is the continuity equation for an incompressible material? Can the above velocity field belong to an incompressible material? (b) Now consider a material where the density is independent of spatial position but not independent of time. what is the volume after the deformation? 11. Consider the velocity field ν = e 1. In words. what is det F ? What is the relationship among the principal values of F in this case? ˜ ˜ � � dV = det F dV0 ∼ (2. (a) Find the increase in length of the diagonal AB. r = r(t). (b) What is the percent change in volume of the cube? (c) Sketch the result. r is independent of space and time. 0). 0.

A unit cube. is subject to the displacement field u1 = k2 (2a1 + a2 ) u2 = k(a2 − a2 ).KINeMATICS oF A CoNTINuuM 49 (a) What is the unit elongation in the c2 direction? In words. what does elongation measure? (b) Find the maximum unit elongation for this element. What is the percent error in strain and angle change when using the infinitesimal strain assumption for each of these values of k? Comment in the differences. respectively? 16. with edges parallel to the coordinate axes. and k = 10 using both basic principles and infinitesimal strain. Recalculate the strain and -2 change in angle for k = 1 ´ 10 . How would you find the direction associated with this elongation? (c) What is the change in angle between the sides initially parallel to the c1 and c3 axes. 2 1 2 (a) Given the material vector d a = da1 e 1 + da2 e 2 originating from the point (1. (Hint: What is the relationship between stretch and strain?) • • • • . k = 1. what is the change in volume if k = 1 ´ 10 ? (d) What is the change in angle between material elements initially parallel to the c2 and c3 axes? -4 17.1. -3 (c) At the point (1. we let k = 1 ´ 10 .0). and u3 = 0 . You do not need to simplify these expressions.0).1. find � � � �� �d x � the ratio � �� . In the last demonstration of this chapter. � � �d a � ® ® ® (b) Find the infinitesimal strain tensor.

.

e.51 CHAPTER 3 Stress Internal forces in real matter are those among molecules. and subatomic particles..1) t n has units of force over area (e. Consider a body as shown in Figure 3. electrostatic). DF is the resultant force acting on the small area � � DA around P. 2. � A plane.. In continuum theory. S. Body forces—These are forces that act throughout the volume of a body (e.1: Pictorial depiction of the stress vector. . due to contact with another body. Surface forces —These forces act on a surface (real or imagined). Consider a por® tion of the original body to be a free body. gravity. with normal unit vector n passes through an arbitrary point..1 STReSS VeCToR (“TRACTIoN”) The surface force at a point on a surface can be described by the stress vector.g. internal forces are considered to be 1.1. FIguRe 3. separating parts of the body (e. 3. N/mm2 = MPa). pressure).g. atoms.. P . t n .g. i. Then. at point P on plane S as � → � � ΔF t n = lim ΔA→0 ΔA → (3. Let us define the stress vector.

2) ∼ where the stress tensor. T32. while the second number denotes the unit normal of the plane on which it is acting. T33. are the normal stresses.2). T . T31. then the stress vector. The off-diagonal ele∼ ments of T . åMA = 0. on the plane is given by Cauchy’s formula � � � tn = T n (3. T .3. When looking at the components of the ∼ stress tensor. T23. Consider the free body diagram of a differential parallelepiped shown in Figure 3. A. We will find the moment of all forces about an axis passing through the center point. T11 is the normal component (normal stress) of e1 ∼ � ® the stress vector t �1 on the plane whose normal is e1. where we take the positive moment defined as counter-clockwise with thumb pointing out of the page. T21 and T31 are the in-plane components (shear e � ® stress) of the stress vector t �1 on the plane whose normal is e 1 (see Figure 3. T11. � � � � Δx1 Δx1 ∑ MA = T21 ( Δx2Δx3 ) 2 + ( T21 + ΔT21 )( Δx2Δx3 ) 2 � � � � Δx2 Δx2 −T12 ( Δx1 Δx3 ) − ( T12 + ΔT12 )( Δx1 Δx3 ) =0 2 2 FIguRe 3. keep in mind what the subscripts mean. ® . e So. the diagonal elements of T . T22.2: Components of the stress vector on a plane defined by unit normal e1.2 � STReSS TeNSoR ANd ITS CoMPoNeNTS � � If n is a unit normal vector to a plane. � � � � � From t � = T e1 = T11 e1 + T21 e2 + T31 e3 .52 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 3. t n . T13. T21. For equilibrium. The first number denotes direction. are the shear stresses.3 PRINCIPLe oF MoMeNT oF MoMeNTuM (PRooF oF STReSS TeNSoR SyMMeTRy ) We will show that the stress tensor. T12. is a linear transformation (second-order tensor) known as the Cauchy ∼ stress tensor. Then. 3. is generally a symmetric tensor by the use of the moment of ∼ momentum equations for a differential element. and parallel to the x3-axis.

T13 = T31 and T23 = T32. ∼ λ 3 − I1 λ 2 + I2 λ − I3 = 0 where the individual scalar invariants. The normal stresses are the principal stresses (eigenvalues). Ii’s. Thus. We will not consider such materials. T12 = T21 (3. 3. obtained from the characteristic equation of T (recall Eq. Also. On these planes.e. it is true in general (see problem 10).3) Similarly.5) I3 = det(T ) = (T11 T22 T33 ) + 2(T12 T23 T13 ) − (T11 T 2 + T22 T 2 + T33 T 2 ) 23 13 12 ∼ . considering the other directions. there are only six independent components of T . ∑ MA = ( T21 − T12 )( Δx1 Δx2Δx3 ) = 0 for equilibrium.STReSS 53 FIguRe 3. The planes having these directions as their normals are the principal planes.3: Balance of the moments of momentum acting on a parallelpiped about point A. T is symmetric. An exception to stress tensor symmetry is for a special class of materials known as polar materials. no shear stresses). Though we have shown the stress tensor is ∼ ∼ symmetric for an equilibrium situation without body forces. (1. Hence.77)). the stress vector is normal to the plane (i.4) I1 = T11 + T22 + T33 � � � � � � �T T � �T T � �T T � � 11 12 � � 11 13 � � 22 23 � I2 = � �+� �+� � � T21 T22 � � T31 T33 � � T32 T33 � (3.. are given by (3. Dropping terms containing small quantities of higher order. there exists at least three mutually perpendicular principal directions (eigenvectors).4 PRINCIPAL STReSSeS For any symmetric tensor.

8) � � proj n t n = n = Tn � |n |2 ® ® Noting that ½n ½= 1.7) Recall from vector analysis that the projection of a vector u onto a vector n . . the magnitude of the normal stress. 3. from Eq. � |ν |2 � � �� � tn · n � � � (3. T2. (1. the normal stress is given by projn t n .4).76)) ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ t1 n1 T1 0 0 n1 T1 ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ (3. e 3 be the principal directions of T . T3 be the principal stresses.2).54 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Principal stresses include the maximum and minimum values of normal stresses among all planes passing through a given point. Consider a ∼ ® ® plane whose unit normal is n = nie i (Figure 3. (3. by � � �� � � � �Tn � = t n · n = n2 T1 + n2 T2 + n2 T3 � (3. is given � � �� u·ν � ® � � by projν u = ν (Figure 3. and let T1.5). The components of the stress vector on the plane.5 ® ® MAXIMuM SheAR STReSS ® Let e 1.6) ⎣ t2 ⎦ = ⎣ 0 T2 0 ⎦ n2 ⎦ = ⎣ n2 T2 ⎦ ⎣ 0 0 T3 n3 t3 n3 T3 or � � t n = n1 T1 e1 + n2 T2 e2 + n3 T3 e3 ® ® ® ® � � � (3.4: Decomposition of the stress vector into normal and shear components on a plane defined ® by unit normal n . projn u. ½T n½ on the plane defined by n is given . e2. are given by (recall Eq.9) 1 2 3 � � ® ® � FIguRe 3. In our case.

n3) such that ½T S½ is stationary.or 0. the magnitude of the total shear stress. √ . ½TS½ .0. It can be shown [2] that the values of minimum shear ® stress.13) On the planes given by Eq. ± √ . (3. ® Then. from Eq.0. ± √ . 0 .1) The maximum shear stress occurs on one of the planes given by � (3. (3. (3.9) into Eq.13). occur on the principal planes. |TS |2 = � � � (T1 − T2 )2 4 1 � 1 � � for n = √ e1 ± √ e2 2 2 1 � 1 � � for n = √ e1 ± √ e3 2 2 1 � 1 � � for n = √ e2 ± √ e3 2 2 (3.11) To obtain the values of (n1.1. or (0.7) and (3.14) (T1 − T3 )2 |TS | = 4 2 |TS |2 = (T2 − T3 )2 4 .10) Eqs. on the plane is (using the Pythagorean Theorem) � ��� � � � � �TS � = � |t n |2 − |Tn|2 (3. n3) for which ½T S½ is a maximum or minimum.0). we find values ® 2 for (n1. √ . n2. 0. (0. ± √ 2 2 2 2 2 2 ® 2 (3.5: Projection of a vector u on to a vector n .11).12) n= � � � � � � � 1 1 1 1 1 1 √ . n2. (3.STReSS 55 ® FIguRe 3.10) yields � ® 2 2 2 |TS |2 = n2 T 1 + n2 T 2 + n2 T 3 − ( n2 T1 + n2 T 2 + n2 T3 ) 2 1 2 3 1 2 3 ® 2 (3.0). the values of ½T S½ are. specifically ½T S½ = 0. n = (1.

it can be inferred that.1: States of stress for an object with its axis coincident with the x1 axis LoAdINg CoNdITIoN Hydrostatic pressure Uniform tension or compression Simple shear Pure bending Pure torsion Plane stress T11 T12 T13 T22 T23 T33 -p 0 0 0 0 0 0 T13 = x2 f (r) 0 -p 0 0 0 0 T22 = T22 (x1. t. on the plane of maximum shear. s. 2 2 2 From this example.15) Further.56 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ½ 1 - T 2½ ½ 1 - T 3½ ½ 2 - T 3½ T T T Thus. x2) . It acts on the plane that bisects the angle between the directions of the maximum and minimum principal stresses. the maximum shear stress corresponding to a stress tensor is given by |TS |max = → min |T max Principal − T Principal| 2 (3. p. 2.1 shows the components of the stress tensor for some idealized loading conditions. or ¾¾¾ . ¾¾¾ . This table allows one to form the stress tensor. and c are constants and r 2 = x22 + x32. T . x2) 0 0 0 0 0 0 -p 0 0 0 0 0 s 0 cx2 0 t 0 T12 = -x3 f (r) T11 = T11 (x1. Table 3. for the general case. The principal planes have zero shear stress. for each of these states of stress. and thus find the principal stresses. The maximum shear stress is equal to one-half the difference between the maximum and minimum principal stresses. 1. In this table. x2) T12 = T12 (x1. maximum and minimum ∼ TABLe 3. 3. the maximum shear stress is the largest of ¾¾¾ .16) Summary. it can be shown that the normal stress is |Tn | = → min |T max Principal + T Principal | 2 (3.

g.STReSS 57 shear stresses. this table does not refer to any particular material model. x3 + Δx3 ) − t� (x1 . x2 .6 eQuATIoNS oF MoTIoN (CoNSeRVATIoN oF LINeAR MoMeNTuM) We wish to determine the differential equations of motion for any continuum in motion. � � 1 � � t (x1 + Δx1 . x2 + Δx2 . As it was for Table 2. x3 ) e2 Δx2 e2 � � � � 1 � � + t� (x1 . . Consider the stress vectors acting on the six faces of a small rectangular element of the continuum as shown in Figure 3. x3 ) − t� (x1 . weight) per unit mass. and a = acceleration at position xi . x2 .. Every particle of the continuum must satisfy Newton’s law of motion. and normal stress on the plane of maximum shear. ® ® ® Let B = Bi ei be the body force (e. x2 .6: Stress vectors acting on the six faces of a rectangular parallelepiped located within a continuum. x2 .17) FIguRe 3. x3 ) + ρ B Δx1 Δx2 Δx3 e3 Δx3 e3 � = (ρ a)Δx1 Δx2 Δx3 � (3. 3.1. x3 ) − t� (x1 . x2 . We need a constitutive equation to determine the strains generated in a given material in response to any of these states of stress. x3 ) e1 e1 Δx1 � � � 1 � � + t� (x1 . Applying Newton’s second law of motion. r = mass density at xi .6.

23) In Section 3.19) div(T ) + ρ B = ρ a ∼ → → (3. T = stress tensor. (3. quasi-static conditions) Eqs. We can show that � � � � t n = force vector per unit area (surface traction). and n = unit vector normal to the ∼ boundary.22) 3. If acceleration vanishes (e.. This tensor is referred to the current configuration. (3.20) which.18) becomes ∼ ® j ∂ Tij � � � ei + ρ Bi ei = ρ ai ei ∂ xj In invariant form. (3. (3. Furthermore. is ∂ Tij + ρ Bi = ρ ai ∂ xj (3.21) Eqs.21) are known as Cauchy’s equations of motion. (3. Eq.21) reduce to the equilibrium equations div(T ) + ρ B = 0 ∼ → ∂ Tij + ρ Bi = 0 ∂ xj (3. Cauchy’s equations of motion refer to the current configuration (spatial *3.20) and (3.23) is called the stress boundary condition.7 BouNdARy CoNdITIoN FoR The STReSS TeNSoR Surface tractions are applied distributive forces on the actual boundary of some body.17) by Dx1 Dx2 Dx3 and takingnthe lim . knowing te = T ej = Tij ei .2. Eq. t� = T n ∼ n � (3.8 ALTeRNATIVe STReSS deFINITIoNS .20) and (3. in component form for a Cartesian coordinate system. ΔA Δxi →0 ΔA � → → ∂t � ∂ x1 ® ® ® � e1 + ∂t � ∂ x2 � e2 + ∂t � ∂ x3 � e3 + ρB = ρa � � (3.g.18) But. xi’s. we introduced the Cauchy stress tensor. (3.58 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ΔF t = Dividing Eq.

also known as the nominal or Lagrangian. The equations of motion can be ~ expressed in terms of ~ as P � � ∂ Pij (3. ∼ Let us define the surface traction on a given surface referred to the body’s original configuration as → t → ≡ Pn 0 ∼ n0 → (3. Thus. Then. and ~ ® n 0 is the unit normal to that surface in the reference configuration.30) + ρ0 Bi = ρ0 αi ∂ aj .15).STReSS 59 description).25) In the current configuration. (2. the two expressions for d f must also be equal.27) becomes (detF ) ∼ ∼ P ∼ 1 → → F T ndA = T ndA ∼ (detF ) ∼ ∼ (3. r. stress tensor. as the area of the surface goes to zero.28) from which we see P is related to T by ∼ ~ P = (detF )TF −T ∼ ∼ ∼∼ (3. the differential force acting on the same surface is df = t → dA n → → (3. such that quantities. Other measures of stress refer back to the body’s original configuration (material description). the resultant force on that surface is equal to the traction times the area of the surface.. Because the force must be ® equal.24) where P is the first Piola–Kirchhoff. and they can be expressed in terms of T . e.27) It can be shown that n0 dA0 = → 1 → F T ndA .26). In the limit.g. are evaluated in the current configuration as well. (3.25) and (3. Equating Eqs. we get ® → t → dA0 = t →dA = Pn0 dA0 = TndA ∼ ∼ n0 n → → → (3. d f = t → d A0 n0 → → (3. Eq.26) The resultant force on a surface does not depend on the description used. and substi� ® � tuting the definitions for t n and t n 0. (3.29) where F is the deformation gradient introduced in Eq.

Solution. there is no shear component. Consider a rectangular block inside a body (see Figure 3. not. t� = T n ⇒ t� = −pI n = −pn . It can be more convenient to express some constitutive equations using the second Piola–Kirchoff stress tensor. (3. 3. This means that the stress vector ∼ ∼ n n is normal to the plane. is formed from the first Piola–Kirchhoff stress tensor by P2 = F −1 P ∼ ∼ The Cauchy stress tensor is related to the first and second Piola–Kirchhoff stress tensors by (3.31) T = J −1 PF T = J−1 F P2 F T ∼ ∼∼ ∼ ∼ ∼ (3. and a the material element’s acceleration (a used here to not confuse the acceleration with the material coordinates). 2. i. whereas the first Piola–Kirchoff stress tensor ~ is.. where p is a scalar. the second Piola–Kirchoff stress tensor. Finally. .32) where we have used the common notation J = det F (Chapter 2.7).7: Rectangular block with sides parallel to the coordinate axes.9 deMoNSTRATIoNS 1.e. The second Piola– ~ Kirchoff stress tensor is symmetric if T is symmetric. another closely related stress tensor. The state of stress at a certain point is T = -pI .2).60 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where aj’s are the material coordinates of the points in the body. From Eq. � � � � � FIguRe 3. Show that there is no ~ ~ shear on any plane containing this point. problem 10). in general. This state of stress is called hydrostatic pressure. r0 the original density.

8: Pressure distribution on faces of rectangular block. 0. ® n = (1. the total force is F x1 =0 = t�dA = n � � � � � � � � 1 � � p − ρ gx2 dA e1 = p dA − ρ g x2 dA e1 .0. where dA = cdx2. 0. p. 0.0.2). 0) ® t n = (0. p - rgx2) ® t n = (0. n � �� � � � � � � � � p − ρ gx2 dA e1 = p dA − ρ g x2 dA e1 � .0).0) ® t n = (-p + rgx2. Fx1 =0 = pbc − ρ gb2 c . ® Pictorally. -1).0.0) ® t n = (0.8)? (b) Find the total resultant forces acting on faces x2 = 0 and x1 = 0. ⎢ . 0) ® t n = (0. ® n = (0. -p + rgx2) ® ® ® ® ® ® ® FIguRe 3. Solution. t n = (p - rgx2. r. 1. -p + rgb.STReSS 61 The distribution of stress inside the body is given by ⎤ ⎡ 0 0 −p + ρ gx2 � � ⎥ or Tij = (−p + ρ gx )δij with constant g. ® n = (0. 0. 2 � � � � � � t�dA = p dA e2 = p (ac) e2 . 0. and p. -1. T =⎣ 2 0 0 −p + ρ gx2 ⎦ ∼ 0 0 −p + ρ gx2 (a) What is the distribution of the stress vector on the six faces of the block (see Figure 3. (3. (b) On x2 = 0. 0). ® n = (0.0). ® n = (0. 0). t n = T n ∼ � � x1 = 0 x1 = a x2 = 0 x2 = b x3 = 0 x3 = c n = (-1. (a) On � From Eq. Thus. 1). the total force is Fx2 =0 = � � � � �� On x1 = 0.

The re� � � 1 2 2 (T11 + T22 ) ± (T11 − T22 ) + 4T 12 ≡ T1. and letting A = � � a2 A 2 + 1 = 1 ⇒ 2 a1 = ± � a2 = ± � A 1 T12 λ − T11 . est of 2 2 2 2 .70).1). and T12 are functions of x1. the principal values and directions can vary from point to point.2 . T22.69) and (1. ¾ . x2 ) 0 ⎦. one principal value. x2 ) T22 (x1 . = 1 + 2 A 2a = 2 T12 A2 + 1 A2 + 1 Note that. ® ® ® ® (T11 − λ ) a1 + T12 a2 = 0 T12 a1 + (T22 − λ ) a2 = 0 −λ a3 = 0 The last of the above equations implies a3 = 0. the principal direction is n = e 3 (because T e 3 = le 3 = 0). (b) Determine the maximum shear. l = 0 is one eigenvalue.e. l = T3 = 0).4) and (3. and T11. λ − T11 T12 � � λ − T22 ⇒ 1. find the principal values and corresponding principal directions. l[l2 - (T11 + T22 )l + (T11 T22 - T 12)] = 0.62 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 3. the maximum shear is the great� |T1 − T2 | 1 T1 ½ T2 ½ 2 (T11 − T22 )2 + 4T12 .5). (a) For plane stress. Thus. x2 ) 0 � � ⎥ ⎢ (a) The stress matrix for plane stress is T = ⎣ T12 (x1 .2 = 2 The eigenvectors corresponding to these eigenvalues are given by Eqs. T22. maining eigenvalues are λ1. and T12 are functions of only x1 and x2 (see Table 3. ½ ¾ and ½ ¾ = ¾ . and x2. ⎤ ⎡ Solution.. Using Eqs. So. (b) Because the third eigenvalue is zero (i. x2 ) T12 (x1 . For l = 0. (1. λ − T22 T12 a2 = a2 . T11 (x1 . ∼ 0 0 0 2 (3. since T11. a1 = With a12 + a22 = 1. A state of plane stress is one for which T13 = T23 = T33 = 0. ~ For l = T1 or l = T2.

22). ∂ xj ∂ xj ∂ T1 j ∂ T11 ∂ T12 ∂ T13 = + + = 2ν x1 − 2ν x1 + 0 = 0 ∂ xj ∂ x1 ∂ x2 ∂ x3 ∂ T2 j ∂ T21 ∂ T22 ∂ T23 = + + = −2ν x2 + 2ν x2 + 0 = 0 ∂ xj ∂ x1 ∂ x2 ∂ x3 ∂ T3 j ∂ T31 ∂ T32 ∂ T33 = + + = 0+0+0 =0 ∂ xj ∂ x1 ∂ x2 ∂ x3 So. (3. ~ ~ (c) Show that. (b) Determine the magnitudes of the normal and shear stresses on this plane. ⎤ 2 2 2 −2ν x1 x2 0 x2 + ν (x1 − x2 ) ⎥ ⎢ 2 2 2 T=⎣ −2ν x1 x2 x1 + ν (x2 − x1 ) 0 ⎦ ∼ 2 2 0 0 ν (x1 + x2 ) ⎡ ∂ Tij ® ∂ Tij + ρ Bi = 0 . 3..1. and ½T n½ is the magnitude of the normal component of ® the stress vector (i.0.0). in the general case. From Eq. With B = 0 Þ = 0 . (b) Find a relationship between the principal values (eigenvalues) of T and T 0. traction vector) ® (a) Given a general stress tensor.0). . and (0. is defined by T 0 = T − Tii ∼ . does the stress distribution given by satisfy the equations of equilibrium? Solution. show that ½T n½= Tijninj . Now. ⎣ ⎦ ∼ (a) PRoBLeMS 4 0 −6 Find the stress vector at this point acting on the plane that intersects the coordinate axes at (1.10 1 1. the principal directions of the stress and the deviatoric ⎡ ⎤ stress coincide.0.e. T . this stress distribution satisfies the equations of equilibrium. 3. The state of stress at a point is given by T = ⎢ −6 8 0 ⎥ MPa. yes. ~ ~ ~ ∼ 3 I (a) Show that the first invariant of the deviatoric stress vanishes. where ni are the compo~ ® ® nents of a normal vector.e. the component of the stress vector in the direction of n ).1). the deviatoric stress..STReSS 63 4. T 0. 4 −6 4 � � 2. In the absence of body forces. n. (0. Regarding the stress vector (i. For any stress state T.

The stress tensor at a point is given by T = ⎣ 0 5 12 ⎦ MPa.2 MPa. ∼ 0 0 T33 ⎡ ⎤ 4 0 0 � � ⎢ ⎥ 5. The stress tensor at a given point is given by T = ⎢ 0 3 0 ⎥ MPa. The principal values of a stress tensor are T1 = 6 MPa. and unit normal vector.5547 e3 � � � � � n3 = ± 0. solve for ½T n½ given the following tensor. acting on this same plane. T2 = 0... solve again for ½T n½ to verify your answer from part (b). ® (d) Using your answer from part (c). (c) What is the maximum value of shear stress at this point? ® (d) Find the plane corresponding to the maximum shear stress (i. n ). 2 1 2 � (e) What is the normal stress vector on the plane given by the unit normal n = � + � + � e1 e2 e3 3 3 3 2� 1� 2� � n = e1 + e2 + e3 ? 3 3 3 ⎡ ⎤ 5 0 −6 � � 6. find its unit ® normal. ∼ 0 12 −5 ® (b) Using the expression derived in part (a).5547 e2 − 0. 4. Find ⎤ ⎡ T11 2 0 � � ⎥ ⎢ the values of T11 and T33 if a matrix of the stress is given by T = ⎣ 2 1 0 ⎦ MPa. and T3 = 10 MPa.83205 e2 + 0.e.64 INTRoduCTIoN To CoNTINuuM BIoMeChANICS (c) Find the stress vector.83205 e3 l3 = -13 . ⎣ ⎦ ∼ −6 0 −1 (a) What is the maximum value of normal stress among all planes passing through this point? (b) What is the maximum value of shear stress among all planes passing through this point? The eigenvalues and corresponding eigenvectors of this matrix are ® ® l1 = 4 n1 = ± e1 � � � � � l2 = 13 n2 = ± 0. n ). find its unit normal. n . ® T. ~ ⎡ ⎤ ⎡ ⎤ 2 7 0 2 � � 1⎢ ⎥ ⎢ ⎥ � T = ⎣ 7 4 6 ⎦MPa n = ⎣2⎦ ∼ 3 1 0 6 −1 ® ® (a) What is the maximum value of normal stress at this point? (b) Find the plane corresponding to the maximum normal stress (i. t n .e.

98) and the divergence theorem (which converts the surface integral into a volume integral) we get D Dt ��� R ρ εijk xj νk dV = ��� R ρεijk xj Bk dV + ��� R εijk ∂ (xj Tpk ) dV ∂ xp . ∼ 0 0 x3 D Dt ��� R ρ x × v dV = � � ��� R ρ x × B dV + � � �� S x × t� dA n � � Using Eq. and all other Tij = 0.1... � � � (c) Calculate the normal stress on the plane of maximum shear by |T n|2 = | t �|2 − |T S |2 n � � min and show this equals 1 T max . (a) Find the principal values and principal directions for simple shear. find its unit normal. n ). Find the maximum shear stress and the plane on which it acts. -2. Newton’s second law can be stated as the sum of forces equals the time rate of change in linear momentum. n ).e.STReSS 65 (a) What is the stress at the point (-3. then T12 = T21 = t. the sum of moments equals the time rate of change in angular momentum. 10. (b) Find the max shear stress and the plane on which it acts. find the principal stresses and the principal directions of these stresses. Principal + T Principal 2 9. find its unit ® normal. 1)? (b) What is the max value of normal stress among all planes passing through this point? (c) What is the max value of shear stress among all planes passing through this point? (d) What is the normal stress on the plane of maximum shear stress? ® (e) Find the plane corresponding to the max normal stress (i. Similarly. Simple shear is the state of stress in which the only nonvanishing components are a single pair of shear stresses. (1. If shear is occurring in the x1 direction on planes of constant x2. Conservation of angular momentum in a body leads to the equation (c) What is the normal stress on the plane of maximum shear stress? (d) Find the plane corresponding to the maximum normal stress (i. A stress field is given by T = ⎣ x1 x2 0 ⎦ MPa. (e) What is the shear stress on the plane from part (d)? ⎡ ⎤ −3x2 x1 0 � � ⎢ ⎥ 7. (f ) What is the shear stress on the plane from part (e)? 8.e. For the stress state of uniform compression or tension given in Table 3. In this problem we will more formally investigate the fact that the stress tensor is symmetric.

66

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

This result must be true for all regions in the body, so the integrals can be dropped. Hence, we finally have � � � ∂ (xj Tpk ) . D� ρ εijk xj νk = εijk ρ xj Bk + Dt ∂ xp Starting from here, (a) Expand the derivatives to show this reduces to � � � � ∂ Tpk εijk Tjk + xj + ρ Bk − ρ ak − ρ νj νk = 0 ∂ xp

(b) Use Cauchy’s equation of motion and calculate eijk vj vk to show that eijk Tjk = 0. (c) Finally, show that this implies Tij = Tji. 11. Show that Eq. (3.30) implies Eq. (3.21). Eq. (3.32) and Chapter 2, problem 10 will help. 12. Write the equations of motion if the stress compositions have the form Tij = (-p + rgx2 )dij where p = p (x1, x2, x3, t ). Do NOT assume acceleration or body forces equal zero. Give your answer in both invariant and component forms. 13. Given the stress distribution for T below, find T12(x1, x2) in order for the stress ˜ distribution to be in equilibrium given that the stress vector on the plane x1 = 1 is � � → → 2 → t x1 =1 = x2 e1 + 5 − x2 e2 . Assume no body forces.

⎤ ⎡ 2x1 x2 − x2 T12 (x1 , x2 ) 0 � � ⎥ ⎢ T = ⎣ T12 (x1 , x2 ) x2 − x2 0⎦ ∼ 1 0 0 x2

**14. A body of density r is subject to body forces described by B = -g e 2 and a stress distribution as follows:
**

®

®

T11 = (x1 + x2 ) T22 = T12 T33 x2 x1 x3 = T12 x1

T12 = x2 (T11 − x2 ) T13 = T23 = 0

Find the magnitude of the acceleration field within the body as a function of x1 and x2. ® Write the vector, a , at the point (1,1,0). • • • •

67

CHAPTER 4

elasticity

4.1 SuMMARy uP To Now

So far, we have studied the fundamentals of continuum mechanics, including kinematics of deformation, strain, and stress. In addition, we have developed three important principles of continuum mechanics: • Continuity equation (Conservation of mass)

ρ div ν +

•

�

Dρ =0 Dt

Symmetry of stress tensor (Balance of moment of momentum) T = T T ˜ ˜

•

**Equations of motion (Conservation of linear momentum)
**

® div(T ) + rB = ra ˜ ®

We will now begin to investigate how to use these equations to solve problems in continuum mechanics. To do so, we will need to relate the stress in the body to some measure of deformation, which will be accomplished through the introduction of constitutive equations. Thus far, we have studied concepts that apply to materials in general. Now, we will examine specific principles governing the behavior of different types of materials. We will begin with the most basic of models describing the linear elastic behavior of a real solid under infinitesimal deformations. Such material is described by a linear stress–strain relationship, called the constitutive equation for a linear elastic solid. However, before we get to linear elasticity of infinitesimal deformations, we will look at general elasticity, focusing on hyperelasticity. We will then apply simplifying assumptions to the general theory to arrive at the linear, infinitesimal theory. We do this to highlight the assumptions that are inherent in linear elasticity. Furthermore, the constitutive equations we will study in the remaining chapters are also derived from more general theories. The more general theories are, by necessity, more mathematically complicated and beyond the scope of this introductory text.

68

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

*4.2

geNeRAL eLASTICITy

An elastic material is a material for which the current stress in the body is solely determined by the current configuration of the body; the stress does not depend on how the current configuration came to be, such as the rate at which the deformation took place or other aspects of the deformation history. These latter considerations are associated with fluids and viscoelastic materials, as will be seen in subsequent chapters. Mathematically, the constitutive equation for general elastic materials relates the stress to the deformation gradient, Eq. (2.15). This can be expressed functionally as (4.1) T = T (F ) ˜ ˜ ˜ where T is the Cauchy stress tensor. Based on the principle of material objectivity (or material ˜ frame indifference), which says that the stress in the body should be the same regardless of the reference frame from which it is measured, it can be shown that Eq. (4.1) must depend on F through ˜ the left Cauchy–Green deformation tensor, B , or, equivalently, X , the Eulerian strain tensor. Recall ˜ ˜ Eqs. (2.17) and (2.18) which define B and X , respectively. Thus, Eq. (4.1) becomes ˜ ˜ T = T (B) (4.2) ∼ ∼ ∼ We will restrict our discussion to isotropic materials. In an isotropic material, the rotation of a particle has no influence on the stress tensor. With this in mind, the representation theorem [3] gives the most general functional representation of T as ˜ T (B) = α 0 ∼ + α1 B + α2 B 2 I (4.3) ∼ ∼ ∼ ∼ where the ai’s are real functions of the scalar invariants of B ˜ αi = αi (I1 , I2 , I3 )

(4.4)

Recall Eqs. (1.78), (1.79), and (1.80), or Eqs. (1.81), which define the scalar invariants of a secondorder tensor. Explicitly combining Eqs. (4.3) and (4.4) yields

T(B) = α0 (I1 , I2 , I3 ) ∼ + α1 (I1 , I2 , I3 ) B + α2 (I1 , I2 , I3 ) B 2 I ∼ ∼ ∼ ∼

(4.5)

In terms of the second Piola–Kirchoff stress tensor, Eq. (3.31), isotropy and material objectivity yield P 2 as the following function of C , the right Cauchy–Green deformation tensor, Eq. (2.16), ˜ ˜

P2 (C) = β0 (I1 , I2 , I3 ) ∼ + β1 (I1 , I2 , I3 ) C + β 2 (I1 , I2 , I3 ) C 2 I ∼ ∼ ∼ ∼

(4.6)

where I1, I2, I3 are now the scalar invariants of C . In passing, we note that the scalar invariants of B ˜ ˜ and C are identical, but the functions a and b are, in general, different. Eq. (4.5) or Eq. (4.6) are ˜ the constitutive equations of general isotropic elasticity. What remains is to determine the coefficient functions, a or b. We will now turn to a specific subset of elastic materials commonly employed in biomechanics, so-called hyperelastic materials, which will help us to determine the coefficient functions.

eLASTICITy

69

4.2.1 hyperelasticity

The basic concept of hyperelasticity is that the material behaves elastically (i.e., recovers original shape) even at large deformations. It does this by storing the energy used to deform it as strain energy, which is released upon release of the applied load. This is in contrast to plasticity at large or small deformations, in which the original shape is not recovered. Hyperelastic materials are defined by a Helmholtz free-energy function, y, which describes how the strain energy is stored. Hyperelastic materials are a specific subset of elastic materials, but not all elastic materials are hyperelastic. As mentioned, the stress–strain relationship for a hyperelastic material is based on a strain energy function. Based on the principle of material objectivity, the strain energy function is expressed as a function of C . It can be shown that ˜ ∂ψ T T = 2 J −1 F F (4.7) ∼ ∼ ∂C ∼ ∼ where y is the energy stored ( per unit initial volume) Recall that J = det F . Now, we can ensure that the constitutive equation defined by Eq. (4.3) de˜ scribes an isotropic material if y is a function of the invariants of C , i.e., ˜

ψ = ψ (I1 , I2 , I3 )

Using the chain rule,

(4.8)

∂ψ ∂ ψ ∂ I1 ∂ ψ ∂ I2 ∂ ψ ∂ I3 = + + ∂C ∂ I1 ∂ C ∂ I2 ∂ C ∂ I3 ∂ C ∼ ∼ ∼ ∼

Furthermore [4],

(4.9)

∂ I1 =∼ I ∂C ∼

∂ I2 = I1 ∼ − CT = I1 ∼ − C I ∼ I ∼ ∂C ∼

(4.10)

∂ I3 = (C 2 )T − I1 C T + I2 ∼ = I3 (C −1 )T = I3 C −1 I ∼ ∼ ∼ ∼ ∂C ∼

so that

Eq. (4.7) becomes

� ∂ψ ∂ψ ∂ψ ∂ψ � I1 ∼ − C + = I+ I ∼ I3 C −1 ∂C ∂ I1 ∼ ∂ I2 ∂ I3 ∼ ∼ T = 2 J −1 F ∼ ∼ � � � ∂ψ ∂ψ ∂ψ � I1 ∼ − C + I+ I ∼ I3 C −1 FT ∼ ∂ I1 ∼ ∂ I2 ∂ I3 ∼

(4.11)

(4.12)

(4.5). a2) to one (y ). we wish to calculate the expression relating the Cauchy stress to the left Cauchy–Green strain tensor. Tendons can undergo finite deformations during the extension of joints. I3 = 1 (see Chapter 2 problem 10).12) takes the form ˜ � � ∂ψ � ∂ψ � −1 ∂ ψ 2 I1 B − B + T = 2J B+ I3 I ∼ ∼ ∼ ∂ I1 ∼ ∂ I2 ∂ I3 ∼ � � � � � � ∂ψ −1 ∂ ψ −1 ∂ ψ −1 ∂ ψ = 2J I3 ∼ + 2 J + I1 B − 2 J I B2 ∼ ∂ I3 ∂ I1 ∂ I2 ∂ I2 ∼ Comparing Eq. (4. we have reduced the number of unknown functions from three (a0. Furthermore. Table 4. it is not sufficient to simply set I3 = 1 and ¶y = 0. and then we wish to derive an expression for the stress in the x3 direction for the state of uniaxial tension.14) Though we do not yet have expressions for the coefficients. I2 . I2 .13) to Eq. If the ¶I3 material is incompressible.70 INTRoduCTIoN To CoNTINuuM BIoMeChANICS In terms of B . the stress field can only be determined up to an arbitrary hydrostatic pressure.1 gives some expressions for ˜ free-energy functions that have been used to describe various biological tissues. a1.13).15) Recall that J = det F = 1 for incompressible materials. . becomes � ∂ψ ∂ψ ∂ψ 2 T = −pI + 2 + I1 B − 2 B ∼ ∼ ∼ ∂ I1 ∂ I2 ∂ I2 ∼ � (4. we see T = T (B ) and identify ˜ ˜ ˜ ∂ψ α0 (I1 . there has been ample research into different forms of y that one may choose to describe a tissue. (4. I3 ) = −2 J −1 ∂ I2 (4. p. I2 ) = α e β (I1 −3) − αβ (I2 − 3) 2 (4. I2 . Eq.13) (4. Eq. I3 ) = 2 J −1 I3 ∂ I3 � � ∂ψ −1 ∂ ψ α1 (I1 . Demonstration. Consider the following hyperelastic energy function for an incompressible material [7] ψ (I1 . and y cannot depend on I3. (4. However. For incompressible materials.16) First. For a purely incompressible material. for reasons beyond the scope of this book. I3 ) = 2 J + I1 ∂ I1 ∂ I2 ∂ψ α2 (I1 .

Thus.15) yields For uniaxial tension or unconfined compression of an isotropic incompressible tissue.17) F λ3 0⎦ 0 ∼ = ⎣ � � T = −pI + αβ 2eβ (I1 −3) − I1 B + αβ B 2 ∼ ∼ ∼ ∼ 0 0 λ3 . (4. μ is the Lamé constant of linear elasticity (discussed below in Section 4. F has the fol˜ lowing form [8] ⎤ ⎡ −1/2 λ3 0 0 ⎥ ⎢ −1/2 (4. The expression relating the Cauchy stress to the left Cauchy –Green strain tensor for an incompressible tissue is given by Eq.eLASTICITy 71 TABLe 4. (4. I2.5) **n = a + 2b Solution. I3) Compressible neo-Hookean μ λ (I1 − 3) − μ ln J + (ln J )2 2 2 μ (I1 − 3) 2 α (I1 − 3) + β (I2 − 3) γ α (I1 −3)+β (I2 −3) ne I3 � � γ eα (I1 −3) − 1 + β (I2 − 3) + g (I3 ) Incompressible neo-Hookean* Mooney–Rivlin Model of Mow and Holmes**[5] Model of Veronda and Westmann[6] *In this model. we need to calculate ∂ ψ and ∂ ψ . y = y (I1.1: Some hyperelastic models used in biomechanics ModeL NAMe FRee eNeRgy eXPReSSIoN.15). ∂ I1 ∂ I2 ∂ψ = αβ e β (I1 −3) ∂ I1 ∂ψ αβ =− ∂ I2 2 Substituting these expressions into Eq.

and the infinitesimal strain tensor. By direct calculation. the finite Eulerian strain tensor. (2.23). but geometrically nonlinear situation. respectively).2. Using the result from part (b) of problem 3 in Chapter 2.6) becomes � � P2 (E ∗ ) = β0 ∼ + β1 (I + 2E ∗ ) + β2 (I + 4E ∗ + 4(E ∗ )2 I ∼ ∼ ∼ ∼ ∼ ∼ ∼ = (β0 + β1 + β2 )I + 2(β1 + 2β2 )E∗ + 4β2 (E ∗ )2 ∼ ∼ ∼ (4. B and B 2 are ˜ ˜ ⎤ ⎤ ⎡ −1 ⎡ −2 λ3 0 0 λ3 0 0 (4. X. T » P2) . 4. Through the relationship Eq. E *.20) . geometrical and physical. (2. This is most easily accomplished by considering Eq.5) ˜ ˜ ˜ can be written as � � (4. Physical nonlinearity refers to whether the constitutive equation relating stress to strain is of linear form. and ˜ ˜ (2.19) T E = γ0 ∼ + γ1 E + γ2 E 2 I ∼ ∼ ∼ ∼ Though geometrically linear. (4. E (see Eqs. Eq. We have thus far presented three strain tensors. (2. one can ap 3 ˜ preciate the nonlinear nature of hyperelastic constitutive equations. this equation is physically nonlinear in E and in the coefficients (they ˜ still depend on I2 and I3). (4. Geometrical nonlinearity refers to whether or not the strains generated in the body are infinitesimal or finite.18). in the x3-direction. Under the assumption of infinitesimal deformations. let us examine a physically linear. (4. Eq.2 Approximations Leading to Linear elasticity Elasticity has two inherent forms of nonlinearity. both the Eulerian and Lagrangian strain tensors reduce to the same infinitesimal strain tensor.6) and developing it to first order in terms of C and its ˜ invariants.72 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where l3 is the principal stretch ratio in the x3-direction (direction coincident with load application). and the infinitesimal assumption. -1 2 T33 = -p + ab[2eb(2l3 +l3 - 3) - (2l-1 + l2 )]l2 + (ab)l4 3 3 3 3 where we have made the substitution I1 = tr(B ) = 2l-1 + l23. finite ˜ Lagrangian strain tensor.18) ⎥ ⎥ ⎢ ⎢ − − B = ⎣ 0 λ3 1 0 ⎦ and B 2 = ⎣ 0 λ 3 2 0 ⎦ ∼ ∼ 2 4 0 0 λ3 0 0 λ3 Thus.18). In other words. Examining this expression.26). Now. X » E * » ˜ ˜ E (also.

.3 eXPeRIMeNTAL oBSeRVATIoNS oF INFINITeSIMAL LINeAR eLASTICITy Consider a slender cylinder of cross-sectional area A under static tension. (4. ˜ (4. i. Applying geometric linearity to the physically linear Eq. (E *)2. developing the invariants of C to first order in the invariants of E * ˜ ˜ C C C yields I 1 » 3 + 2I1E *. We also ˜ neglect the higher-order term. Eq.21) where l and m are Lamé constants formed from the b coefficients. Thus. (4. one can conceive of a situation which is simultaneously linear in both the geometric and physical senses.20) reduces to [9] P2 (E ∗ ) = λ tr(E ∗ )I + 2μ E ∗ ∼ ∼ ∼ ∼ ∼ (4. Finally. and I 3 » 1 + 2I1E *.22) ∼ ∼ ∼ ∼ ∼ We now turn to an in depth discussion of this very equation. P.eLASTICITy 73 Per problem 5 of Chapter 2. versus extension.21). Figure 4. E * » E and ˜ ˜ P2 » T . we get ˜ ˜ T (E) = λ tr (E)I + 2μ E (4. Dl. . 4. Assuming an unstressed reference state for the tissue. I 2 » 3 + 4I1E *.1: Load–deflection curve for a slender cylinder subjected to a uniaxial tension test.21) is a linear relationship between the stress and the state of deformation. the b coefficients become functions of only the first invariant of E *.1 shows an idealized plot of applied load.e. where the tensor to which the invariant refers ˜ ˜ ˜ ˜ ˜ ˜ is superscripted. but remains geometrically nonlinear due to the presence of large strain tensors. created during tension. FIguRe 4. Eq.

Dl) plot can be replotted as (P/A. In addition to the stress–train test described above. and the specimen exhibits elasticity.2). but with an increased proportional limit (material has work-hardened or been cold-worked). other characteristic tests can be performed as described below: (a) To measure the change in volume of a homogeneous. the material is inhomogeneous (versus homogeneous). isotropic material under uniform pressure p. The (P. called the Poisson’s ratio.74 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Segment OA is the linear portion (proportional range). and the lateral strain is ed. for which the state of stress is Tij = -pdij . Reapplication of the load from C results in elastic behavior with the same slope as OA.e.24) If a specimen is cut at different orientations at a sufficiently small neighborhood and still exhibits the same stress–train behavior.” K. the material is anisotropic. recall the change in volume is the dilatation. line OA is retraced. and define the “bulk modulus. ν=− εd εa (4. . The axial strain is ea. e = Eii. e) to obtain “material” (as opposed to “structural”) behavior that is independent of geometry (see Figure 4.. plasticity. Dl/l ) or (s. if the load is removed. If properties change at different neighborhoods. we traverse OABC and find “permanent deformation” OC. i. Past point A. as FIguRe 4.23) εA The ratio –ed/ea is another material coefficient. The slope of OA is a material coefficient called Young’s modulus (or modulus of elasticity). If properties change with direction. σ EY = (4.2: Stress–strain and axial versus lateral strain curves for a slender cylinder subjected to uniaxial tension. the material is called isotropic.

m) using three different experiments.28) reduce to six.27) T11 = C1111 E11 + C1112 E12 + . k.28) where Cijkl are the components of a fourth-order tensor (“elasticity tensor”) that represents the mechanical properties of an anisotropic Hookean elastic solid. + C1233 E33 . for a linear anisotropic elastic solid. Finally.. define the “shear modulus. T33 = C3311 E11 + C3312 E12 + . these equations can be written as Tij = Cijkl Ekl (4..... as Mt l μ= (4.25) (b) To measure the torsional response of a cylindrical bar of circular cross-section of radius R and length l... + C1133 E33 T12 = C1211 E11 + C1212 E12 + . (4.. + C3333 E33 Alternatively. we can write the basic relationship T = T (E) ∼ ∼ ∼ where T is a single-valued function of E with T (0) = 0. Thus. subjected to torsional moment Mt with resulting twist angle q. 2 Thus... n.eLASTICITy 75 p K = −e (4. .. From (a). solid.” m. But. deformations disappear completely (d) deformations are very small 4.4 LINeARLy eLASTIC SoLId Let us now formulate the constitutive equations of the linearly elastic.. (c).26) Ip θ 4 πR where Ip = is the polar moment of inertia. we have introduced four material properties (E Y . Based upon features (b). the three experiments described share the following features: (a) relation between applied load and a quantity measuring deformation is linear (b) rate of load application does not affect linear relationship (c) upon removal of loading. for any isotropic linear elastic material. and (d). or Hookean elastic. Because T and E are symmetric. the ˜ ˜ nine equations of Eq. we need no more than 36 material constants.. are these constants independent? We shall see shortly.. we can write ˜ ˜ ˜ (4.

5 ISoTRoPIC LINeARLy eLASTIC SoLId If the material is isotropic.31) T11 T22 T33 T12 T13 T23 = λ Ekk + 2μ E11 = λ Ekk + 2μ E22 = λ Ekk + 2μ E33 = 2μ E12 = 2μ E13 = 2μ E23 (4. is also isotropic.30) becomes (4. (4.76 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 4. and Hijkl = dildjk . Eq..30) T = λ eI + 2μ E ∼ ∼ ∼ Written out in long form. then Cijkl = Cijkl (i. (4. as dij are the same ˜ for any Cartesian basis. (4.28) yields Tij = Cijkl Ekl = λ eδij + (α + β )Eij If we let a + b = 2m.32) . (4. which means the tensor is isotropic. in invariant form.. components of elasticity tensor must not change regardless of how the bases are changed). I.e. Bijkl = dikdjl . One can also define three other isotropic fourth order tensors: Aijkl = dijdkl . i. we write Cijkl as Cijkl = λ Aijkl + α Bijkl + β Hijkl Moreover. then Tij = λ eδij + 2μ Eij or. (4. Note that identity tensor.e.29) Aijkl Ekl = δij δkl Ekl = δij Ekk = δij e Bijkl Ekl = δik δjl Ekl = Eij Hijkl Ekl = δil δjk Ekl = Eji = Eij Substitution of Eq.29) into Eq. Using a theorem (beyond the scope of this book) that states that any fourth-order isotropic tensor can be represented as a linear combination of the above three tensors. material properties are independent of direction.

only one normal stress component is nonzero. and (4.33) e = Eii = 1 Tkk 3λ + 2 μ (4.30). Eq. Eq. l and m. (4. a linear isotropic elastic solid has only two independent material constants.eLASTICITy 77 Eqs.33) becomes E11 E22 � � 1 λ λ +μ = T11 = T11 T11 − 2μ 3λ + 2 μ μ (3λ + 2μ ) λ λ T11 = − E11 = E33 = − 2 μ (3λ + 2μ ) 2(λ + μ ) E12 = E13 = E23 = 0 In the tensile test from Section 4.6 MATeRIAL PRoPeRTIeS oF eLASTIC MATeRIALS We will now express the material properties of a linear isotropic elastic solid (as identified in Section 4. we defined the Young’s modulus as EY = σ μ (3λ + 2μ ) T11 = = εa E11 λ +μ (4.30) gives Eij = Also. (4. 4. Assume T11 = s ¹ 0. from Eq.3) using the constitutive equation. The constants.34) Now.36) . Thus. (a) Uniaxial tension (or compression). the Poisson’s ratio was defined as ν =− εd E33 λ =− = εa E11 2(λ + μ ) (4.32) are different forms of the constitutive equation for a linear isotropic elastic solid. let us consider the tests we used previously to determine the properties of a linear elastic solid. � � 1 λ Tkk δij Tij − 2μ 3λ + 2 μ (4. (4.33). are called the Lamé constants (material properties).30).3.35) Similarly. For this state of stress. Inverting Eq. (4. (4. Then.31). (4.

38) (b) In simple shear.35) and (4.36).40) Thus. (4.42) . (4. Note that while there are three material properties.41) e= 2 μ + 3λ So. Plugging this into Eq.35) and (4. (4.78 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Substituting Eqs. from Eqs. from our definition for the bulk modulus. (c) The state of hydrostatic pressure is given by T = -pI. Thus. that μ= EY 2(1 + ν ) (4. Eq. only one pair of off-diagonal elements is nonzero.33) simplifies to E12 = E21 = from which we identify T12 2μ (4.36) into Eq. This is because we can show. Let T12 = T21 ¹ 0. (4.37) Eq.39) μ= T12 2E12 (4. the Lamé constant m is typically called the shear modulus (as identified in our torsion test). (4. p 2 2 μ + 3λ K=− = =λ + μ e 3 3 (4. This is the usual form used to describe an isotropic linear elastic solid. (4. only two are independent.33) yields E11 = E22 = E33 = E12 = E13 = E23 = 1 [T11 − ν (T22 + T33 )] EY 1 [T22 − ν (T11 + T33 )] EY 1 [T33 − ν (T11 + T22 )] EY 1 T12 2μ 1 T13 2μ 1 T23 2μ (4.37) is also referred to as the generalized Hooke’s law.34) gives us ˜ ˜ −3p (4.

We will consider only cases of small motions. In such a case. µ) (E Y . n) or (l. where every particle is always in a small neighborhood of the natural state. m). Thus. n) l µ K EY l µ 2 λ+ μ 3 ν EY (1 + ν )(1 − 2ν ) EY 2(1 + ν ) EY 3(1 − 2ν ) μ (3λ + 2μ ) λ+μ λ 2(λ + μ ) EY n n In general. we use the set (E Y . then material and spatial coordinates are approximately equal.43) . Table 4. ∂ 2 ui so ai = .eLASTICITy 79 TABLe 4.2 provides material property relationships for an isotropic linearly elastic solid. we can state the ∂t 2 equations of motion for linear isotropic elasticity as ρ ∂ Tij ∂ 2 ui = ρ Bi + 2 ∂t ∂ xj (4.2: Relationships among different material constants used in linear elasticity MATeRIAL CoNSTANT (l . 4. under infinitesimal strain conditions (small deformations).7 eQuATIoNS oF The INFINITeSIMAL TheoRy oF eLASTICITy ρ ai = ρ Bi + ∂ Tij ∂ xj Recall the equations of motion from Chapter 3.

On the other hand. u2. the necessary and sufficient conditions for the existence of single-valued continuous solutions ui(x1. Eqs.47) are called the equations of compatibility (or integrability). it is not guaranteed that a displacement field exists that satisfies Eq. the given Eij are not compatible. u3. ∂ 2 E11 ∂ 2 E22 ∂ 2 E12 + =2 2 2 ∂ x1 ∂ x2 ∂ x2 ∂ x1 ∂ 2 E22 ∂ 2 E33 ∂ 2 E23 + =2 2 2 ∂ x2 ∂ x3 ∂ x3 ∂ x2 (4.44).44) into Eq. x3) are ∂ 2 E33 ∂ 2 E11 ∂ 2 E31 + =2 2 2 ∂ x3 ∂ x1 ∂ x1 ∂ x3 � � ∂ 2 E11 ∂ ∂ E23 ∂ E31 ∂ E12 = − + + ∂ x2 ∂ x3 ∂ x1 ∂ x1 ∂ x2 ∂ x3 � � 2E ∂ 22 ∂ ∂ E31 ∂ E12 ∂ E23 = − + + ∂ x3 ∂ x1 ∂ x2 ∂ x2 ∂ x3 ∂ x1 � � ∂ 2 E33 ∂ ∂ E12 ∂ E23 ∂ E31 = − + + ∂ x1 ∂ x2 ∂ x3 ∂ x3 ∂ x1 ∂ x2 Hence.43) represents three equations (one for each i ) for nine unknowns (three ui and six Tij). we can always determine the six strain components using Eq. (4. x3) are continuous functions.44).46) Tij = λ eδij + 2μ Eij ti = Tij nj Notice that Eq.45) (4. u3 are given. which govern infinitesimal linear elasticity (see problem 4).45) (which relates the T ’s to the u’s). Theorem.44) (4. (4.43) gives three equations in the three unknowns. u1. (4.8 CoMPATIBILITy CoNdITIoNS FoR INFINITeSIMAL STRAIN CoNdITIoNS When any three displacement functions u1.80 INTRoduCTIoN To CoNTINuuM BIoMeChANICS The strain field is the stress field is and the surface tractions are 1 Eij = 2 � ∂ uj ∂ ui + ∂ xj ∂ xi � (4. (4.47) . This process yields Navier’s equations of motion. (4. The remaining six equations come from substituting Eq. (4. 4. This is because we do not necessarily obtain unique displacement components when integrating the strain components. into Eq. If Eij (x1. In this case. if Eij are arbitrarily prescribed. u2. x2. x2. Substituting that result. (4.

˜ ˜ ˜ Thus. Strain components from Eq.49) FIguRe 4. (4. (1.9.32)).3 only for convenience) is under the action of equal and opposite normal traction. tn = T n = n2(T ®2) + n3 (T ®3) = 0 . (4. the traction-free lateral surface condition is satisfied. and the bar is at rest. Body forces are zero. The lateral surface is free of any surface traction. (4. 3.9 4.eLASTICITy 81 4. Boundary condition on the end faces is obviously satisfied.1 CLASSICAL PRoBLeMS IN eLASTICITy Simple Infinitesimal extension of a Linear elastic Solid A cylindrical elastic bar of arbitrary cross-section (drawn as circular in Figure 4. ® e 2.37) are ® E11 = E22 = E33 = E12 = E13 = E23 = σ 1 [T11 − ν (T22 + T33 )] = EY EY 1 σ [T22 − ν (T11 + T33 )] = −ν EY EY 1 σ [T33 − ν (T11 + T22 )] = −ν EY EY 1 T12 = 0 2μ 1 T13 = 0 2μ 1 T23 = 0 2μ (4. the stress components are T11 = σ T22 = T33 = T12 = T13 = T23 = 0 Let us show that this is a possible solution.48) ∂ Tij =0 ∂ xj are identically satisfied. By considering points on the boundary surface. Equations of equilibrium. (4. From Eq. s.46) (recall Eq.3: Longitudinal and axial cross-sections of a cylindrical elastic bar in uniaxial tension. 1. On the lateral surface. . n = 0 ®1 ® ® ® ® e e e + n2®2 + n3e 3. Eq. at its end faces.43).

(4. Assume now that the cross-sectional area of the bar is A.50) Thus. then the effects of the two distributions are the same sufficiently far removed from the region of force application. is Δd = −ν Fd AEY (4. Let L0 be the undeformed length of the bar and Dl equal the elongation.47). Saint Venant’s principle. Eqs.51). Thus.52) Δl = FL0 AEY (4.82 INTRoduCTIoN To CoNTINuuM BIoMeChANICS The conditions of compatibility. we can show ⎤ F/A 0 0 ⎢ ⎥ [T ] = ⎣ 0 0 0 ⎦ ∼ 0 0 0 ⎡ (4. (4.50) and (4. so there are single-valued displacement field components. 0.53) Let d be the undeformed length of a line in the transverse direction. in practice. are automatically satisfied.50) and (4.51). Hence. From Eqs. we have completed the solution of the simple extension or compression problem. the exact value of the distribution of surface traction is not known. σ x1 EY νσ u2 = − x2 EY νσ u3 = − x3 EY u1 = (4. Only the resultant force is known. If a force distribution is replaced by another one that has the same resultant force and moment. from Eqs.54) It is important to note that. when a bar or similar object is pulled. The contraction in the transverse direction. and the maximum shear stress is |Ts|max = F/2A (acting on planes at 45° angle to the normal cross-sections). 0. the maximum normal stress is ® ® ˜ |Tn|max = F/A (acting on the cross-sectional area). . C. The surface traction s on either side gives rise to a resultant force F = σA passing through the centroid. This leads to the question of whether the solution we obtain is acceptable or not. the principal stresses are F/A. (4. (4.51) Because T is diagonal.

the displacements (ui = xi – ai) associated with simple extension can be found.49). we again recover Eqs.. the displacements are ui = (1 – li)xi. combining conservation equations with constitutive equations can reduce the number of unknowns needing to be kept track of. With this in mind. we are left with determining C0.58). By the geometry of the problem. is dui = C0 dxi (4.57) yields C1 = 0. then the first boundary condition on u1 is u1(x1 = 0) = 0.1. (4.55) Referring to Table 2. and u3 = u3(x3). u1 = u1(x1). as each displacement is only a function of its respective coordinate. Navier’s equations reduce to d2 u 1 =0 2 dx1 Each of these equations has the solution d2 u2 =0 2 dx2 and d2 u3 =0 2 dx3 (4. Assuming we fix one end of the bar at x1 = 0. We need two boundary conditions for each of u1. Hence. (4. applying Eqs. Applying each of these boundary conditions to Eq. 2. u2. Navier’s in this case) . Under quasi-static conditions (i.e. u2 = u2(x2).eLASTICITy 83 Let us briefly consider an alternative approach to this problem.1.. you will derive Navier’s equations for infinitesimal elasticity. The nonzero equations are E11 = E22 σ du1 = dx1 EY du2 du3 σ = E33 = = = −ν dx2 dx3 EY (4. The first derivative of Eq. and u3. those equations are (λ + μ ) ∂ 2 ui ∂ 2 uk +μ 2 =0 ∂ xi ∂ xk ∂ xk (4.56) ui (xi ) = C0 xi + C1 (4.58) where we have changed ¶ to d. (4. By symmetry arguments. In each direction.50). ignoring inertial forces).59) Thus.57) where C0 and C1 are integration constants to be solved for based on the boundary conditions. In problem 4 at the end of this chapter. This method of solution points out three important concepts: 1.e.57). which again leads to Eqs. we expect u2(x2 = 0) = u3(x3 = 0) = 0. Hence. the boundary conditions on u2 and u3 will be identical. (4. Referring to Table 3. (4. we see that T11 = s. the resulting equation (i.

Define x1 = 0 and x1 = l as the left and right faces of the bar. . causing bending of the bar. to satisfy equilibrium ® ∂ T11 =0 ∂ x1 Thus. We will see these principles put to use again in later chapters.9. T11 = T11 (x2 . we must specify a state of stress corresponding to (a) a traction-free lateral surface (b) some distribution of normal surface tractions on x1 = 0 and x1 = l equivalent to the following bending couples MR = M2 ®2 + M3 ®3 e e ® ® ML = -MR Note: M1 is absent because it would result in a twisting (torsion) couple. 4. The x1 axis passes through the centroid of the bar. and 3.84 INTRoduCTIoN To CoNTINuuM BIoMeChANICS can then be solved with appropriate boundary and initial conditions. Then. x3 ) (4.4: Longitudinal and axial cross-sections of a beam in pure bending. To determine a linear elasticity solution. A beam is in pure bending. We assume T11 is the only nonzero stress component and is a function of xi . or simple bending. it is often necessary to translate stress boundary conditions into displacement boundary conditions and vice versa.60) FIguRe 4. respectively.4).2 Pure Bending of a Beam A beam is a bar acted upon by forces/moments in an axial plane. if acted upon by end couples only (see Figure 4.

Thus. as the body is in equilibrium. let us consider the boundary tractions. (4. a = 0.64) M3 = −β I33 − γ I23 (4. x2dA = x3dA = 0. tn = T n = T e1 = T 11 e1. i. the reaction force is given by R = òtn dA.61) into Eqs. Together these 2 2 ∂ x2 ∂ x3 ∂ x2 ∂ x3 T11 = α + β x2 + γ x3 (4. ® ® Because R = R1e1 passes through the centroid. (4. I23 are the moments and products of inertia. ® ® ˜ ˜ ˜ Using Eq. R1 = � T11 d A = α � dA + β � x2 d A + γ � x3 d A = α A (4. its effect (of simple extension) can be superposed onto that of bending.62). (4. = 0.e.65) Note: The negative signs result from the right-hand rule. Now. from the statement of the problem. We stated that the lateral surface is traction-free. Recall that an eccentric force (a force not coinciding with the centroid of the cross-section) is tantamount to a bending couple plus a centric force.62) can be written as T11 = β x2 + γ x3 (4. M1 = 0. Eq. and. I33.47).63) Note: The first moment of area about a centroidal axis is zero. and = 0 . x1 = 0: The resultant force system is equal and opposite to that on x1 = l. ò ò M2 = and � x3 T11 d A = α � x3 d A + β � x2 x3 d A + γ � x2 d A = β I23 + γ I22 3 (4. What about the two end surfaces? ® ® ® ® ® ® x1 = l: On this surface.eLASTICITy 85 and E11 = E22 1 1 [T11 − ν (T22 + T33 )] = T11 EY EY ν = E33 = − T11 EY (4. we find that equations imply ∂ 2 T11 ∂ 2 T11 ∂ 2 T11 = 0. let us check if the strains are compatible. Now..61) E12 = E13 = E23 = 0 Because we assumed the state of stress.62) Now. Thus. Similarly calculated.66) . Substituting Eqs. I22. R2 = R3 = 0. (4.

For small q. we find that Eq.68) Integrating Eqs.67) Let us now describe the resultant deformations. (4. (4. such that I23 = 0. we choose axes x2 and x3 to coincide with the principal axes of the cross-sectional area. (4.68). one can find the displacement components [10] Consider displacements for x1 = constant (cross-sectional planes). For simplicity.86 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Without any loss of generality. (4. By setting the axes this way. The displacement of the neutral axis is used to define the deflection of the beam as shown in Figure 4. i. Using the strains from Eq. i. (4.67).66) yields the stress distribution for the cylindrical bar. T11 = M3 M2 x3 − x2 I22 I33 (4. the cross-sectional plane remains plane and is rotated by an angle q.65) becomes β =− Similarly.e. let M3 = 0. (4.e. � � M2 x1 u1 = x3 EY I22 M2 x1 x3 I22 EY ν M2 u2 = − x2 x3 I22 EY � �� M2 � 2 x1 − ν x2 − x2 u3 = − 2 3 2I22 EY u1 = (4.61) and substituting in Eq. with all other Tij = 0. u1 is a linear function of x3. θ ≈ tan θ = u1 M2 x1 = x3 EY I22 .69) (4.64) becomes M3 I33 γ= M2 I22 Substituting these equations into Eq.70) Hence.. Then. Eq.5. along lines of symmetry.. we find E11 = M2 x3 I22 EY E22 = E33 ν M2 =− x3 I22 EY (4.

Therefore. FIguRe 4.eLASTICITy 87 FIguRe 4.3 Torsion of a Circular Cylinder Consider the elastic deformation of a cylindrical bar of circular cross-section (radius a. length l ). 4.9. twisted by equal and opposite moments Mt as shown in Figure 4. It is important to note that the results of this analysis apply only to circular cross-sections. and ui is the displacement field associated with the rotation q. The absence of shear in pure bending means that the cross-sectional planes remain perpendicular to the neutral axis. the neutral axis experiences a state of zero stress.6: A circular cylindrical bar loaded in torsion. i.5: Displacement of the neutral axis in a beam subjected to pure bending. Assume that the motion of each cross-sectional plane is a rigid body rotation about x1.6. unlike the previous two problems. q(x1). q is a small rotation angle (of each cross-sectional plane)..e. .

given u.72) T12 = −μ x3 T13 dθ dx1 dθ = μ x2 dx1 (4. we do not need to check the compatibility condition. (4.88 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 4. the nonzero strains are 1 dθ E12 = − x3 2 dx1 1 dθ E13 = x2 2 dx1 and the nonzero stresses are (4. is Eq.71) Thus. The displacements are (see Figure 4.22).73) ® Notice that.7: Circular plane cross-section of cylindrical bar loaded in torsion illustrating displacements.73) into the equilibrium equation in the absence of body forces. (3.73) a possible state of stress? Substituting Eq. However. (4.7) u1 = 0 u2 = −θ x3 u3 = θ x2 (4. we find ∂ T1j ∂ T11 ∂ T12 ∂ T13 = + + =0 ∂ xj ∂ x1 ∂ x2 ∂ x3 ∂ T2j ∂ T21 ∂ T22 ∂ T23 d2 θ = + + = − μ x3 2 = 0 ∂ xj ∂ x1 ∂ x2 ∂ x3 dx1 ∂ T3j ∂ T31 ∂ T32 ∂ T33 d2 θ = + + = μ x2 2 = 0 ∂ xj ∂ x1 ∂ x2 ∂ x3 dx1 . Eq.

nlateral = The surface traction of Eq.75) R1 = R2 = R3 = M1 = � � � � T11 d A = 0 T21 d A = −μθ � T31 d A = μθ � � � x3 d A = 0 x2 d A = 0 � � (x2 T31 − x3 T21 )dA = μθ � M2 = M3 = 0 � x2 + x2 d A = μθ � Ip 2 3 FIguRe 4.e. On the face x1 = l. the end faces (see Figure 4.74) dx1 where q ¢ is a constant.9).8: Surface normal defining lateral surface of circular cylindrical bar. i. dθ = θ� (4. let us examine the surface tractions. .74). the lateral surface (see Figure 4. t� � μ � (−x2 x3 θ � + x2 x3 θ � ) e 1 = 0 ⇒ a traction free lateral surα ® ® face. So.eLASTICITy 89 Thus.68) gives rise to the resultant forces t� � nend = T e 1 = T21 e2 + T31 e3 ∼ � � � (4.8). Now. (4. n = e1. ⎡ ⎡ ⎤ ⎤⎡ ⎤ 0 T12 T13 x2 T12 + x3 T13 0 1⎢ 1⎢ � ⎥ ⎥⎢ ⎥ � t� = [T ][ nlateral ] = ⎣ T21 0 0 ⎦ ⎣ x2 ⎦ = ⎣ 0 ⎦ ∼ nlateral α α T31 0 0 0 x3 Using Eqs. (4.73) and (4.. Now. equilibrium is satisfied iff the increment in angular rotation (twist per unit length) is constant. First.

provides most of the mechanical integrity of the long bones. We will focus our interest on cortical bone. Os- . and bioapatite (mineral phase. surrounded by concentric layers of mineralized bone tissue. On the end face x1 2 = 0. which. and other supportive tissues.76) ⎢ ⎢ ⎢ M x t 3 ⎢ [T ] = ⎢ − ∼ Ip ⎢ ⎢ ⎣ Mt x2 Ip ⎡ 0 − (4. On either end face. to maintain equilibrium. The basic structural unit of cortical bone is the osteon.9: Shear stress components on end faces. which creates a twist per unit length given by θ� = So. there is a counterbalancing moment. Bone is a hard tissue composed primarily of water.90 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 4. M1 = Mt. chemically similar to hydroxyapatite.10). The osteon consists of a Haversian canal. Ca10(PO4)6(OH)2). There are two major types of bone: cortical (compact) bone and trabecular (cancellous or spongy) bone. even though it comprises only a shell surrounding the trabecular bone. the resultant is a twisting moment. –mq’Ip. which contains blood vessels. type I collagen (organic phase). πα 4 where Ip = is the polar second moment of area for a circle of radius a. nerves. the stress inside the bar is Mt μ Ip ⎤ Mt x3 Mt x2 Ip Ip ⎥ ⎥ ⎥ 0 0 ⎥ ⎥ ⎥ ⎥ ⎦ 0 0 (4. with osteocytes imbedded in those layers (see Figure 4. Torsional loading of the tibia.77) Demonstration.

As one might expect.587 rad = 33.5 MPa 4 Ip πα π (0.015)4 Mt l 2(400)(0. Let us use the above example for torsional loading as a starting point.77) and substituting in the appropriate values for the moment. compact bone has a Young’s modulus of 18 GPa and an ultimate tensile strength of 140 MPa.5 cm and length l = 35 cm. However. Will the tibia fail as a result of this accident? What is the maximum angle of twist in the bone caused by the accident? Solution. the skier’s tibia undergoes a quasi-static twisting. polar moment of inertia (based on radius).35) ∼ 0. (4. teons (and the collagen fibers within them) are oriented in a longitudinal direction. and the maximum angle of twist is relatively large. Let us describe the tibia as a circular cylinder with radius a = 1. and the radius. Under uniaxial tension.6◦ = = 9 4 μ Ip (3 × 10 )π (0. the shear modulus of compact bone is only 3 GPa. whose foot is anchored into a ski.m. this anisotropic arrangement makes bones more susceptible to damage due to transverse or shear loads. During a skiing mishap. Assume that the moment generated the skier’s twisting motion is approximately 400 N. with the foot locked in a stationary point as a force is applied at the end of the ski.015) = = = 75.015) θ �l = Thus.eLASTICITy 91 FIguRe 4. one obtains the following values for the stress and maximum angle of twist. the tibia breaks (Ouch!). T13 = Mt x2 2Mt α 2(400)(0. Taking any of the stress terms from Eq. and the shear strength is approximately 50 MPa.10: Micrograph of an osteon showing the Haversian canal and lamellae (concentric layers of mineralized bone. . Consider the lower leg of a skier.

+ ∂ x1 ∂ x2 Because E33 = 0. E11 = E22 E12 ∂ u1 ∂ x1 ∂ u2 = ∂ x2 � � 1 ∂ u1 ∂ u2 = + 2 ∂ x2 ∂ x1 (4.78) Eq. Eq. (4.92 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 4.1 Plane Strain The state of plane strain can be defined as deformation of a cylindrical body such that there is no axial component of displacement. (4. e u1 = u1 (x1 .79) Substituting Eq. we can identify the nonzero strain components. Let ®3 be the unit vector corresponding to the cylindrical axis.78) is the definition of plane strain.10.78). (4.79) into the generalized Hooke’s law.32) yields T11 = λ Ekk + 2μ E11 T22 = λ Ekk + 2μ E22 T33 = λ Ekk T12 = 2μ E12 T23 = T13 = 0 � ∂ u1 ∂ u2 where Ekk = .10. PLANAR APPRoXIMATIoNS (2d SIMPLIFICATIoN) 4. Substituting the stresses into the equations of equilibrium gives � ∂ T11 ∂ T12 + =0 ∂ x1 ∂ x2 ∂ T21 ∂ T22 + =0 ∂ x1 ∂ x2 ∂ T33 =0 ∂ x3 (4.37) yields T33 = n (T11 + T22). x2 ) u2 = u2 (x1 .80) . (4. From Eq. using the equation for E33 from Eq. x2 ) u3 = 0 (4. (4.

u1 and u2 can be calculated.81) into Eqs. ultimately.11: Examples of plane strain. (4. there is no axial component of displacement. Some examples of plane strain are shown in Figure 4.37) and using the results in Eq. and. (4.x2). (4. Given a f. In particular. we must check the compatibility equations. Eij .83) generates a possible elastic solution. with the sixth being ∂ 2 E11 ∂ 2 E22 ∂ 2 E12 + =2 2 2 ∂ x1 ∂ x2 ∂ x2 ∂ x1 Substituting Eqs. the equilibrium Eqs. The equation is known as the biharmonic equation.eLASTICITy 93 To reduce the number of equations from two to one (the third is already satisfied). Tij . FIguRe 4. (4.x2) that satisfies Eq. such that T11 = T22 = ∂ 2φ 2 ∂ x2 ∂ 2φ 2 ∂ x1 (4.11.x2). we find (4. For plane strain. f(x1. we introduce the Airy stress function.83) Any Airy stress function f(x1. Five of six are satisfied identically. any third-degree polynomial (that also generates a linear stress and strain field) may be used.81) ∂ 2φ T12 = − ∂ x1 ∂ x2 � � ∂ 2φ ∂ 2φ T33 = ν + 2 2 ∂ x1 ∂ x2 For arbitrary f(x1. (4.82).82) ∂ 4φ ∂ 4φ ∂ 4φ +2 2 2 + =0 ∂ x4 ∂ x4 ∂ x1 ∂ x2 1 2 (4. .80) are automatically satisfied. Because the problem is cast in terms of stresses.

(4. anisotropy of a tissue is an easily approachable subject. *4. The various anisotropic constitutive equations are obtained by considering fewer symmetry requirements.10.11 ANISoTRoPIC LINeAR eLASTICITy Discussion of a general theory of anisotropic elasticity is beyond the scope of this text. Furthermore. We begin by recalling Eq. this is a case of ideal symmetry.28).. under infinitesimal theory.g. e. A material with one plane of symmetry is known as a monoclinic material. . skin or heart valves.e. Eq. Let us now determine how the other 19 constants are eliminated. Hence.e.. Cijkl = Cjikl and Cijkl = Cijlk). ⎡ ⎤ T11 T12 0 ⎢ ⎥ [T ] = ⎣ T12 T22 0 ⎦ ∼ 0 0 0 As before.85) is the same under the interchange of ij with kl. Tij = Cijkl Ekl (4. (4. we refer the interested reader to other texts. i. we use the Airy stress function such that T11 = T22 = ∂ 2φ 2 ∂ x2 ∂ 2φ 2 ∂ x1 ∂ 2φ ∂ x1 ∂ x2 (4. Perhaps the simplest form of symmetry is one plane of symmetry. Another common planar simplification of problems is to use an axis of symmetry. Cijkl = Cklij also.94 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 4. We have seen that only two material constants are required to describe an isotropic linear elastic material. considering the symmetry of both T and E .84) T12 = − An example of plane stress is biaxial testing of soft tissues.. as those problems are best formulated using cylindrical coordinates. isotropy is a special condition in which rotation of a particle has no influence on the tensor. linearly elastic material. However.2 Plane Stress Plane stress is defined as T33 = T31 = T32 = 0. However.85) This equation suggests there are 81 (= 34) material constants needed to describe a fully anisotropic. However. As mentioned. there are only 21 ~ ~ independent material constants (i.

eLASTICITy 95 Let e1 be normal to the symmetry plane. C ¢ = Cijkl . T � = Qmi Qnj Qrk Qsl Tmnrs ijkl We learned in Chapter 1 that Eqs. making good use of the summation convention. � � C� 1223 = C1223 = Qm1 Qn2 Qr2 Q13 Cmnr1 + Q23 Cmnr2 + Q33 Cmnr3 = Qm1 Qn2 Qr2 Cmnr3 � � = Qm1 Qn2 Q12 Cmn13 + Q22 Cmn23 + Q32 Cmn33 = Qm1 Qn2 Cmn23 � � = Qm1 Q12 Cm123 + Q22 Cm223 + Q32 Cm323 = Qm1 Cm223 = Q11 C1223 + Q21 C2223 + Q31 C3223 = −C1223 So. Now.91) . This transformation is given by �� e1 = −e1 � �� e2 = −e2 � �� e3 = −e3 � (4. (4. as one should not determine any differences in material behavior between the ijkl transformed and reference coordinates. with all other components of Q equal to zero. one way of defining a fourth-order tensor is that it obeys the following transformation law for an orthogonal transformation.86) where ¢ indicates the transformed coordinates. is ˜ ⎡ ⎤ −1 0 0 ⎢ ⎥ [Q ] = ⎣ 0 1 0 ⎦ ∼ 0 0 1 (4. the components of Cijkl do not change under the following transformation ® �� e1 = −e1 � �� e2 = e2 � �� e3 = e3 � (4. Hence. For material symmetry with respect to the above transformation.88). (4. let us consider C1223.89) C1223 = −C1223 ⇒ C1223 = 0 (4.86) define a reflection.87) and (4. we have reduced the number of constants from 21 to 13. Then. whose transformation. ~ (4.87) (4. Though we have not discussed it so far.90) Similarly. C1112 = C1113 = C1222 = C1233 = C1322 = C1323 = C1333 = 0. if there are two planes of symmetry perpendicular to one another. Given Eqs. So. then the third plane perpendicular to the other two is also a plane of symmetry. Q .88) We see that Q11 = -1 and Q22 = Q33 = 1.

material constants are independent of location. Let us briefly consider how the governing equations would look for an orthotropic problem. reducing the number of material constants by four more.43) is � � ∂ T11 ∂ T12 ∂ T13 + + + ρ B1 = ρ a1. Given the nine independent elastic constants.96 INTRoduCTIoN To CoNTINuuM BIoMeChANICS A material with this symmetry is known as orthotropic.92) where the coefficient matrix is known as the stiffness matrix. For i = 1. (4. There is no further contribution from reflection across the x3-axis. (4.92).93) If we assume that the tissue is homogeneous. so there is a total of nine constants needed to define an orthotropic material. Bone is often treated as an orthotropic tissue. Eq.93) becomes C1111 ∂ 2 u1 ∂ 2 u2 ∂ 2 u3 + C1122 + C1133 ∂ 2 x1 ∂ x1 ∂ x2 ∂ x1 ∂ x3 � + C1212 ∂ 2 u1 ∂ 2 u2 + ∂ 2 x2 ∂ x2 ∂ x1 � + C1313 � ∂ 2 u1 ∂ 2 u3 + ∂ 2 x3 ∂ x3 ∂ x1 � + ρ B1 = ρ a1 . Given the x2-axis is an axis of symmetry.. we additionally find that C1123 = C2223 = C2333 = C1333 = 0. the six components of the stress tensor are given by the following matrix representation ⎤ ⎡ ⎤⎡ ⎤ C1111 C1122 C1133 0 E11 T11 0 0 ⎢ ⎥ ⎢ ⎥⎢ ⎥ 0 0 ⎥ ⎢ E22 ⎥ ⎢ T22 ⎥ ⎢ C1122 C2222 C2233 0 ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ T33 ⎥ ⎢ C1133 C2233 C3333 0 0 0 ⎥ ⎢ E33 ⎥ ⎢ ⎥=⎢ ⎥⎢ ⎥ ⎢T ⎥ ⎢ 0 0 ⎥ ⎢ E12 ⎥ 0 0 2C1212 0 ⎢ 12 ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ 0 0 0 2C1313 0 ⎦ ⎣ E13 ⎦ ⎣ T13 ⎦ ⎣ 0 T23 E23 0 0 0 0 0 2C2323 ⎡ (4. then Eq. this becomes ∂ x1 ∂ x2 ∂ x3 � � ∂ u1 ∂ u2 ∂ u3 + C1122 + C1133 C1111 ∂ x1 ∂ x2 ∂ x3 � � �� � � �� ∂ ∂ u1 ∂ u2 ∂ ∂ u1 ∂ u3 + + + C1212 + C1313 + ρ B1 = ρ a1 ∂ x2 ∂ x2 ∂ x1 ∂ x3 ∂ x3 ∂ x1 ∂ ∂ x1 (4. Given Eq. (4.e. i. and following what we did for the monoclinic symmetry about the x1-axis.

Use this to show the −3p dilatation is given by e = . 4. These are commonly reported as the combination of two Young’s moduli. A transversely isotropic material is one where there is one plane (plane of isotropy) to which every plane perpendicular to it is also a plane of symmetry. With respect to the Lamé constants. and a shear modulus. So. and shear moduli [2]. C1133 = C2233. and 3. C1122 = C1133. by in˜ ˜ ˜ verting the constitutive equation for an isotropic linear elastic solid. 2. two Poisson’s ratios. Let ®3 be the plane of isotropy. T = -p I . Eqs.95) imply that 1. C1111 = l + 2μ and C1122 = l. and 4. E .95) More complicated mathematics show that. C1111 = C2222. For hydrostatic pressure. These last few equations are presented to demonstrate the increased complexity associated with non-isotropic problems.eLASTICITy 97 or C1111 ∂ 2 u1 ∂ 2 u1 ∂ 2 u1 ∂ 2 u2 + C1212 2 + C1313 2 + (C1122 + C1212 ) ∂ 2 x1 ∂ x2 ∂ x3 ∂ x1 ∂ x2 ∂ 2 u3 ∂ x1 ∂ x3 + ρ B1 = ρ a1 (4. 3. Cartilage can be treated as a transversely isotropic material. C1313 = C2323.12 PRoBLeMS 1. In that case. For this state of stress. 2. a more restrictive class of symmetry is transverse isotropy. the number of independent material constants is reduced from nine to five. C1313 = C1212. to describe an isotropic linearly elastic material. where the plane of isotropy is perpendicular to the tissue surface. find the strain tensor. C1212 = ½(C1111 – C1122). (4. The transformation describing a transe versely isotropic material is given by e1 = (cos β ) e1 + (sin β ) e2 �� � �� e3 = e3 e2 = (−sin β ) e1 + (cos β ) e2 � � � � � � (4. A fully isotropic material is one in which there are two perpendicular planes of isotropy. Moving down. 3. C2222 = C3333. 1. Thus.94) + (C1133 + C1313 ) There are similar equations for i = 2. 3λ + 2 μ . The material constants can be expressed in terms of the more familiar Young’s moduli. there remain only two independent constants needed. Poisson’s ratios. in addition to the constants we have already shown to be zero. C1111 and C1122.

) 3. Write Eq.5x4 + 10x3 ⎥ 3 2⎦ 5x4 3 15x2 1 ⎤ 8.3 and E Y = 4. problem 2. show that K = λ + μ . ⎤ ⎡ 25 30 0 ⎥ ⎢ ⎥ ⎢ 5.96) Derive these equations using Cauchy’s equation of motion and the constitutive equation for an isotropic linear elastic solid. the density (or volume. (Hint: find a possible state of stress that satisfies the equilibrium equations. For an incompressible tissue. . 7.96 invariant form.5x4 + 10x3 3 2 1 −18x8 2 ⎥ 7. Consider Eq. A unit cube made of a linear isotropic elastic solid is subjected to uniaxial compression in ® ® the x3-direction.98 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 2. Assume the cube is at equilibrium and the body force is B = -ge 3. 4.8 GPa. Navier’s equations for an isotropic linear elastic solid are (λ + μ ) ∂ 2 ui ∂ 2 ui ∂ 2 uk +μ 2 =ρ 2 ∂ xi ∂ xk ∂t ∂ xk (4. What is e for an incompressible material (recall Chapter 2. Assume no body forces. Assume a linearly elastic isotropic solid with n = 0. 4. isotropic. given that mass is conserved) is constant.) (b) Show that a displacement field exists.5x4 ∼ 2 ⎣ ⎡ 20x3 1 2. Find the strain tensor corresponding to the stress tensor in Chapter 3. 6.4 MPa and m = 22. Given that the strain tensor at a point is [E ] = 10−6 ⎢ 30 16 20 ⎥ ∼ ⎦ ⎣ 0 20 9 (a) Find the stress tensor if the material properties are l = 2.5x4 2 15x2 7. problem 11)? This must be true for any state of stress. Using the constitutive equation of a linear isotropic elastic solid (in terms of l and m) and 2 our definition of the bulk modulus (K ). (4. (c) Calculate Young’s modulus and Poisson’s ratio.32).3 MPa. (Hint: Recall the experi3 mental test used to find the bulk modulus. Does a displacement field exist corresponding to the following strain tensor? Explain your answer. (a) Calculate the stress field. e (b) Find the corresponding stress vector representing this tensor on the ®1 plane. incompressible material. Use the state of hydrostatic pressure to find n for a linearly elastic. ⎢ [E ] = ⎢ 2.

What is the resultant force on the face x1 = L? 11. Be sure to specify if certain components are zero. Assume cortical bone is a linearly elastic isotropic material with E Y = 10 GPa and m = 4 GPa. substituting aggregate modulus for Young’s modulus if necessary. (c) Show that your stress field is a possible state of stress. Please cite your source. (Hint: There are three conditions. (a) Find the magnitude and location of the greatest normal stress on the beam. (d) Briefly explain. (c) Consider a rectangular prism bounded by x1 = 0. (a) Draw a diagram of the bar. The lateral surface of the bar is free of any surface traction. Be sure to cover all of the important steps. FIguRe 4. 9. Consider a cylindrical bar (with center axis aligned along x1) made of a homogeneous. State any simplifying assumptions. (b) Define a reasonable stress field. . During light exercise.) 12. x1 = L. in words.5e 2N·m is applied to the beam. A right hand couple M = 0. x2 = ± h/2. Clearly label the axes and the normal traction s.12: Rectangular sample of cortical bone for use with problem 12.12 is placed ® ® under pure bending. the articular cartilage of the femoral head can still experience substantial stresses. what each condition that your stress field satisfied in part (c) means. and the average stress in the lateral directions is 50 kPa. linear isotropic elastic solid. Assume that the average normal compressive stress is 750 kPa. Consider the Airy stress function f = ax13 + bx12x2 + gx1x22 + dx23 (a) Does this stress function necessarily generate a possible elastic solution? (b) Obtain the stress tensor corresponding to this state of plane strain.eLASTICITy 99 (c) Determine the surface tractions for all six faces of the cube. Equal and opposite normal traction s is applied to each end of the bar.) 10. The rectangular prismatic sample of cortical bone shown below in Figure 4. x3 = ± b/2. What is the percent decrease in the volume of articular cartilage? (Hint: Find Young’s modulus and Poisson’s ratio for hip articular cartilage in the literature.

e. i.. (Hint: I2 = 4 for a circular cross-section with a centroidal axis. length = 10 mm) is subjected to a twisting moment of 1 N·m. (a) Find the shear modulus of cortical bone.100 INTRoduCTIoN To CoNTINuuM BIoMeChANICS (b) Find the strain tensor at this point of greatest normal stress.1 and 4. let x2 = r and x3 = 0. 2 λz (b) Show that Eq. As stated in the text. material. (a) Find the maximum normal and shear stresses and their locations on the rod. Show that the relationship between the shear stress and angle change is T12 = mtang. (b) Find the principal strains as a function of distance along the x2 axis. (b) An identical circular cylindrical sample of cortical bone is subjected to a pure bending test.1).15 radians. Write Eq.. (a) A hyperelastic material with free energy function given by the Mow and Holmes hyperelastic model is subjected to confined compression. i.e.92) for a material transversely isotropic with respect to the ®3 axis. 13. A circular rod of radius R is under torsion. An incompressible neo-Hookean material is subjected to simple shear (see Tables 2.97) � � n Show that the relationship between Tzz and Tz is given by Tzz = 2ψλz α + 2β − . y is the free energy stored per unit initial volume. What is the equation of motion for i = 3? 16.7) implies T = 2ρ F ∼ ∼ ∂ϕ T F ∂C ∼ ∼ (4. (4. • • • • .m. 14. when the other end face is located at x1 = 0? Assume n = 0. A circular cylindrical sample of cortical bone (radius = 2 mm. For confined compression in the z-direction. j is the free energy stored per unit mass. specific strain energy. The resulting twist at one end face is 0. m.) e 15. Assume cortical bone behaves like a linearly elastic isotropic solid. ⎤ ⎡ 1 0 0 ⎥ ⎢ [F ] = ⎣ 0 1 0 ⎦ ∼ 0 0 λz (4. (4. with an applied moment M2 = 1 N. What is the deflection of the bar at x1 = π R4 10 mm.98) where y = r0j. 17.5.

101 CHAPTER 5 Fluids 5. for any n T n = λn ∼ Because every plane is a principal plane.48) for the transpose of a tensor. What does this mean? This result means that on all planes through a point in a fluid. Thus.3) . Consider any two ~® ® ® ® ® ® ® ® ® ® ® ® ® ® planes. These ~ ~ ~ ~ statements imply � � (5. conform to the boundaries of their container. it follows that the stress vector on any plane at any point is ® normal to the plane. Thus.1 INTRoduCTIoN To FLuIdS Fluids are a class of idealized materials that cannot support or sustain a shear stress and. This is why the class of materials known as fluids is said to not be able to support shear stresses.1) n 1 · T n 2 − n 2 · T n 1 = (λ2 − λ1 ) n1 · n2 ∼ ∼ � � � � � � ® ® ® ® ® (5. in this book. Therefore.. In other words. so n 2 × Tn 1 = l1 n 1 × n 2. polymeric solutions. (5. the left-hand side of Eq.2) is 0. as ~ is symmetric. we are primarily interested in liquids. respectively. Examples include water and air. as a result. While fluids include both gases and liquids. every direction is an eigenvector of T. we can write n 2 × Tn 1 = n1 × T Tn2 = n1× ~ ® ® T T n2. T = −pI ∼ ∼ or Tij = . From the definition of a fluid. fluids can be classified as (a) Newtonian (or linearly viscous fluids)—stress depends linearly on the rate of deformation (b) Non-Newtonian —e. which are incompressible and compressible fluids.pdij . so n 1 × T n 2 = l2 n 1 × n 2. where p is the hydrostatic pressure. Hence. n1 and n2. the normal stresses are all the same. Tn 1 = l1 n 1. (5. In terms of stress behavior. Tn 2 = l2 n 2. not only are there no shear stresses.g.2) ® Recall Eq. blood flowing at low strain rates. Any shear stress applied to a fluid induces flow. (1. Also. We define the normal stress as –p (negative normal ® compression). (l2 - l1)n1 × n2 = 0 Þ l1 = l2.

for an incompressible fluid. (3. (5. Eq.7) Recall the equations of equilibrium.9) where the partial derivative changes to a simple derivative in the last expression because the first two expressions show that p is independent of x1 and x2.9) says that p is a function of x3 alone.e. let the x3 axis point vertically down. and B3 = g. Then.3) into Eq. not a function of space or Dr time).2 hydRoSTATICS ∂ Tij + ρ Bi = 0 ∂ xj (5. Thus. in invariant form. � (5. salt water with nonuniform salt concentration).10) . Eqs. we have (5. (2. ∂ xi or.6) Note: Incompressible fluids are not required to have a spatially uniform density (e.7) yields ∂p = ρ Bi .8) where Bi are components of weight per unit mass. (5. (5. such that B1 = B2 = 0. Mathematically this is expressed as ¾ = 0.39). it is referred to as a homogeneous fluid (r constant everywhere). 5.4) ∂ νi =0 ∂ xi In invariant form.102 INTRoduCTIoN To CoNTINuuM BIoMeChANICS An incompressible fluid is one where the density is constant (i. the pressure difference between two points in the liquid separated by a vertical distance h is p2 − p1 = ρ gh (5. If the density is uniform. Dt Dρ ∂ νi =0 +ρ Dt ∂ xi So. = 0. Eq..g. ∇p = ρ B . Eq. Also. ∂ x3 d x3 (5. Eq. Recall the conservation of mass.5) div(ν ) = 0 � or ∇ · ν = 0 � (5.22).8) becomes ∂p ∂p = 0. (5.. (5. and ∂ x1 ∂ x2 ∂p dp = = ρg.

(5. For the cylinder to be in equilibrium. pbottom A − ptop A − W − T = 0 So. the body will float upward generating tension in the rope.1: Example of hydrostatics. and weight W.14) FIguRe 5. Furthermore.. rate of deformation = 0. such that (5. (5. (5.12) T = A( pbottom − ptop ) − W But. . It is the same for all particles on the same horizontal plane. respectively.1). The weight of a body in a container full of liquid is balanced by the buoyant force exerted by the liquid on the body. What is the tension in the rope? Solution. length l. Let ptop and pbottom be the pressure at the top and bottom surfaces of the body. A body of cross-sectional area A. If the fluid is in a state of rigid motion (i.13) T = ρ gl A − W Note: The first term in Eq. is less than that of the liquid. If the density of the body. from Eq. the sum of the vertical forces must be zero.FLuIdS 103 Thus. but inertial effects are non-negligible).e. pbottom . is tied by a rope to the bottom of a container which is filled with a liquid (see Figure 5. r. static pressure in the liquid depends only on the depth. (5.11) Demonstration.14) is the buoyancy force acting on the body.ptop = rgl. let T be the tension in the rope.10). then the governing equation is − ∂p + ρ Bi = ρ ai ∂ xi (5.

104

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

5.3

NewToNIAN VISCouS FLuId

Recall from elasticity that a solid under shear stress deforms and reaches an equilibrium state with a resultant shear strain; this is a reversible process. A fluid under shear stress deforms and reaches a steady state where it deforms continuously with a rate of shear, a process that is irreversible. Thus, for a fluid, shear stress = f (rate of shear strain). For a Newtonian fluid, this relationship is linear see Figure 5.2. Let us decompose T into two parts. ~

T = −pI + T � ∼ ∼ ∼

or

Tij = −pδij + Tij�

(5.15)

where T ¢ depends only on the rate of deformation, D . Recall the strain-rate tensor, Eq. (2.36). ~ ~ For a rigid body motion (i.e., D T ~ = 0), ~ ¢ = 0, and the stress in the fluid is given by Eq. (5.3). However, for nonrigid body motion, following an approach similar to the isotropic linear elastic solid,

**Tij� = λ Dkk δij + 2μ Dij
**

where

(5.16)

1 Dij = 2

�

∂ νi ∂ νj + ∂ xj ∂ xi

�

The coefficients l and m are material constants (not the Lamé constants) and Tij¢ is the viscous stress tensor. Putting Eq. (5.16) into Eq. (5.15), yields the constitutive equation for a Newtonian fluid.

Tij = −pδij + λ Dkk δij + 2μ Dij

(5.17)

FIguRe 5.2: Relationship between shear stress and deformation rate for a Newtonian field.

FLuIdS

105

where Tij is the total stress tensor. Writing these equations out, one gets

T11 T22 T33 T12 T13 T23

= −p + λ Dkk + 2μ D11 = −p + λ Dkk + 2μ D22 = −p + λ Dkk + 2μ D33 = 2μ D12 = 2μ D13 = 2μ D23

(5.18)

The pressure, p, above is a little bit of ambiguous terminology because p is only part of the total compressive normal stress on a plane. Just remember that the isotropic tensor, -pdij , is the part of Tij which does not depend on the rate of deformation.

5.4

MeANINg oF l ANd m

ν1 = ν1 (x2 ), ν2 = ν3 = 0

(5.19)

Consider the shear flow given by the velocity field

The strain rate tensor for this flow is

D12 = D21 =

with all other Dij = 0. Substituting Eq. (5.20) into Eq. (5.18),

1 ∂ ν1 1 dν1 = 2 ∂ x2 2 dx2

(5.20)

**T11 = T22 = T33 = −p T13 = T23 = 0
**

and

T12 = μ

dν1 dx2

(5.21)

We call m the first coefficient of viscosity. It has units of kg/m.s. It is a proportionality constant relating shear stress to the rate of decrease of angle between two mutually perpendicular material lines. Now, consider the trace of Eq. (5.16),

Tii� = λ Dkk δii + 2μ Dii = Dii (3λ + 2μ )

106

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

Thus,

1 � Tii = 3

2 1 T ¢ is the mean normal viscous stress, and D is the rate of volume change. l+ - m, called - ii ii 3 3 the coefficient of bulk viscosity, is the constant of proportionality relating viscous mean normal stress to the rate of volume change.

� � 2 λ + μ Dii 3

(5.22)

5.5

**INCoMPReSSIBLe NewToNIAN FLuId
**

Tij = −pδij + 2μ Dij

(5.23)

For an incompressible fluid, Eq. (5.5), Dii = 0. Thus, Eq. (5.17) becomes

Eq. (5.23) can be written as

Tij = −pδij + μ

Furthermore, from Eq. (5.23),

�

∂ νj ∂ νi + ∂ xj ∂ xi

�

(5.24)

**Tii = −3p + 2μ Dii
**

which, as Dii = 0 for an incompressible fluid, implies

p=

−Tii 3

(5.25)

Hence, for an incompressible fluid, the pressure is the mean normal compressive stress.

5.6

NAVIeR–SToKeS eQuATIoNS

Let us substitute the constitutive equations for an incompressible viscous fluid, Eq. (5.24) into the equations of motion, Eq. (3.21). The resulting equations are the Navier–Stokes equations.

ρ

�

∂ νi ∂ νi + νj ∂t ∂ xj

�

�

= ρ Bi −

∂ 2 νi ∂p +μ 2 ∂ xi ∂ xj

(5.26)

D νi = where, we have used ai = Dt � � Dν ∂ν � �. � = + (∇ ν ) ν a = Dt ∂t

� ∂ νi ∂ νi + νj . In general, for any coordinate system, ∂t ∂ xj

FLuIdS

107

Note: ¶xj2 is equivalent to ¶xj ¶xj , and is thus a summation. For example, the equation for i = 1 is

ρ

�

∂ ν1 ∂ ν1 ∂ ν1 ∂ ν1 + ν1 + ν2 + ν3 ∂t ∂ x1 ∂ x2 ∂ x3

�

∂p +μ = ρ B1 − ∂ x1

�

∂ 2 ν1 ∂ 2 ν1 ∂ 2 ν1 + + 2 2 2 ∂ x1 ∂ x2 ∂ x3

�

In invariant form,

ρ

�

∂ν � � + (∇ν )ν ∂t

�

�

= −∇p + μ ∇2ν + ρ B

�

�

(5.27)

These are the Navier–Stokes equations of motion for an incompressible Newtonian fluid. At physiological pressures, most biological fluids can be considered incompressible and, thus, can be analyzed with these equations. We have three equations with four unknown functions, n1, n2, n3, and p. The fourth equation is supplied by the continuity equation for an incompressible fluid, Eq. (5.6)

∇ ·ν =

�

∂ ν1 ∂ ν2 ∂ ν3 + + =0 ∂ x1 ∂ x2 ∂ x3

(5.28)

Demonstration. If all particles have their velocity vectors parallel to a fixed direction, the flow is said to be parallel, or unidirectional flow. Show that for parallel flows of an incompressible, linearly viscous fluid, the total normal compressive stress at any point on any plane parallel to and perpendicular to the direction of flow is the pressure, p. Solution. Let the x1 axis be the flow direction (n2 = n3 = 0). From the continuity equation, ¶n1 = 0. Thus, the velocity field is independent of x1 and must be of the form n1 = n1(x2, x3, t). For ¶x1 this flow, D11 = D22 = D33 = 0. Hence, from Eq. (5.23), T11=T22=T33 = –p.

5.7

BouNdARy CoNdITIoN

On a rigid boundary, we consider a “no-slip” condition exists. This means a fluid layer next to a rigid surface moves with that surface. If that surface is at rest, then the fluid velocity at that surface equals zero. This is an experimental fact based on numerous observations; a “no-slip” condition applies to practically all fluids.

5.8

•

IMPoRTANT deFINITIoNS

Steady flow —Any property may vary from point to point, but all properties are constant with respect to time at every point.

108 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ∂ρ Dρ �= 0 because r = r (x1. ∂ (ρνi ) =0 ∂ xi ∇ · ρν = 0 � (5. The “no-slip” boundary conditions are ν1 (0) = 0 ν1 (d ) = ν0 (5.300 ⇒ Laminar f low Re > 2. x3). in indicial form. by Eq. n is the mean fluid velocity. (5.9 CLASSICAL FLowS 5. unidirectional flow of an incompressible viscous fluid between two horizontal planes of infinite extent with no pressure gradient in the flow direction. conserva= 0. So. For internal flow in a pipe Re = ρν d μ (5.300 ⇒ Turbulent f low likely (5. Streamlines are drawn such that they are tangent to the instantaneous velocity of flow at every point. n1 = n1(x2) and n2 = n3 = 0. – Laminar regime ® smooth. Re ≤ 2. the other is moving with constant velocity no. layered flow structure – Turbulent regime ® random fluid particle motion in flow structure – The Reynolds number (Re) is a dimensionless parameter used to distinguish between flow structures.1 Plane Couette Flow Plane Couette flow is steady.31) 5. and m is the fluid viscosity. x2.30) Where r is the fluid density. One plate is fixed.32) .29) • • Streamline —One of four basic representations of a flow field.4). but ∂t Dt tion of mass during steady flow is described by For steady flow. turbulent flow —Characterization of the flow structure within viscous flow regimes. Re is a dimensionless parameter representing the ratio of inertial forces to viscous forces. d is the pipe diameter. Laminar vs. The velocity field is. or.9.

(see Figure 5.2 Plane Poiseuille Flow Plane Poiseuille flow is steady. d2 ν1 =0 2 dx2 Solving Eq. It is a linear function of the distance from the rigid boundary as shown in Figure 5.32) gives (5. the size of the channel in the x1 and x3 directions is much greater than the x2 direction.33) subject to Eq.33) ν1 (x2 ) = νo x2 d (5. 5. unidirectional flow of an incompressible Newtonian viscous fluid between two fixed parallel plates of “infinite extent”.. Let the height of the channel be 2a.4) FIguRe 5. (5. (5.9.4: Geometry of Plane Poiseuille flow.FLuIdS 109 FIguRe 5. i. From the Navier–Stokes equations.34) This is the solution for plane Couette flow.3.3: Plane Couette flow.e. .

x3) 0 0 n2(x2.35) ∂p d2 ν1 =μ 2 ∂ x1 dx2 From Navier–Stokes equations in the x2 and x3 directions.e.1: Some basic flow patterns FLow PATTeRN n1 cx2 n2 0 0 -cx3 ¾¾¾2 2 x2 + x3 n3 0 0 cx2 ¾¾¾2 2 x2 + x3 Simple shearing Rectilinear Vortex* Plane n1(x1.35) is two boundary conditions. we obtain dx1 d2 p =0 2 dx1 TABLe 5. (5. x3. (5. dx1 dp Treating ¾ as a constant. we assume the velocity field is. As before. x2. x3.36) becomes dp d2 ν1 =μ 2 dx1 dx2 Differentiating Eq. From Navier–Stokes equations in the x1 direction.110 INTRoduCTIoN To CoNTINuuM BIoMeChANICS and the width be w. There is a singularity on the vortex line. one for a and one for –a. So. (5. (5. the pressure gradient is a constant along the flow direction. The “no-slip” boundary conditions are ν1 (± a) = 0 Note that Eq. and integrating Eq. i. Using this table. the components of Ñn and the strain rate tensor can be found.t) ® *x22 + x32 ¹ 0.t) n3(x2.37) from which we identify the pressure is only a function of x1. Eq. (5. we get (5. (5.38) dp which means ¾ = constant. but. .38) with respect to x1.. there is also a pressure gradient allowed.36) ∂p = 0 and ∂ x2 ∂p =0 ∂ x3 (5. this time. n1 = n1(x2) and n2 = n3 = 0.38) twice with respect to x2. p = p(x1).

.5: Velocity profile of Plane Poiseuille flow. (5. to Eq. dp Recall that n1 = n1 (x2). (5.43) Eqs. the velocity profile is a parabola with a maximum velocity at mid-channel as shown in Figure 5.FLuIdS 111 μν1 = � dp dx1 � 2 x2 + C1 x 2 + C2 2 (5.39) yields � � dp � 1 � 2 2 a − x2 ν1 = − 2μ dx1 (5. � � dp a 2 (5.e.35).40) Thus.5.39) Applying the boundary conditions..41) and (5.40). Eq. The maximum velocity occurs at x2 = 0. Note that n1 > 0 when ¾ < 0 (i. flow is to the right when the pres dx1 sure is greater on the left than the right).44) FIguRe 5.41) νMAX = − dx1 2μ The volumetric flow rate (integral of the velocity over its area of action) is � � �w �a dp 2a3 w Q= ν1 d x 2 d x 3 = − 3μ dx1 0 −a (5.42) The average velocity is νavg = Q a2 = 2aw 3μ � � dp − dx1 (5. Substituting this into Eq. (5. (5.43) together imply νavg = 2 ν MAX 3 (5.

(5. It is of interest to determine how long it takes for steady flow to develop. x3). but n1 now depends also upon t. Thus. i. Suppose we know the inlet and outlet pressures of our system. The Navier–Stokes equations reduce to − ∂ 2 ν1 ∂ ν1 ∂p +μ =ρ 2 ∂ x1 ∂t ∂ x2 (5.47) where h = 2a. all real channels) has edge effects that cause n1 to depend on both x2 and x3. More complicated mathematics yields the following solution [10] ν1 (x2 .e. Consider the same rectangular channel.Pleft)/L.e.x3 ) = 0 and n1 (x2 ± b ) = 0.t ) = 0 and n1 (x2. Pleft and Pright.t) = 2 ∞ 32 (−1) e cos 4x2 ΔPh 2 ⎜ ⎜ ⎜1 − 2 − ∑ 8μ L ⎝ h (2n + 1)3 π 3 n=0 ⎛ n � −(2n+1)2 π 2 μ t ρ h2 � � (2n+1)π x2 h � ⎞ ⎟ ⎟ ⎟ ⎠ (5.39)–(5. fully developed flow in a rectangular channel where the dimension in the x3 direction was much greater than the dimension in the x2 direction.3 extensions of Plane Poiseuille Flow For plane Poiseuille flow. Using these conditions.Pright . This means that p(x1) = dx1 C3x1 + C4.43). Under steady conditions.45) 5. then we can say p(0) = Pleft and p(L) = Pright. If the channel has length L.46) with n1 (± a.9. we assumed steady. n1 = n1(x2.112 INTRoduCTIoN To CoNTINuuM BIoMeChANICS dp Finally. It can be shown that the steady-state solution for n1 in a rectangular channel of height h and width w is [11] . (5. letting DP = Pleft . i.t). where w = 2b.48) − +μ + =0 2 2 ∂ x1 ∂ x2 ∂ x3 with n1 (± a.. This expression allows us to determine how long is a reasonable time after initiation of flow to assume steady conditions (see problem 2). dp ΔP =− dx1 L This can be substituted into any of the Eqs. 0) = 0. respectively. Furthermore.e. let us consider the consequences of ¾ being a constant. a channel with non-infinite width (i.. we get C4 = Pleft and C3 = (Pright .. the Navier–Stokes equations reduce to � � ∂p ∂ 2 ν1 ∂ 2 ν1 (5. n1 = n1(x2.

- ~ 2 D D .FLuIdS n 3 2 ∞ 32 (−1) cosh cos 4x2 ΔPh 2 ⎝ h h � � ν1 (x2 . For inelastic fluids. x3 ) = 1− 2 − ∑ (2n+1)π w 3 3 8μ L h n=0 (2n + 1) π cosh 2h 113 ⎛ � (2n+1)π x � � (2n+1)π x2 �⎞ ⎠ (5. Non-Newtonian fluids can be classified as inelastic or viscoelastic. For viscoelastic fluids. Second. (4. Turning now to the constitutive equations for a class of non-Newtonian fluids.46) and (5.e. For inelastic fluids. These last two examples serve to illustrate two main points. for the elastic solid) or strain rate (e. the fluids discussed in this chapter).3). we will briefly consider some non-Newtonian fluids. we have m = m(I 2~ ) to describe the dependence of the viscosity on the D strain-rate tensor [13]. and a2 are functions of temperature. (5. ~ (5.g. the viscosity coefficient in the viscous stress tensor depends on the strain-rate tensor through its invariants. We defer a discussion of viscoelasticity until Chapter 7. *5. i. etc. Hence.48) are easily obtained with elementary knowledge of partial differential equations (PDEs. (5. where I 2~ = 1 tr(D2). Though the solutions to Eqs. there seems to be no experimental evidence that the vis1 cosity depends on I 3~. which separated the constitutive equation of a fluid into hydrostatic pressure and viscous stress tensor components.. they illustrate how to simplify the Navier–Stokes equations once the assumptions of the problem have been detailed. (5.10 NoN-NewToNIAN FLuIdS In this section. akin to Eq. the concept of memory effects means that the present state of stress depends not only on the current strain (e.. flow rate. can be calculated. see the work of Haberman [12] for details).g. but also on aspects of the deformation history.49) from which the shear stress. and the scalar invariants of D. Hence. we can write the following equation describing a nonlinear viscous fluid T = −pI + α1 D + α2 D2 ∼ ∼ ∼ ∼ (5. we D know from Eq.5). and more importantly. a1. Eq.50) is an isotropic tensor-valued function. they demonstrate the mathematical complexity that comes with including more details of a problem. I ~ = 0. density. it is not constant as it is for a Newtonian fluid. Eq. they are much more complicated than our simple ordinary differential equation. We will consider two simplifications of this general equation.50) where p. and the magnitude of the edge effects can be determined (see problem 2).15). depending on whether memory effects are significant. For incompressible fluids.38). Such fluids are called Reiner–Rivlin fluids [3]. (5. Furthermore.. First. viscosity depends on the current rate of deformation. Non-Newtonian fluids have nonlinear constitutive equations. Recall Eq.

The 1-D simplifica∂ ν1 tion of Eq. must be exceeded before the material will flow. the I2 apparent viscosity. Pseudoplastics are also known as shear-thinning materials (e. whereas for values of n < 1. The viscosity of a power-law fluid is given by � � �n−1 � � D D ∼ ∼ μ = μ I2 = K I2 (5..53) is T � = τy + μ (see problem 3). and g× is the shear strain-rate in the x1 direction. we have t21 = mappg× 1. drilling mud.g. and blood at low shear strain rates. Bingham plastics and Casson fluids are examples of viscoplastic materials. In a 1-D flow configuration. Finally. Let us now consider another class of inelastic fluids. A rheopectic fluid is one for which m increases with time at constant stress (e. Figure 5. such as clay suspensions. paints). These materials behave like an elastic solid at low shear strain and then as a Newtonian fluid above a critical value. A yield stress.52) apparent viscosity decreases. Figure 5.6 compares the 1 relationship between shear stress and strain rate in a 1-D flow configuration for various fluid models. though both eventually behave as Newtonian fluids.g. Demonstration. toothpaste. for values of n > 1. ty. where t21 is the shear stress in the x2 direction on a plane of constant x1. viscoplastic materials. A thixotropic fluid is one for which m decreases with time at constant stress (e. Consider the plane Couette flow of a Bingham plastic.g. polymers).. increases as strain-rate increases.. A shear stress is applied to the upper plate.6 shows the relationship between shear stress and strain rate for several fluids.53) . 21 ∂ x2 What is the velocity of the upper plate as a function of the applied shear stress? � ⎧� ⎪ τ ⎨ � y + 2μ D ∼ D T� = I∼ ∼ 2 ⎪ ⎩ 0 1 2 tr � � 2 T � 2 ≥ τy ∼ � � 1 �2 2 < τy ∼ 2 tr T (5. Power-law fluids for which n < 1 are known as pseudoplastics. the 2 � � �n−1 D ∼ .g.114 INTRoduCTIoN To CoNTINuuM BIoMeChANICS so that The first simplification is a power-law fluid. K � � �n−1 D ∼ D I2 ∼ (5.51) T � = 2K ∼ ~ Because I D is a monotonically increasing function of the strain-rate. The constitutive equation for the viscous stress tensor of a Bingham plastic is We defer a discussion of Casson fluids until the next chapter. This is an uncommon class of materials. epoxy setting). Power-law fluids for which n > 1 are known as dilatants. (5. some dispersions). the viscosity of a fluid may also be time dependent.. A dilatant is a material that exhibits shear thickening (e.

6: Comparison of shear stress versus strain rate for a Newtonian and several non-Newtonian fluids.FLuIdS 115 FIguRe 5. Ñn . The Bingham plastic does not flow until the yield stress is overcome. its diagonal elements are 0. we solve the equation T � = τy + μ 1 . for ∂ν T ¢ < ty . Thus. can be written as (Eq.55) . μ *5.54) D= ∼ Because W is antisymmetric.35)) ® ∇ν = D + W ∼ ∼ where and � (5. ¢ 21 21 ∂ x2 � � � T 21 − τy x2 subject to the boundary condition n1(x2 = 0) = 0. The solution is ν1 (x2 ) = . μ � � � T 21 − τy d the velocity of the upper plate is ν1 (d ) = . For T21 ³ ty . Hence. W31.11 VoRTICITy VeCToR Recall that the velocity gradient. and W23.the velocity of the upper plate is zero. Solution. the spin tensor is equivalent to a vector w expressed by � � 1 � �� � ��T ∇ν + ∇ν 2 � � 1 � �� � ��T W= ∇ν − ∇ν ∼ 2 (5. Thus. and the only three independent off~ ® diagonal elements are W12. (2.

the flow is called irrotational. Let f(x1. � � 1® ζ2 = 0. ® D � �� � � � � � dx = Ddx + Wd x = Dd x + ω × d x ∼ ∼ ∼ Dt D � �� � � � � dx = ∇ν dx and Eq. 2W. x3. *5. The minus sign means the spinning is clockwise looking from the positive side of x3.56) (5. The angular velocity vector.60) ζ1 = � ∂ ν3 ∂ ν2 − ∂ x2 ∂ x3 � = 0.33). x2. � � � � � � � ∂ ν3 ∂ ν2 � ∂ ν1 ∂ ν3 � ∂ ν2 ∂ ν1 � � ζ = 2ω = − e1 + − e2 + − e3 ∂ x2 ∂ x3 ∂ x3 ∂ x1 ∂ x1 ∂ x2 and 2W is the vorticity tensor. and let the velocity components be derived from f by ® ν1 = − ∂φ ∂ x1 ν2 = − ∂φ ∂ x2 ν3 = − ∂φ ∂ x3 (5. 2 normal to the x2 . t) be a scalar function. From Thus. corresponding to a velocity field is zero in some ~ region of the fluid for some time interval. ζ = −k e 3 .61) . n2 = n3 = 0.54) we get Dt (5.59) Let z = 2 w be the vorticity vector. ~ Demonstration. Solution. (2. (5.58) � � � � � � ω = − W23 e 1 + W31 e 2 + W12 e 3 � � � (5. - is z.57) 1 ω= 2 � ® ® � ∂ ν3 ∂ ν2 − ∂ x2 ∂ x3 � � 1 e1 + 2 � ∂ ν1 ∂ ν3 − ∂ x3 ∂ x1 � � 1 e2 + 2 � ∂ ν2 ∂ ν1 − ∂ x1 ∂ x2 � � e3 (5. w is the angular velocity vector of that part of the motion representing the rigid body rotation of a material particle.x3 plane.116 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Wx =ω ×x ∼ Recall Eq. (5. Find the vorticity vector for the simple shearing flow n1 = kx2. z.12 IRRoTATIoNAL FLow If the vorticity vector. or the vorticity tensor. Then. So. ζ3 = −k.

68) (5. (5.62) Then by Eq. ζ3 = 0 (5.. i.e. ∂ xi arbitrary functions f work if.FLuIdS 117 or νi = − ∂φ ∂ xi (5. they also satisfy the continuity equation.65) ∇2 φ = ∂ 2φ ∂ 2φ ∂ 2φ + + =0 2 2 2 ∂ x1 ∂ x2 ∂ x3 ∂φ and Ñ2f = 0.64) yields (5.60).67) (5. =− ∂ x2 ∂ x3 ∂ x2 ∂ x3 ∂ x3 ∂ x2 ζ2 = 0. in addition to Eq. for irrotational flow.63) ∂φ . x2.56) is the Laplace equation for f.62) into Eq. f. in invariant form. (5. (5. (5. t) defines an irrotational flow field if νi = − Recall that for an incompressible fluid the continuity equation is ∂ νi =0 ∂ xi Eq.1 Irrotational Flow of an Inviscid Incompressible Fluid An inviscid fluid is defined by Tij = −pδij The equations of motion for an inviscid fluid are � � ∂ νi ∂ νi ∂p ρ + νj = ρ Bi − ∂t ∂ xj ∂ xi or. x3. � Dν ρ = −∇p + ρ B Dt � (5.12. there exists a scalar. a scalar function f(x1. the vorticity vector becomes � � ∂ ν3 ∂ ν2 ∂ 2φ ∂ 2φ ζ1 = − + = 0.64) ∂ 2φ =0 ∂ x2 i Eq. Of course. (5. ∂ xi (5. Thus. which satisfies νi = − 5.66) So. (5.69) .62).

(5. speed squared). so =0 + + + Ω . If the flow is steady. (5. then ν2 p + +Ω= C 2 ρ (5.75) where C is a constant. i.70) into Eq. for irrotational flow.75) is Bernoulli’s equation. νj ∂ ∂ xi or � � p ∂φ ν2 + + +Ω =0 − ∂t 2 ρ (5. where n 2 = n 1 + n 2 + n3 ∂ xj 2 ∂ xi is the square of the magnitude of the velocity (i. (5.68) results in � � ∂ νi ∂ νi ∂ p =− +Ω + νj ∂t ∂ xj ∂ xi ρ (5. Hence.e. Ω = gx3 yielding B1 = B2 = 0 and B3 = –g. This equation applies when the effect of viscosity can be neglected. in the case of gravity with positive x3 pointing up.72) ∂ νj ∂ν 1 ∂ � 2� But. ∂ νi = ∂ νj (see problem 5).73) − ∂φ ν2 p + + + Ω = f (t) ∂t 2 ρ (5.62).e.e. (5. (5.70) For example. Eq. at every fixed point nothing changes with time. ∂ xj ∂ xi 2 ∂ xi j ∂ xj ∂ xi ∂ νi 1 ∂ν 2 2 2 2 = . For an incompressible fluid with homogeneous density. irrotational flows are always possible provided that the body forces acting are derivable from a potential W. Eq. Eq. But. For whatever func- .71) (5.72) becomes � � ∂φ ∂ νi ∂ ν2 p from Eq. i.. νi = − ∂ xi ∂t ∂ xi 2 ρ Note that nj2 = njnj .. Bi = − ∂Ω ∂ xi (5..74) is the general form of Bernoulli’s equation. (5.118 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Eq.74) where f (t) is some arbitrary function of t. so that νj i = νj = ν . Eq. (5. which is a summation.68) is known as Euler’s equation of motion.

(5. which means C = 0 . constant gravitational force. (c) Find the pressure distribution for an incompressible. the equations of motion can be integrated to give ∂ xi Bernoulli’s equation! Let us briefly consider the assumptions leading to Eq.0). (a) Show that f satisfies the Laplace equation.0. along a streamline subject only to a 5. so that ∇2 φ = ∂ φ + ∂ φ + ∂ φ = 6x + −6x + 0 = 0 (a) 2 2 2 2 2 2 ∂ x1 ∂ x2 ∂ x3 ∂ x1 ∂ x2 ∂ x3 φ ∂ 2φ ∂ 2φ + + = 6x + −6x + 0 = 0 2 2 2 ∂ x2 ∂ x3 1 ∂φ (b) From νi = − . incompressible flow of an 2.FLuIdS 119 ∂φ tion f. We have assumed 1. At other points. steady 4. (b) Find the irrotational velocity field.0.75). ∂ xi ∂φ = −3x2 + 3x2 1 2 ∂ x1 ∂φ ν2 = − = 6x1 x2 ∂ x2 ∂φ ν3 = − =0 ∂ x3 ν1 = − (c) At (0. BerFrom Bernoulli’s equation. Solution. inviscid.0) = p0 and W = gx3. Given f = x13 – 3x1x22. p = p0. Demonstration. and W = gx3 = 0. ν2 p p + + Ω = C . = −6x. homogeneous fluid if p(0. inviscid fluid that is 3. = 0 . 2 2 2 ∂ 2φ ∂ 2φ ∂ 2φ = 6x. as long as νi = − and ∇2 φ = 0 . 2 ρ ρ noulli’s equation yields ρν 2 − ρΩ 2 � � ρ 2 2 p = p0 − ν1 + ν2 − ρ gx3 2 � � �2 ρ � 2 p = p0 − 9 x2 − x2 + 36x2 x2 − ρ gx3 1 1 2 2 p = ρC − . we have n1 = n2 = n3 = 0.

e. i. where ν1 (x2 ) is the steady-state solution solved for in the s. (c) Consider Eq. .26). Next. Show also that. m decreases with time. (a) Show how Eqs. (5. Simplify Eqs. For a thixotropic fluid. Consider plane Poiseuille flow of a thixotropic. . Eq.40) with a = h/2. where T 12 (x2 = −h/2) is the wall shear stress ∞ T 12 (x2 = −h/2) � 1 w/2 for a channel of infinite width.) 3. (5.46) and (5. Derive the Navier–Stokes equations. Then. Note that avgT12 (x2 = −h/2) = T12 (x2 = −h/2. yields an average wall shear stress within 5% of the solution when we considered the width to be infinite? Within 1%? avg T12 (x2 = −h/2) ∞ (Hint: Calculate .53) to find T12 for a ¢ power-law fluid and for a Bingham plastic.47) reduces to Eq. t) s. (5. use Mathematica©’s ‘N[ ]’ command to evaluate the sum..46). What aspect ratio.49) reduces to Eq. as w → ∞. how long does it take for the velocity at mid-channel to be within 5% of its final value? Within 1%? (Hint: First calculate ν1 (x2 . what values of K and n yield the constitutive equation for a Newtonian fluid? In the case of a Bingham plastic..s.48) are obtained from the Navier–Stokes equations. let the x3 axis point vertically down such that B1 = B2 = 0.15).) (d) Consider Eq. x3 )dx w −w/2 . In the case of a power-law fluid.) T12 (x2 = −h/2. we will look at various aspects of the equations for the extensions of plane Poiseuille flow.e.52) and (5. Find the shear stress at the wall (x2 = -h/2). how much greater does w have to be than h. Assuming m depends on time as m(t) = 10e-3t + 2. (5.s. consider gravity as the only acting body force (i. (5. (b) Show in the limit as t → ∞ that Eq. In terms of h. and B3 = g) and the constitutive equation for the fluid to be Eq. from Cauchy’s equation of motion and the constitutive equation for an incompressible Newtonian fluid. Consider a flow described by n1 = n1(x2). μ. and r. (5. Note that these equations apply only to an incompressible Newtonian fluid. (5. x3 )dx3 . Show that the equation of motion in the x3 direction is ρ � ∂ ν3 ∂ ν3 + νi ∂t ∂ xi � = ρg − � ∂ Ti3 ∂p + ∂ x3 ∂ xi 2. (5. what value of ty results in a Newtonian fluid? 4. (5. In this problem. (5. ν1 (x2 ) 1 /2) = w � w/2 −w/2 text. incompressible Newtonian fluid.120 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 5.40).49).13 PRoBLeMS 1. Eq.

Show that this implies ∂ νj ∂ νi = .e. where Wij’s are components of the antisymmetric ® ® spin tensor. 8) A velocity profile for blood plasma.. which is written as ® ® ® ® z = -2(W23e1 + W31e2 + W12e3) = -eijk Wjkei . all properties independent of time) (b) Irrotational flow (i. z = 0. (d) Find the acceleration field in terms of the pressure field. Simplify the Navier–Stokes equations by successively making the following assumptions: (a) Steady-state flow (i.. z . (b) Find the stress tensor. B = -ge3 = -gÑx3 ) Now. directly from the Navier–Stokes equations.e.. derive the pressure distribution. For irrotational flow. with p = p0 at the origin. Irrotational flow is where fluid elements in the flow field do not undergo rotation. which relates pressure changes to changes in elevation and velocity. Rotational flow is described by the vorticity vector. What is the effect of m on wall shear stress? 5. c constants (a) Determine the relationship between c and k. and eijk is the permutation symbol. ¶ni / ¶xi = 0) ® ® (d) When the only body force is gravity (i. This result is a powerful and widely used equation. ∂ xj ∂ xi 6. 7. known as Bernoulli’s equation. is as follows: ν1 = 4x2 x2 − 9x2 1 2 ν3 = 4x1 x2 x3 ν2 = 12x3 − 6x1 x2 3 2 (a) Is this velocity profile consistent with the assumption of an incompressible fluid? (b) What is the simplest linear constitutive equation you would use in this situation that relates stress to strain rate? .e. ¶ni / ¶xj = ¶nj / ¶xi ) (c) Incompressibility (i. a viscous fluid. integrate the expression..e. Given the following flow field of an incompressible Newtonian fluid ν1 = kx1 ν2 = −3kx2 ν3 = cx3 with k.FLuIdS 121 plot the velocity at mid-channel from t = 0 to t = 5. (c) Neglecting body forces.

x3 = 0) is PBC. one asks how you were able to handle the pressure change occurring from base camp to the summit. Being a fellow engineer. Everest..122 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 9) Over spring break. The equation for fluid hydrostatics is Ñp = rB. the temperature can be written as a linear function of x3. Everest. Furthermore. the only force ® ® acting on this column of air is gravity (i. your friend wants precise information regarding the pressure distribution you encountered while climbing to the ® World’s highest point. has the equation of state p = rRT. While regaling your friends with stories from your adventure. In this column of air. B = -ge3 with positive x3 directed upward). You may assume that the acceleration due to gravity did not change appreciably during your climb.e. Consider air to be an ideal gas. with appropriate scaling. • • • • . T = TBC – ax3 (where TBC is the cold base camp temperature). which. you decided to scale Mt. We wish to calculate the pressure as a function of distance from base camp in the column of air surrounding Mt.e. The pressure at base camp (i..

2 BASICS ANd MATeRIAL PRoPeRTIeS oF BLood Blood is a viscous fluid mixture of plasma and cells. serves to regulate temperature by transporting heat from the body’s core to the exterior. we will develop governing equations for blood flow. while the forces opposing blood flow are shear forces due to viscosity and turbulence.1 kg/m. our goal is to study the rheological behavior of blood to shed insight into its bulk transport capacity. we must start with some basic properties and concepts. pressure gradients. blood has several other important roles. To accomplish this task.s = 1 dynes/cm2 Force ® 1 N = 105 dyne 6. and muscle contraction (veins). of which 3 L are plasma and 2 L are cells. lactic acid) to other organs for breakdown and carbon dioxide back to the lungs for exhalation.1 INTRoduCTIoN Blood is the main fluid of the circulatory system. functioning in transport and helping to maintain homeostasis. Blood is involved in controlling the body’s pH by acting as an extensive buffer.s/m2 = 0. and frictional loss in vessels. However. The pressure at a given point is the sum of static pressure due to gravity.g.1 N. Hence. and is approximately 40%. pressure due to beating of the heart..1 Pa. The remainder is made up of platelets . The forces driving blood flow are gravity.123 CHAPTER 6 Blood and Circulation 6. some useful unit conversions when talking about blood flow are the following: Viscosity ® 100 cP = 0. including constitutive equations and expressions for flow rate. The hematocrit is defined as the fraction of blood volume due to cells. RBCs. and is the staging ground for many cells of the immune system. In this chapter. However. The blood volume is ~5 L. mainly red blood cells (RBCs). proteins and carbohydrates) to cells throughout the body and then carries away waste (e.g. blood delivers oxygen picked up in the lungs and nutrients from internal stores (e. account for more than 95% of blood’s cellular composition. The flow of blood has often been described in the c-g-s system instead of the m-kg-s system.. also known as erythrocytes.s = 0. In terms of transport.

If the cellular components of blood are removed. 6. except clotting proteins have been removed. so RBCs must deform and flow in single file. Figure 6. The effects of platelets and WBCs are negligible on macroscopic flow. Red blood cells exhibit r = 1.16) for Newtonian fluids. turbulent flow ensues.024 g/cm3. in general. serum is essentially the same as plasma. or.056 g/cm3. and has been strongly implicated in atherogenesis. Their diameter is ~7. when Re > 2. (6. using an older notation.4 NoN-NewToNIAN BehAVIoR oF BLood Recall Eq. one is left with plasma.1%). If we look at only at plasma. Consider the simple 1-D case. or white blood cells (WBCs).124 INTRoduCTIoN To CoNTINuuM BIoMeChANICS (~5%) and leukocytes (~0.2 cP and r = 1. resulting from a higher friction coefficient. T12 = 2mD12.1). m. This low viscosity value is at high shear rates and 37°C. flow is laminar in vessels that are sufficiently small. hence.7 mm. Tij¢ = lDkkdij + 2mDij . In Eq.000 Small arteries and veins ® Re ~ 1 Microvessels ®Re < 1 Capillaries ® Re ~ 10-2 Therefore. that capillaries are ~4–10 mm in diameter. and their thickness is ~1–2. (5. and ~2% various organic and inorganic substances.1) . Whole blood is considered to exhibit a viscosity around m = 3 to 6 cP. and g is the strain rate (the dot over g indicating a time rate of change). where the viscosity. Within the human circulatory system. however.6 mm. 6. m = 1. One obtains serum if the plasma is allowed to clot. ~7–8% protein.1 shows shear stress as a function of strain rate for blood. whole blood viscosity can approach 100 cP. Note. from which we can see . is constant. Under different conditions. which is 90% H2O. Turbulence dissipates energy. we learned that. which most often affects larger vessels. typical values of Re are as follows: Human aorta ® Re > 3.098 g/cm3.300. ˙ τ = μγ (6. The density of whole blood is r = 1.65 ± 0. t is the shear stress.3 ReyNoLdS NuMBeRS FoR BLood In Chapter 5.

though others have been put forth. m is not constant. Figure 6. blood is classified as a thixotropic Casson fluid. Thus. This is the simplest constitutive equation for blood. This portion of the curve resembles a pseudoplastic for which as ˙ ↑γ⇒↓μ Furthermore. The yield stress of blood is ~0. . occurs at about g > 700 s-1 for normal values [14]. blood.. and its effect on facilitating RBC adhesion.BLood ANd CIRCuLATIoN 125 FIguRe 6. However.e. similar to a Bingham plastic. the curve is essentially linear. g plot. is largely responsible for the non-Newtonian behavior of blood. . ty. Based on these empirical observations. yielding the viscosity as a function of shear strain rate. Aggregation of RBCs is strongly dependent on shear strain rate. This behavior of blood is represented by the very beginning of the curve. as they break up as strain rate increases. The slope of the curve is not constant. has a yield stress. indicating that the viscosity of blood is not constant. creasing strain rate. the viscosity of blood decreases. that the relationship is clearly nonlinear. RBCs stack into structures called rouleaux. i.05 dyne/cm2 [15]. which can clump into larger aggregates. although we will neglect time dependence in our analysis. Fibrinogen. The end result is the curved part of the t vs. g curve. representing a Newtonian fluid. with in. at high strain rates. The transition of blood to Newtonian behavior (constant m) is a function of hematocrit and .2 shows the derivative of the t vs.1: Shear stress versus strain rate for blood. At low strain rates.

2) .η = � √ τy + ˙ |γ | � ˙ η |γ | �2 (6. incompressible Newtonian . we note that as g becomes large. For blood. when t and g = 0.6 BLood RheoLogy If the shear strain rate is sufficiently high. blood behaves as an isotropic. fluid. .05 dyne/cm2 (which is very small). and g the shear strain rate. m ® h.3). One can use the Casson equation for hematocrits £ 40%. The isotropic assumption here means that.3) 6.5 CASSoN eQuATIoN √ τ= � Blood seems to obey the following empirically derived equation τy + . ty is the yield stress. (6. (6.τy .126 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 6. � � ˙ μ γ . h a constant. and ty » 0. h » 4 cP (and depends on the hematocrit).2: Derivative of the t versus g curve for blood. (6.1) into Eq. Substituting Eq. 6. From Eq. where t is the shear stress.2) yields � ˙ ηγ (6. blood cells have no preferred .

BLood ANd CIRCuLATIoN

127

orientation. Furthermore, blood is incompressible at physiological pressures. Recall the equation for an incompressible Newtonian flow, Eq. (5.23)

Tij = −pδij + 2μ Dij

(6.4)

How can we generalize this Newtonian equation to account for blood’s non-Newtonian behavior? We will introduce a “positively valued” second invariant, J2. Recall Eqs. (1.78)–(1.81), which say that a symmetric tensor has three invariants

**I1 = Tii 1 (Tii Tjj − Tij Tji ) 2 I3 = det(T ) ∼ I2 =
**

Let (6.5)

J2 =

1 2 1 I − I2 = Tij Tji 2 1 2

(6.6)

Because I1 = 0 for incompressible fluids, J2 is essentially a sign change for I2. For the strain-rate tensor, 1 J2 (D) = Dij Dij (6.7) ∼ 2 where we have switched the order of the last indices, as D is symmetric. For non-Newtonian behavior, the viscosity can be expressed as a function of J2 such that

**Tij = −pδij + 2[ μ( J2 )]Dij
**

For a simple 1-D flow,

(6.8)

˙ γ = 2D12

and

(6.9)

J2 =

Substituting Eq. (6.10) into (6.9),

� 1� 2 2 2 D12 + D21 = D12 2 � ˙ γ = 2 J2

(6.10)

(6.11)

128

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

In general, Eq. (6.11) into Eq. (6.3) plus some algebra yields

μ ( J2 ) =

where J2 is given by Eq. (6.7).

�� � 1/2

τy 2

� �1/4 + η 2 J2 √ J2

�2

(6.12)

6.7

SuMMARy

Blood viscosity can exhibit three behaviors based on three flow regimes. 1. Elastic. Before the yield stress is overcome, blood behaves like an elastic solid, e.g., a linearly elastic solid may be proposed. Below the yield stress, blood does not flow and shear strain rate is zero. However, at some point, the yield stress will be overcome. We can describe the yield condition using J2(T 0), as is often done in the theory of plasticity [14], where T 0 is ~ ~ the deviatoric part of the stress tensor. Recall Chapter 3, problem 1,

**1 Tij0 = Tij − Tkk δij 3
**

The yield condition is given by

(6.13)

J2 (T 0 ) = ∼

1 0 0 T T 2 ij ij

(6.14)

For J2(T 0) < K, blood remains elastic (obeys Hooke’s law, and Dij = 0), while for J2(T 0) ³ K, ~ ~ blood yields and flows. K ~ 4–25 ´ 10-6 N2/m4 and depends on the hematocrit. The constitutive equation is (recalling Eqs. (4.30) and (4.38) applied to an incompressible material)

Tij =

2EY Eij 3

(6.15)

2. Non-Newtonian flow. Under this regime, blood yields and flows. Given J2(T 0) ³ K, the ~ constitutive equation is given by Eqs. (6.7), (6.8), and (6.12)

. when J2(D ) < c. From Eq. (6.11), and a transition around g > 700 s-1, c = 122,500 s-2. ~

⎧� � ⎫ �1/4 2 ⎪ ⎪ � τy �1/2 � 2 1 ⎪ ⎪ ⎪ ⎪ 2 + η 2 Dkl Dkl ⎬ ⎨ �� D Tij = −pδij + 2 � ⎪ ij ⎪ 1 ⎪ ⎪ Dkl Dkl ⎪ ⎪ 2 ⎭ ⎩

(6.16)

BLood ANd CIRCuLATIoN

129

TABLe 6: Summary of the constitutive behavior of blood RegIMe CRITeRIA CoNSTITuTIVe eQuATIoN

Elastic Non-Newtonian fluid

J2 (T 0) < K ˜ J2 (T 0) ³ K ˜ J2 (D 0) < c ˜ J2 (T 0) ³ K ˜ J2 (D) ³ c ˜

Tij =

2EY Eij 3

Tij = -pdij + 2[m( J2(D ))] Dij ˜ Tij = -pdj + 2Dij

Newtonian fluid

K ~ 4–25 ´ 10-6 N2/m4 and c ~ 125,000 s-2.

. 3. Newtonian flow. At high shear strain rates, past the transition point of ~ g > 700 s-1 with J2(T 0) ³ K and J2(D ) ³ c, the constitutive equation is given by Eq. (5.23) ~ ~

Tij = −pδij + 2μ Dij

(6.17)

where m » 4 cP. Table 6.1 summarizes the constitutive behavior of blood. We will now use these concepts to examine the flow of blood in a tube, e.g., an artery. Though, perhaps, seemingly overcomplicated, the development of Eq. (6.16) is an essential step in describing blood rheology. The Casson equation is sufficient for problems in which the flow is simple, but complicated flows require this more robust constitutive equation for their analyses.

6.8

LAMINAR FLow oF BLood IN A TuBe

Consider the flow of blood in a circular cylindrical tube of diameter d (e.g., an artery). We will make the following assumptions: 1. laminar, 2. steady, 3. axisymmetric, and 4. fully developed flow (no entrance effects). Considering assumption 4; for fully developed flow in a cylindrical tube the entrance length is given by

Le = (0.06)Re d

(6.18)

where Re is the Reynolds number and d is the tube diameter. Le describes the minimum distance of flow upon entrance into the tube for which variations of velocity in the direction coincident with the tube’s long axis can be neglected. Note that the distance between successive artery bifurcations may not satisfy assumption 4. Furthermore, for laminar flow in cylindrical tubes, Re < 2,100 [10].

130

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

FIguRe 6.3: Geometry for Poiseuille flow in a cylindrical tube.

In the derivations below we will use cylindrical coordinates, r, q, and z (see Figure 6.3). Assuming axisymmetric flow, all variables are independent of q. Furthermore, past the enter once length, nz can no longer depend on z. Hence, nz = nz(r). Newtonian flow regime. We have cylindrical Poiseuille flow (see Section 5.9), where Poiseuille flow is laminar, steady, pressure-driven flow of an incompressible Newtonian fluid. In cylindrical coordinates (Ñ has different expressions depending on the coordinate system used) the Navier– Stokes equations give for 1-D flow, analogous to Eq. (5.38),

� � dp μ d dνz = r dz r dr dr

The “no-slip” boundary condition is

(6.19)

νz (r = a) = 0

(6.20)

The other boundary condition is a symmetry condition describing the fact that nz is maximum at the center of the tube (r = 0). Mathematically,

From Eqs. (6.19)–(6.21), it can be shown (see problem 1) � � dp � 1 � 2 2 a −r νz (r) = − 4μ dz

� dνz � � =0 dr �r = 0

(6.21)

(6.22)

Q=

�

A

nz dA = 2p

�a

0

p a4 nz rdr = 8m

�

dp − dz

�

(6.23)

44). At rc.4). � � − (P2 − P1 ) t (2p r)L + P2 p r 2 = P1 (p r 2 ) ⇒ 2t = r L As L ® 0. .24) � � dp 1 − = ν MAX dz 2 (6. Casson (non-Newtonian) flow regime. Consider the force balance on a fluid element (see Figure 6. 2 dz From Eq. 1 dp .22)–(6. Let this distance be rc.BLood ANd CIRCuLATIoN 131 ν MAX = νavg = a2 8μ a2 4μ � � dp − dz (6. (6.26) Note that τ |r=0 = 0 Þ τ |r=0 = 0 < ty ! The plot of the shear stress profile is shown in Figure 6.25) Note the similarity of Eqs.27) At some radial distance from the tube’s center. the shear stress will overcome the yield stress. (5.4: Force balance on a fluid element in cylindrical coordinates.40)–(5. (6.26). τ =− r dp 2 dz (6. This example illustrates how the geometry of the problem can affect the solution.28) FIguRe 6. the shear stress is equal to the yield stress and is given by rc τy = 2 � � dp − dz (6. the shear stress at the wall of the cylinder is � � a dp τw = − 2 dz where the slope of the line is − (6.25) with Eqs.5.

yields � � 1 dνz (r) dp �√ √ �2 (6.. (6.e. With the positive z-axis into the right in Figure 6.2).22) for small rc. (6.31) − a − r − rc dz 3 which reduces to Eq.28) into Eq. a > rc) and substituting Eqs.27) and (6. (6. and (6.5: Plot of shear stress as a function of distance from the center tube. .20). Eq.29) into Casson’s equation.30) can be integrated to yield 1 νz = 4η � � �� � dp 8 1/2 � 3/2 2 2 3/2 a −r + 2rc (a − r) for rc < r < a (6. (6.31) give aτw νz = 2η � �� � r �2 � 8 � τ � � r �3/2 � 2τ � r� y y 1− − 1− 1− + for rc < r < a (6.30) = − − r − rc dr 2η dz Using the “no-slip” boundary condition of Eq.32) a 3 τw a τw a . −τ = μ z .29) Assuming tw > ty (i. dν Recall the notation t =mg . (6.28). Alternatively. Eq.26). dr from which we identify ˙ γ=− dνz dr (6. (6. Eqs.132 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 6.4. (6. (6.

As e ® 0.6: Velocity profile of a Casson fluid in a circular cylinder. F(e) ® 1. (6. Using the first equality of Eq. (6.25).33) is shown in Figure 6.31) and (6. This is a model for blood flow in artery. and Q approaches that of an incompressible Newtonian fluid with m = h. (6.33) Compare this last result to Eq.37) The last of which is similar to Eq. FIguRe 6. we get a complicated equation for the volumetric flow rate.BLood ANd CIRCuLATIoN 133 At r = rc (also 0 < r < rc). As before we can calculate 2τy /a τy �= ε= � τw −dp/dz (6.36) νavg a2 = 8η � � dp − dz �� F (ε ) (6.7 shows a plot of F(e). .35) n MAX = n (rc ) and (6. Due to the shape of this profile.31) as nz(rc) 1 νz = 4η � � �� 2 dp 8a √ rc − rc a + 2rc a − a2 − for 0 ≤ rc ≤ r dz 3 3 (6.23). The flow profile of Eqs. which can be expressed more conveniently as � � �� π a4 dp (6. (6.6.24). the core velocity is given by Eq. (6.34) Q= − F (ε ) 8η dz where F (ε ) = 1 − 16 1/2 4 1 ε + ε − ε 4. 7 3 21 Figure 6. rc is known as the plug radius.

and α = a . due to the beating of the heart. denotes the real part of the imaginary number. and nr = nq = 0? .7: Plot of F(e) versus e. μ which describes the ratio of unsteady forces to viscous forces.g. (6. when � � N N ∂p ΔP ΔP =− + ∑ [ψn cos (nω t) + φn sin (nω t)] = − + ∑ R ϒn en(iω t) ∂z L L n=1 n=1 of a Newtonian fluid. This completes the task we set out with. Of course. show how to get Eqs.134 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 6. R[ ] .9. J0 is a zero� ωp order Bessel function of the first kind. a is known as the Womersley parameter. (Hint: Look up the Navier–Stokes equations in cylindrical coordinates) What do they become when nz = nz (r). In the manner of Section 5.. t. i. t) = 1 − � 3/2 1/2 � ⎠en(iω t) ⎦ 4μ L nωρ J0 i n α n=1 where ¡n = yn . i = (6. The solution for pulsatile flow. √ 6. namely.e. For example.38) (6. blood at high shear rates. in a cylindrical tube is [15] � 3/2 1/2 � ⎞ ⎡ ⎛ ⎤ � N J0 i na α r i ϒn ⎝ ΔP � 2 a − r2 + R⎣ ∑ vz (r. e.9 PRoBLeMS 1. and z.25).ifn. we have pressure as a function of both time. we can increase the complexity of our analysis in several ways.39) −1. to develop the constitutive equations for blood flow and an equation for the flow rate.22)–(6.

q. In this chapter. in cylindrical coordinates are (∇α )r = ∂α ∂r (∇α )θ = 1 ∂α r ∂θ and (∇α )z = ∂α ∂z (6.40) You will need to know that the components of the divergence of a tensor. Solve for pressure-driven flow of a Casson fluid between two parallel plates of infinite extent. Solve for steady Couette flow of a Casson fluid. cylindrical coordinates are (∇ · A)r = ∼ ∂ Arr 1 ∂ Arθ ∂ Arz Arr − Aθ θ + + + ∂r r ∂θ r ∂z ∂ Aθ r 1 ∂ Aθ θ Arθ + Aθ r ∂ Aθ z (∇ · A)θ = + + + ∼ ∂r r ∂θ r ∂z ∂ Az r 1 ∂ Azθ ∂ Az z Azr (∇ · A)z = + + + ∼ ∂r r ∂θ r ∂z (6. ~ q. Solve for the flow in a cylindrical tube of a power-law fluid (i. in cylindrical coordinates is � � ⎡ ∂ ar 1 ∂ ar ∂ ar ⎤ − aθ ⎢ ∂r r ∂θ ∂z ⎥ ⎢ ⎥ � � ⎢ ⎥ ∂ aθ ⎥ ⎢ ∂ aθ 1 ∂ aθ � [∇ a] = ⎢ ⎥ + ar ⎢ ∂r r ∂θ ∂z ⎥ ⎢ ⎥ ⎣ ⎦ ∂ az 1 ∂ az ∂ az ∂r r ∂θ ∂z (6. a (r. (Hint: Reread the demonstration of Section 5.41) The components of the gradient of a scalar. z. What is the shear stress at any height above the fixed plate? Find the velocity of the upper plate as a function of the shear stress..10) 3.42) ® The gradient of a vector. we solved for the flow in a cylindrical tube of a Newtonian fluid and a Casson fluid. z. in A(r. How does this compare to the solution for a Newtonian fluid? 4. t). a(r. t).BLood ANd CIRCuLATIoN 135 2. z. find nz(r)) and show that the volumetric flow rate is nπ Q= 3n + 1 � ΔP 2KL �1 n a (3n+1) n (6. t). q.e.43) .

) 6. only physical intuition. Solve for the flow in a cylindrical tube of a Bingham plastic. a disease process in which the body’s arteries clog. Show that π a4 Q= 8μ L when ΔP > pB ≡ � 1 p4 4 B ΔP − pB + 3 3 ΔP 3 � (6. (a) What is the mathematical expression for wall shear stress in cylindrical coordinates of a Newtonian fluid? (b) Evaluate the expression in (a) to find the wall shear stress for Newtonian flow of blood in an artery. where J1 is a first-order Bessel function dr 0 of the first kind.136 INTRoduCTIoN To CoNTINuuM BIoMeChANICS The components of ~ in cylindrical coordinates are D ⎥ ⎢ ⎥ ⎢ [D] = ⎢ Dθ r Dθ θ Dθ z ⎥ ∼ ⎦ ⎣ Dzr Dzθ Dzz � � ⎡ 1 1 ∂ νr ∂ νθ ∂ νr νθ + − ⎢ ∂r 2 r ∂θ ∂r r ⎢ ⎢ � � ⎢ 1 1 ∂ νr ∂ νθ νθ νr 1 ∂ νθ =⎢ + ⎢ 2 r ∂θ + ∂r − r r ∂θ r ⎢ ⎢ � � � � ⎣ 1 ∂ νr ∂ νz 1 ∂ νθ 1 ∂ νz + + 2 ∂z ∂r 2 ∂z r ∂θ ⎡ Drr Drθ Drz ⎤ 1 2 1 2 � � ∂ νr ∂ νz + ∂z ∂r ⎥ ⎥ �⎥ 1 ∂ νz ⎥ ∂ νθ ⎥ + ∂z r ∂θ ⎥ ⎥ ⎥ ⎦ ∂ νz � ⎤ ∂z (6. narrow. This question is concerned with wall shear stress.44) 5. What is the volumetric flow rate when DP < pB? (This last quesR tion requires no math.) .45) 2Lτy . and harden. (c) What is the wall shear stress for non-Newtonian blood flow based on the Casson equation? (d) What is the time-dependent part of the wall shear stress during pulsatile flow? (You d will need to know that [J (kr)] = −kJ1 (kr). Shear stress along the walls of blood vessels has been implicated in atherosclerosis.

Clinical Hemorheology 16(4). Read Hemorheological Observation on 139 Cases of Essential Hypertension by Casson Equation. Label the plot clearly.4 dyne/cm2 and h = 4. Although we neglected time dependence in our analysis. referring to Figures 5–8 on p. 568). 8. ty = 0. Also assume steady. 4 (p. and identification of statistically significant differences where applicable.e. (a) Make a bar graph comparing the yield stress of blood for male and female normal and hypertension subjects. 9. how would viscosity actually change with respect to time (i. For blood flowing in the vena cava. specifically the first sentence? Explain.BLood ANd CIRCuLATIoN 137 7. • • • • . 559–570. meaning its viscosity is time dependent. is blood flow laminar or turbulent? (b) As mentioned in class. standard deviations. Assume that the inferior vena cava (IVC) has an inner diameter of 1. sample size(s). What does this mean when considering whether to use a Casson fluid or Newtonian fluid to model blood in this situation? (b) Calculate the flow rate for blood in the IVC assuming it is a Newtonian fluid. pp.5 cm and experiences a pressure drop of 1. Also find the max velocity and sketch the profile. 567. laminar flow. (a) Determine the values of e and F(e) for blood flow in the IVC. 1996. blood is a thixotropic fluid.0 cP. (c) Repeat part (b). including units. this time modeling blood as a Casson fluid. would it increase or decrease)? (c) Briefly explain why/how the non-Newtonian behavior of blood flow is attributed to red blood cells. Concepts from biomechanics of circulation.0 mmHg over 10 cm. (b) What inference(s) about the usefulness of measuring yield stress can you make? (c) Do you agree with discussion point no. Shi et al. (a) In the majority of blood vessels in the human body.

.

this definition is described using an integral form. or temporomandibular joint disc (d) Strain-rate-dependent properties—when the stress–strain behavior of a material depends on the strain rate Example: Straight pulls of a soft tissue to a given strain. We mentioned viscoelasticity in the section on nonlinear fluids in Chapter 5. viscoelastic constitutive equations provide appropriate models for many tissues and polymers.139 CHAPTER 7 Viscoelasticity 7.1 INTRoduCTIoN Thus far. done at several different strain rates 7.2 deFINITIoN oF VISCoeLASTICITy A viscoelastic material exhibits creep. including soft tissues. A viscoelastic material is defined as a material where the stress depends on the strain history of the material. stress relaxation. and strain rate-dependent material properties.g. where the integral is known as a hereditary integral. we are going to begin discussion of materials that simultaneously exhibit characteristics of both elastic solids and viscous fluids. cells.1–7. skin under tension (c) Hysteresis—dissipation of energy during loading–unloading cycles Example: Cyclic tensile loading of articular cartilage. and biological and synthetic polymers. Newtonian viscous fluids. viscoelastic materials are often referred to as materials with memory. we have discussed the material behavior of elastic solids. fall into this class of materials and exhibit the following behaviors (see Figures 7. knee meniscus. For this reason. Now. tendon. Thus. Mathematically. the stress in the body decreases over time Example: Stress relaxation of a soft tissue. Many materials of interest in biomechanics. which are empirically known to exhibit such behaviors. and some nonlinear viscous fluids. e. hysteresis.4): (a) Creep—continued deformation over time of a body maintained under a constant stress Example: Creep indentation of articular cartilage (b) Stress relaxation—when a body is suddenly strained and the strain is maintained. where it was introduced that the current stress in the material depends on aspects of the deformation history. . ligament..

and look at how they can be combined to model more complicated behavior. and the stress is monitored over time.2: Stress relaxation of a viscoelastic material. Let us first consider the two components that will make up our mechanical circuits. The first is a Hookean spring. FIguRe 7. it is expedient and appropriate to use mechanical circuit models to represent the behavior of simple. differential operators and mechanical circuit equivalents are used to describe viscoelastic materials.1: Creep of a viscoelastic material. and the strain is monitored over time. .3 1-d LINeAR VISCoeLASTICITy (dIFFeReNTIAL FoRM BASed oN MeChANICAL CIRCuIT ModeLS) If one uses the definition that stress in a viscoelastic material depends both on strain and time derivatives of the strain.140 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 7. A constant strain is applied. 7. idealized viscoelastic materials. which is described by σ=η dε dt The dashpot contributes a viscous element to our viscoelastic material. An alternative definition is a material where the stress is a function of both the strain and time derivatives of the strain. A constant stress is applied. which is described by the equation σ = Eε The Hookean spring provides the elastic component of our viscoelastic material. We will describe three canonical spring and dashpot models. In this case. The second component is a dashpot.

the same path is not traced. stiffness increases with increasing strain rate. i.2) Differentiating Eq. . a spring and dashpot in series. Because the two components are in series.1) yields the constitutive equation for the Maxwell fluid ˙ ε= 1 1 ˙ σ+ σ E η (7.. (7. (7.4: Strain-rate dependence of material properties of a viscoelastic material.1) ε = εspring + εdashpot (7. These relationships can be written as σ = Eεspring = η and dεdashpot dt (7.2) and substituting in Eq. 7.VISCoeLASTICITy 141 FIguRe 7.e. the stress is the same in both the spring and the dashpot. This is known as the Maxwell fluid.1 Maxwell Fluid Let us consider the simplest mechanical circuit consisting of two components.3. In this graph. The loading and unloading curves are not identical.3) FIguRe 7.3: Hysteresis of stress–strain curve for a viscoelastic material subject to cyclic loading. and the strain is the sum of the individual strains for each element.

we have ε = ε0 H(t ) 1 so ε = ε0 . that in each case above. when subjected to a step strain.7) Thus.142 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Let us examine the Maxwell fluid during a creep test. Using this with Eq. (7.9) σ (t ) = Eε0 e−(E /η )t (7. the stress and strain are linearly related. (7. From Figure 7.3) we get (7. (7. Note. σ = σ 0 H (t) where H(t) is the Heaviside step function. s (7. simplifying. (7. by Eq.8) (7. we see that. For the stress relaxation response.1.g.9) is (7. the stress in a Maxwell fluid relaxes in an exponential fashion in time back to zero. What about stress relaxation? For stress relaxation (see Figure 7. Substituting this into Eq. (7. where 1 L{ } denotes the Laplace transform)..11) = Y(t ) = Ee−(E /η )t ε0 . σ = σ0 .6) is ε (t) = σ0 σ0 + t E η (7. s Eε0 σ= E s+ η The inverse Laplace transform of Eq.5) and rearranging. in a creep test.2).10) Thus. σ = L{σ } . we can write σ (t ) (7. Taking the Laplace transform of Eq. and rearranging yields ε= σ0 E � � � � 1 σ0 1 + s η s2 (7.6) The inverse Laplace transform of Eq.4) sε = 1 1 sσ + σ E η (7. the Maxwell fluid will deform unboundedly in a manner linearly dependent on time. Now.5).5) where the barred over variable is the Laplace transform of the variable (e. when subjected to a step load creep test. with a time constant equal to h/E.4).

s(t ¢) = s0. While the stress relaxation response of the Maxwell fluid is reasonable and approximates most relaxation behavior seen in viscoelastic materials. and the strain is given by e(t) = s0 J(t).5: Stress and strain versus time for a Maxwell fluid subjected to a standard viscoelastic test. These behaviors are why this model is known as the Maxwell fluid. continued deformation under a load and complete stress relaxation (“fluid cannot sustain shear stress”). We know that at t = t ¢. for the creep response. where t ¢ is the time at which stress relaxation is initiated. The relaxation phase is more difficult. we find ε (t ) 1 1 = J(t ) = + t σ0 E η (7. FIguRe 7.5). let us look at what would happen in a so-called standard test where we impose a creep phase followed by a stress relaxation phase (see Figure 7. . the creep response is linear in time. The stress is constant at a value s0. The creep phase is easy. Finally. We define a new variable. which is in contrast to the curves normally seen from creep test data.12) where J(t) is known as the creep compliance.VISCoeLASTICITy 143 where Y(t) is known as the relaxation modulus. Similarly. namely.t ¢. t = t . We let e(t ¢ ) º e1.

We can accomplish this with the Heaviside function shifted in time by t ¢. only the conditions have changed. The stress relaxes back to zero in an exponential fashion based on the relaxation time constant. we want the strain to remain constant at e1. σ (τ ) = σ0 e(−E/η )τ ⇒ σ (t) = σ0 e(−E η )(t−t ) .t ¢. we see that for a step stress (i.e.5. In terms of the new variable. 7. So.13) � In Figure 7. Let us now solve the problem for times t ³ t ¢.3. Furthermore. which implies t ³ 0. for the standard test of a Maxwell fluid we have ⎧ ⎧ ⎨ ⎨ 1 + 1 t creep σ0 creep. for stress relaxation. Mathematically. there is an initial jump in strain followed by strain increasing linearly with time. and the stress is the sum of the individual stresses in each element. these conditions are s(0) = s0 and e(0) º e1. e(t) = e1H (t .2 Kelvin–Voigt Solid The next canonical model is the Kelvin–Voigt solid. ε1 We have ε (τ ) = ε1 H(τ ) ⇒ ε = . s0 held constant). the spring and dashpot have the same strain. The Kelvin–Voigt solid consists of a spring and a dashpot arranged in parallel. from Eq. while allowing the stress to relax. where we have retained the initial conditions that result when E E η taking the Laplace transform of the first derivative of a function.144 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where e1 is the last value of strain that was reached during the creep phase. σ0 = σ= σ s+ 1 0 E E η Eη s E η� � � � �E � � � � � Eη 1 1 η 1 sη + E 1 σ0 = σ0 = σ0 . (7. t = t . We then stop the creep phase and hold the current level of strain. or σ = s+ sη + E E sη + E s − (−E/η ) E η Eη Taking the inverse Laplace transform.. Thus. Because the components are in parallel. This is written as ε= σspring E and σdashpot dε = dt η (7. we have 1 1 1 sε − ε (0) = sσ − σ (0) + σ . 0 ≤ t < t � σ (t) = and ε (t) = E η ⎩ σ e(−E/η )(vt −t � stress relaxation.15) . The constitutive equation is the same. σ= σ .14) σ = σspring + σdashpot (7.3). s ε1 − ε = 1 sσ − 1 σ + 1 σ . t ≥ t � ⎩ 0 ε1 stress relaxation (7.t ¢) = e1H (t) turns on the strain e1 at t = t ¢. Re-arranging. Making substitutions into the Laplace transformed � � � � s 1 sη + E 1 1 constitutive equation.

17) J (t) = � 1� 1 − e−(E /η )t E (7. (7.. However. . as opposed to instantaneously as would be for an elastic solid => delayed elasticity) and stress relaxation to a residual value. the relaxation response consists of an impulse followed by a constant plateau.14) gives us the constitutive equation for the Kelvin–Voigt solid ˙ σ = Eε + η ε (7. This arrangement is known as the Maxwell form.18) The creep response of the Kelvin–Voigt solid makes sense. suggesting instantaneous relaxation (which is typically not observed in solids. Arranging a Kelvin–Voigt solid in series with a spring. though possible in fluids). These are Y(t ) = E + ηδ (t ) and (7. i.6). It consists of a spring and dashpot in series.16) Following the same methodology used for the Maxwell fluid.VISCoeLASTICITy 145 Substituting Eq.3. the Maxwell fluid.3 Standard Linear Solid One of the most widely used models in linear viscoelasticity is the standard linear solid. These behaviors are why this model is described as a solid. we can find the stress relaxation and creep responses for the Kelvin–Voigt solid. called the Voigt form of the standard linear solid. yields equivalent behavior (see problem 3). in parallel with another spring (see Figure 7. FIguRe 7.13) into (7.e. 7. limited strain under a finite stress (though the final value is approached gradually. namely.6: Standard linear solid (Maxwell form).

two Kelvin–Voigt solid elements in series yield the four-parameter solid. In addition.e. They are Y(t ) = E2 + E1 e−t /τε and (7. Instead. In invariant form. or the relaxed modulus.22) The standard linear solid model captures behavior typical of many viscoelastic materials. We see that. For example. it can be shown that the constitutive equation for the standard linear solid is � � ˙ ˙ T � + τε T � = ER E � + τσ E � ∼ ∼ ∼ ∼ (7. whereas a Voigt form . the same techniques used to find the constitutive equations for the Maxwell fluid and Kelvin–Voigt solid can be applied here. E2. i. particularly ones of biological interest. the strain response approaches a bounded value. are used. the prime indicates that only the deviatoric stress and strain are being considered. 7.19) The parameters E1.4 Beyond the Canonical Models By adding additional elements. Using Laplace transforms. The constant E2 is usually referred to as ER. These time constants are given as η τε = E1 η (E1 + E2 ) τσ = (7. one for stress relaxation (te ) and one for creep (t s ). the stress approaches a residual value.3. the contributions of dilatational effects are neglected. The results yield the following constitutive equation (see problem 2) ˙ (E1 + E2 )ε + E1 E2 E1 ˙ ε= σ +σ η η (7. other material constants based on these three parameters are used.20) E1 E2 Note that the creep time constant is larger than the relaxation time constant. we can create models with four or more parameters.21) J(t ) = 1 E1 e−t /τσ − E2 E2 (E1 + E2 ) (7. and during stress relaxation.146 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Again.. and h are typically not reported. under a step load. one can solve for the relaxation modulus and creep compliance. two time constants.23) In this case.

Let us further demonstrate this by analyzing the Wiechert model. Eq. So. where E0 is the stiffness of the lone spring. Eq. The constitutive equation for the ith Maxwell 1 1 ˙ ˙ element is.25) Now the strain in a parallel arrangement is equal in each element. from Eq. yield the Wiechert model.26) is the associated viscoelastic constitutive equation for the Wiechert model. Laplace transforms are very useful in problems based on the differential form of viscoelasticity. (7.26) . sspring = E0e.23) is σ total = σ spring + ∑ σ i i=1 n (7. (7.27) The Laplace plane representation of the time domain constitutive equation is known as the associated viscoelastic constitutive equation. σ spring = E0 ε . we get sε = σ i + σ i = Ei ηi Ei ηi can rewrite this as where τi = ηi Ei σi = � s Ei sEi �ε = � �ε 1 1 s+ τ +η i i s (7. Lastly.24) becomes � ε = ⎝E0 + ∑ � �⎠ε σ total = E0 ε + ∑ � 1 1 i=1 s + τ i=1 s + τ i i n sEi ⎛ n sEi ⎞ (7. Demonstration. and all in parallel with a single spring. What is the stress response in a Wiechert model containing five Maxwell elements? . Hence. Because stress sums for elements placed in parallel.24) where si is the stress in the ith Maxwell fluid element. n Maxwell fluid elements in parallel with each other. Given a strain profile. (7. As we have seen. Taking the Laplace trans� � Ei ηi s s 1 1 ˙ + σ i .3). (7. the total stress in the Wiechert model is σtotal = σspring + ∑ σi i=1 n (7. (7. and noting that s(t = 0) = e (t = 0) = 0 . Making appropriate substitutions.26) to solve for stress as a function of time. The Laplace transform of Eq. We form. A ligament or tendon is pulled in tension at a constant rate of strain. ε = σi + σi (no summation on i here).VISCoeLASTICITy 147 standard linear solid in series with a dashpot yields a four-parameter fluid. we can take the inverse Laplace transform of Eq.

(7.29) where Y(t) is the relaxation modulus. The Boltzmann superposition principle states that the effect of multiple causes is the sum of the effects of each individual cause. and the creep compliance is a function of time only. Similarly. (7. (7.4 1-d LINeAR VISCoeLASTICITy (INTegRAL FoRMuLATIoN) Our other definition of a viscoelastic material was when the stress depends on the strain history of the material. the stress can be written as (7. To find the behavior of a linear viscoelastic material under an arbitrary stress or strain history. Taking the inverse Laplace with five elements. we must invoke the Boltzmann superposition principle. consider an experiment where a viscoelastic material is instantaneously deformed to some strain and then instantaneously returned to its undeformed state after some arbitrary span of time. As we have seen. The strain function may be written as a superposition of two step functions ε (t) = ε0 [H(t) − H(t − t1 )] Using the Boltzmann superposition principle and Eq. the strain is directly related to the stress. Substituting. These relationships for creep and stress relaxation assume that a step stress or strain is applied. creep in a linear viscoelastic material can be described by ε (t) = σ0 H(t )J(t ) (7.28) where J(t) is the creep compliance.26) becomes σ total = 2 + ∑ � 1 s i=1 s s + transform yields σ (t) = (E0 R)t + ∑ (Ei R)τi (1 − e−(t/τi ) ) . and s2 5 Ei R E0 R � . ε = . the stress relaxation behavior of a linear material can be written as σ (t ) = ε 0 H(t)Y(t ) (7. stress and strain are linearly related. As an example. and the relaxation modulus is a function of time only. Thus. For a linear material. Again.31) .30) σ (t ) = ε 0 [Y(t ) − Y(t − t1 )] This can be extended to any arbitrary strain history e(t) as in Figure 7. Eq.148 INTRoduCTIoN To CoNTINuuM BIoMeChANICS R Solution: A constant rate of strain is described by e = Rt.28).7. See reference [16] for an example of the Wiechert model in experimental biomechanics. i=1 5 τi 7.

32) is the incremental change in stress brought about by an infinitesimal strain pulse in the past. we obtain the equation σ (t ) = − �t 0 ε (τ ) dY(t − τ ) dτ + Y(0)ε (t) dτ . (7.7: Arbitrary strain history formed from an infinite sum of step functions. (7.. ε (t) = ε (τ ) [H(t − τ ) − H(t − τ + Δτ )] Using our relationship between stress and strain for stress relaxation under a step strain. as well as the Boltzmann superposition principle. the principle of causality).e.32) dY(t − τ ) Y(t − τ + Δτ ) − Y(t − τ ) = lim Δτ →0 dτ Δτ one can rewrite Eq. If we integrate over time to include all pulses up to the present time t (i. We will deal with a particular interval of this history. we can write the incremental change in stress due to this strain “pulse” as dσ (t ) = ε (τ ) [Y(t − τ ) − Y(t − τ + Δτ )] Given the definition of the derivative. (7. from time t - t to t - t + t.VISCoeLASTICITy 149 FIguRe 7.33) Eq.31) as dσ (t ) = −ε (τ ) dY(t − τ ) dτ dτ (7. which can be written as the difference of two step functions.

(7.. the stress has not had time to relax. It represents the contribution of all strains experienced from t = 0 to the present.150 INTRoduCTIoN To CoNTINuuM BIoMeChANICS The second term. not created. Eq. most notably the restriction that the rate of energy dissipation is nonnegative. Tij (t) = �t 0 dEkk λ (t − τ )δij dτ + dτ �t 0 2μ (t − τ ) dEij dτ dτ (7. i. The right-hand side of the equation is often referred to as a hereditary.34) which is known as the Boltzmann superposition integral. providing the general constitutive equation Tij (t ) = �t 0 Cijkl (t − τ ) dEkl dτ dτ (7. t = t. the final strain is generally not zero.e.35) reduces to a form analogous to the isotropic linear elastic solid. integral. i. or historical.33) and (7. Y(0)e(t).35) Based upon Eqs. for our arbitrary strain history. The increment of stress resulting from the last step is then given by the second term..5 3-d LINeAR VISCoeLASTICITy The arguments used in the 1-D integral formulation case can be easily extrapolated to the general 3-D problem. any number of functions could potentially satisfy the general creep and stress relaxation relationships for a linear viscoelastic material. at the last step. Thus.e.34). arises because. these functions are not arbitrary. energy is only conserved (stored) or dissipated. and there are limitations on the creep and stress relaxation functions.36) Making the assumption of isotropy. Using the same arguments. 7. However.37) . (7. The above integral can be cast in the following form σ (t ) = �t 0 Y(t − τ ) dε (τ ) dτ dτ (7. �t 0 ε (t ) = J(t − τ ) dσ (τ ) dτ dτ (7. Restrictions are placed upon these functions using fundamental physical principles. one can arrive at a similar integral expressing strain as a function of stress.

and E Y is the Young’s modulus of the material. finding the solution to the viscoelastic counterpart of the elastic problem with the correspondence principle is accomplished as follows: 1. we can rewrite Eq. (7. however. The correspondence principle is restricted to problems where the boundaries under prescribed loads and displacements do not change with time. (4. or problems with a constant Poisson’s ratio. L0 is the original length of the bar. the solution to the corresponding problem in linear viscoelasticity can be obtained by substituting for the parameters in the linear elasticity solution that can depend on time. whereas L0 is prescribed. identify which variables can depend on time.53). 2.37).40) F (t) = Y(t ) L0 . we obtain the equation Δl A (7. Recall the example for simple extension of a bar that we previously discussed for linear elasticity. we assume small deformation so that A remains approximately constant. the solution to the problem. we invoke the correspondence principle. Dl is the change in length of the bar. (7.37) as Δl FL0 = s sYA (7. Using the correspondence principle (as described above) and Laplace transforms.6 BouNdARy VALue PRoBLeMS ANd The CoRReSPoNdeNCe PRINCIPLe A boundary value problem is a problem which has values assigned on the physical boundary of the domain in which the problem is specified. The correspondence principle states that if a solution of a problem in linear elasticity is known. a process that generally yields a solution with more ease than from first principles. was found to be ε= Δl F = L0 EY A (7. take the Laplace transform of the solu_ tion and substitute sY for E Y . take the inverse Laplace transform to get the viscoelastic solution. substitute for Y(t) or J(t) from the particular viscoelastic model being used. For 1-D problems. To solve boundary value problems in linear viscoelasticity. In that example.39) The term Dl/s arises from the fact that this is a step deformation. In Eq. e and F can depend on time. Eq.38) where F is the applied force. and 4. 3. Let us look at this process in practice. Rearranging the above equation and taking the inverse Laplace transform. Furthermore. and the Laplace transform for the step function H(t) is 1/s. A is the cross-sectional area of the bar.VISCoeLASTICITy 151 *7. the loads and displacements themselves may be time-dependent.

we are interested in the instantaneous and equilibrium behavior of the cell. For further information on the correspondence principle.41) Note that Y(t) for a Maxwell fluid. we direct the reader to the references. i. and incompressible. (7.. particularly its application in multi-dimensional problems.39) yields the final solution for stress relaxation.75 kPa and E2 = 1. The cell is stretched to 10% of its undeformed length (Dl = 5 mm). (7. Kelvin–Voigt solid. F (0) = π a 2 Δl (E2 + E1 ) = 100 nN L0 π a 2 Δl E2 = 45 nN L0 F (∞) = Plugging in the appropriate values for a and L0 and solving for the elastic constants yields E1 = 1. or other model could be used if more appropriate for the material of interest.152 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where Y(t) is the relaxation modulus. F (t ) = � Δl A � E2 + E1 e−t /τε L0 (7. what would its Young’s modulus be immediately after loading and after the cell has come to a quasi-resting state? In this case. Typically.e. if the cell were an elastic material. Consider a tendon fibroblast (tenocyte) held at each end by a micropipette.43 kPa.18 kPa Equilibrium (relaxed) modulus = E2 = ER = 1. Let us assume that the shape of the cell approximates a circular cylinder with radius a and length L0. If we consider the viscoelastic material to be a standard linear solid. This has been only a brief introduction to the correspondence principle. Let us assume that we are testing a tendon fibroblast with radius a = 10 mm and L0 = 50 mm. Demonstration. isotropic. The cell is considered to be a linear viscoelastic solid (modeled with the constitutive equation for a standard linear solid) that is homogeneous. the properties would be reported as Instantaneous modulus = E0 = E1 + E2 = 3. substituting Eq.20) into Eq.43 kPa . The instantaneous load and load at equilibrium are measured and found to be F(0) = 100 nN F(t → ∞) = 45 nN What are the elastic constants E1 and E2? Solution.

7..42) to explicitly converge. we can obtain the relation σ (t ) = ε0 eiω t ⎣Ee + ω ⎡ ∞ � 0 ˜ E(t � ) sin ω t � dt � + iω ∞ � 0 ˜ E(t � ) cos ω t � dt � ⎦ ⎤ (7. We will endeavor to develop the theoretical framework for analyzing viscoelastic materials under dynamic loads. (7. and h) that best fit the experimental data (see problem 15).7 dyNAMIC BehAVIoR oF VISCoeLASTIC MATeRIALS Dynamic or oscillatory testing of viscoelastic materials is a common testing modality.VISCoeLASTICITy 153 In actual experiments. Ee = lim E(t). σ (t ) = −∞ �t E(t − τ ) dε dτ dτ (7.45) . Curve fitting algorithms would be used to obtain the parameters (E1.42). rather than focusing solely on the instantaneous and equilibrium responses of the cell. we must decompose the relaxation function E(t) into ˜ E(t ) = E(t ) + Ee (7.e. We rewrite the Boltzmann superposition integral with a lower limit of −∞ (as sinusoidal functions have no starting point). E2. The specimen is subjected to an alternating stimulus while simultaneously measuring the output.41).44) where Ee is the equilibrium modulus. (7. (7. a given solution for the creep or stress relaxation behavior would be fit to all of the available data. and t→∞ (7.42) where w is the angular frequency (in radians per second). Combining Eqs.43) σ (t ) = Ee ε 0 eiω t + iωε 0 −∞ �t ˜ E(t − τ )eiωτ dτ If we substitute t’ = t - t and rewrite the equation using sine and cosine functions.43) To allow for Eq. We can write a sinusoidal strain history as ε (t ) = ε0 eiω t (7. i.

t).48) tan δ = E �� (ω ) E � (ω ) (7.47) that the stress is sinusoidal. Another mechanical property often cited is the dynamic modulus.47) The resultant stress–strain relation can then be written as One can see from the form of Eq. (7.48) is referred to as the loss tangent. However. given by � � |E ∗ (ω )| ≡ �E � (ω ) + iE �� (ω )� (7.51) There is another way to look at dynamic viscoelasticity.52) where periodic loading is considered. the storage modulus E � (ω ) ≡ Ee + ω and the loss modulus ∞ � 0 ˜ E(t � ) sin ω t � dt � (7. The phase angle.154 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Let us now define two frequency-dependent mechanical properties. we can write the stress–strain relationship in response to a sinusoidal strain as σ (t) = |E ∗ (ω )| ε0 ei(ω t+δ ) (7. between the stress and strain functions can be found by � � σ (t ) = E � (ω ) + iE �� (ω ) ε (t ) (7.t) (7. the stress and strain are not in phase. A linear relationship exists between stress and strain of the form σ (ω .t) = E ∗ (iω )ε (ω .53) . d. and thus s and e are harmonic functions of (w.49) Eq.50) Using the dynamic modulus and loss tangent.46) E (ω ) ≡ ω �� ∞ � 0 ˜ E(t � ) cos ω t � dt � (7. (7. The modulus is given by E ∗ (iω ) = E � (ω ) + iE �� (ω ) (7.

(7. (7.61) .57) into (7. (7.54) σ (t ) = σ0 sin (ω t + δ ) where w is the angular frequency.55) σ (t ) = σ0 sin (ω t) cos (δ ) + σ0 cos (ω t) sin (δ ) Introducing moduli E ¢ and E ² defined as (7.57) tan (δ ) = Eqs.54) results in E �� E� (7. Eq.54) can be rewritten as (7. ε (t) = ε0 sin (ω t) with (7. If we use a complex representation of the input and output. (7.58) says that the s –e relationship is defined by E ¢ in phase with e and E ² which is 90° out of phase with e. σ (t ) = σ0 ei(ω t+δ ) then and ε (t) = ε0 eiω t (7.58) σ (t ) = ε0 E � sin (ω t) + ε0 E �� cos (ω t) (7.56) E� = Eq. be out of phase. in general. Because of the effect of delayed elasticity and viscous flow.60) E∗ = σ σ0 iδ σ0 = e = (cos δ + i sin δ ) = E � + iE �� ε ε0 ε0 (7.56) implies σ0 σ0 cos (δ ) and E �� = sin (δ ) ε0 ε0 (7. Let us consider a viscoelastic material under a sinusoidal strain.59) Eq.VISCoeLASTICITy 155 where E*(iw) is the complex modulus of the material. s and e will.55)–(7. and d is the phase lag.

8 LIMITINg CASeS oF LINeAR VISCoeLASTICITy ARe The hooKeAN SoLId ANd NewToNIAN VISCouS FLuId We would like to take a brief moment to elucidate how linear viscoelasticity relates to linear elasticity and Newtonian viscous fluids.64) as a complex number (i. E ¢ is the storage modulus (in phase with e) associated with the energy stored in the specimen due to applied e.63) into Eq. and loss tangent for a Maxwell fluid.7.156 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where E* is the complex modulus. loss modulus.62) h . From our discussion of linear elasticity. E ² is the loss modulus (out of phase with e) describing the dissipation of energy due to internal friction.1 gives the dynamic viscoelastic properties of the Maxwell fluid and Maxwell form of the standard linear solid. (7. Eqs. 7. and tan d is the loss tangent.64) Now. We now let E σ (t ) = σ0 eiω t (7.e. Table 7. 7.62) and (7.. a + ib) yields the storage modulus.63) Rewriting Eq. we can write the constitutive equation for a Maxwell fluid as ˙ ˙ σ + λ σ = Eλ ε where l = (7. (7.1 dynamic Maxwell Fluid From Eq. (7.3). we know there is no time .65) Writing the right-hand side of Eq. Problem 12 is reserved for the Kelvin–Voigt solid.61) yield iωλ E + iE = E 1 + iωλ � �� � � (7.60) as s = e (E ¢ + iE ² ) yields ε= σ σ0 eiω t = (E � + iE �� ) (E � + iE �� ) (7. (7. (7.

to arrive at linear elasticity from linear viscoelasticity. Eq. e¢ LoSS ModuLuS.e² LoSS TANgeNT. (7. (7. at each time t.66) λ δij Ekk |t 0 �t + 2μ Eij �0 = λ δij Ekk (t ) + 2μ Eij (t ) where we have assumed an unstrained state at t = 0. we set the relaxation functions equal to ld(t - t) and 2md (t - t). To arrive at a Newtonian viscous fluid.30).67) dEij (t ) dEkk (t ) = λ δij + 2μ = λ δij Dkk (t ) + 2μ Dij (t ) dt dt . Then. Thus. Eq. (4.VISCoeLASTICITy 157 TABLe 7. Thus. the constitutive equation is the same as for linear elasticity.36) becomes �t 0 Tij (t ) = λ δij dEkk δ (t − τ ) dτ + 2 μ dτ �t 0 δ (t − τ ) dEij dτ dτ (7.36) becomes �t 0 Tij (t ) = λ δij = dEkk dτ + 2 μ dτ �t 0 dEij dτ = λ δij dτ �t 0 dEkk + 2μ �t 0 dEij (7.1: Dynamic viscoelastic properties of the Maxwell fluid and Maxwell standard linear solid VISCoeLASTIC ModeL SToRAge ModuLuS. Eq. we make the relaxation functions independent of time. Thus. where d(t - t) is the Dirac delta function. TAN d Maxwell fluid E � ω 2λ 2 1 + ω 2λ 2 � E � ωλ 1 + ω 2λ 2 E2 1 η � � 1 ωλ Standard linear solid E1 ω 2 E2 + 2 E1 + ω2 η2 � ω E2 1 + ω2 η2 ωη E2 1 E2 E2 + ω 2 η 2 (E1 + E2 ) 1 dependence in the constitutive equations.

(5. Based on the results of the creep test. we recover the viscous part of the constitutive equation for a Newtonian fluid. 4. i.) The constitutive equation for the Maxwell form of the standard linear solid. The springs and dashpots introduced in this chapter have analogs in circuit theory. E1 E2 These two expressions are similar to capacitors in an electrical circuit. For further discussion of viscoelasticity. (c) Suppose the strain or stress history is arbitrary. . not a step displacement or load. it should be noted that the results of this chapter apply to infinitesimal deformations. (a) Derive the constitutive equation for the standard linear solid. we refer the interested reader to references [19-21]. · · · · can be written as s + C3s = C1e + C2e. 5. such as Y.158 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where we have used the fact that the time derivative of the strain tensor is the strain rate tensor under infinitesimal displacements [18].C. how does this model represent fluid behavior? Solve for stress relaxation of the Wiechert model when subjected to a step strain.17).18). Eq. (7. (b) Choose one of the following: (i) Calculate Y(t) or (ii) calculate J(t) for the standard linear solid subjected to a step strain or step stress.. Finally. Note that. = (b) Show that two springs in series have an equivalent stiffness given by 1 Eequiv 1 1 + . 2. What is the relaxation modulus? This particular relaxation modulus is known as (a) Show that the equivalent stiffness of two springs in parallel is Eequiv = E1 + E2. e = e0H(t). though the lumped parameters are a collection of different constants.” Show that the Voigt form of the standard linear solid (a Kelvin–Voigt solid in series with a spring) can be written the same way. though the lumped parameters will be a collection of different constants. where the Ci ’s are “lumped parameters. Eq. yielding identical material behavior! Show that the three-parameter fluid (a Kelvin–Voigt solid in series with a dashpot or a Maxwell fluid element in parallel with a dashpot) has the constitutive equation of the form · · ·· s + C3s = C1e + C2e. 7.9 PRoBLeMS 1. at each time t. Eq. In this problem you will more formally investigate the Maxwell form of the standard linear solid model. respectively. What are the Ci ’s in terms of the E and h’s? Obtain the strain as a function of time during a creep test. the governing equations are the same. Thus. should be employed when finite strains prevail.4.18). Fung’s QLV (quasi-linear viscoelasticity) theory [14] and hyperviscoelasticity [7].e. Extensions of this theory. (7. 3. How would you calculate the stress or strain as a function of time for the arbitrary history? (Hint: Reread Section 7.

elastic modulus E Y . and a step load (F = F0H(t)) is applied perpendicular to the beam at its midspan. d(t). . The elastic solution for the deflection of a beam of length L. respectively. plot d (t) versus t.t) J (t)dt = t. The beam is made of a material that can be adequately modeled as a Kelvin–Voigt solid. A beam is simply supported at both ends. (a) Assuming that n = 0. and d is the indentation depth. hydrated tissue. find the deflection of a standard linear solid beam. find the indentation depth as a function of time. The material is modeled as a semi-infinite elastic solid indented with a circular cylindrical punch. Show that òo J(t . q(t). (Hint: What is the relationship of J(t) and Y(t) in the Laplace domain?) 7. the material is incompressible).VISCoeLASTICITy 159 the “Prony series” expansion.5 (i. Plot the displacement versus time at the point of max deflection (x2max).) 9. The elastic solution for this problem relating applied force to displacement is F= EY (2a) d 1 − ν2 where F is the applied force.t) Y (t)dt = òo Y(t . What is the response of the Wiechert model to an arbitrary strain history? 6. where J(t) and Y(t) are the creep compliance and relaxation moduli. under pure bending. (Hint: Recall J(s)Y(s)=1/s 2) (b) Using your solution from part (a). and cross-sectional moment of inertia I subject to this loading is t t x2 (x1 ) = � F � 3 4x1 − 3L2 x1 48EY I What about the graph indicates “solid-like” behavior? (Hint: Use an intermediate relationship you get in problem 6. Using the correspondence principle. Write the 3-D integral form of the constitutive equation for (a) An orthotropic material (b) A transversely isotropic material (Hint: Recall Section 4.. for a step load F(t)=F0 H(t) applied to a standard linear solid.e. Consider indentation of a soft. a is the radius of the indenter. Evaluate your expression for an applied moment of M(t) = ct. 10. What is the displacement immediately upon loading? At equilibrium? See reference[21] for an example of viscoelastic testing of cells and the correspondence principle.11) 8.

Consider a sinusoidal stress given by s (t) = s0sin(wt). and loss tangent (d) for the Kelvin–Voigt solid. In this problem. dt V (b) Evaluate the energy per volume for a single loading cycle. W =òF × dl . How does the material behave at high frequencies? 12. but not by out-of-phase strain..160 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 11.e.8 to plot the behavior of the Kelvin–Voigt and standard linear solids when subjected to the standard test described in the section on the Maxwell fluid. ® ® FIguRe 7. 13. (a) Verify the storage modulus (E ¢).e. per volume can be written W = òs d e dt. Show that the work. to show that energy is conserved by in-phase strain. loss modulus (E ²). Comment on the instantaneous and equilibrium behaviors. what is the in-phase strain? Out-of-phase strain? From basic physics. the work done is equal to the integral of force dot product with displacement.8: Stress and strain versus time plots for use with problem 14. In the tenocyte demonstration. we found the instantaneous and relaxation modulus during a stress relaxation experiment. MatlabÓ’s “cftool” command is one possible method for doing this. 15. Use the following data to find the viscosity. of this tenocyte. integrate from 0 to 2p /w. . 14. i. (7. i. This result is why viscoelastic materials exhibit hysteresis. loss modulus (E ²). The out-of-phase energy is lost to the environment as heat. Use the graph shown in Figure 7.. we will look at energy dissipation during cyclic loading of a viscoelastic material. and loss tangent (d ) given in Table 7.1. (b) Sketch a plot of the loss tangent versus frequency. equivalent to energy through the work-energy principle. A proteoglycan solution can be modeled as a Maxwell fluid. (a) Using Eq.58). Find the storage modulus (E ¢). h.

4 75.5 6.3 45.8 46.95 45 .75 13 16.9 45 44.5 91.8 84.9 45.7 68.3 2 2.5 4.VISCoeLASTICITy 161 TIMe (s) FoRCe (nN) 0 0.65 1.9 60.15 45 44.5 87.4 51.8 55.25 25 40 55 80 90 155 220 250 (» ¥) • • • • 100 95.5 9.

.

Thus. such that a change in temperature can cause a change in stress or vice versa. we can consider what happens to the fluid pressure in a tissue when it is loaded or how changing the temperature can induce strain. For example.1 Terzaghi’s Principle of effective Stress Poroelasticity is the study of the behavior of fluid-saturated elastic porous media. the strain field inside a material affects the fluid pressure and solid stress.g. poroelasticity and thermoelasticity..1 INTRoduCTIoN This chapter is concerned with coupled theories. A porous material is a solid containing an interconnected network of pores (voids) filled with a fluid (liquid or gas).2. The solid matrix and the pore fluid are assumed continuous. Poroelasticity has been used to describe several biological tissues.e.. have been described with poroelastic theory. a change in fluid pressure can effect a change in solid stress. in particular.. both naturally occurring (i. such as bone. Mathematically (recall we define the normal stress as –p in a fluid). foams and ceramics). For example. soils.2 PoRoeLASTICITy 8. 8. Materials. who developed it while performing pioneering work on the 1-D consolidation of soils. the advancement comes at the expense of somewhat more complicated mathematics. a sponge). cartilage. the temperature field and stress field affect one another.1) . the principle of effective stress states that the total stress at a given point in the material is the sum of the solid stress and fluid pressure. while thermoelasticity has been more widely used with biomaterials. the principle of effective stress is ascribed to Terzaghi. T = T � − pI ∼ ∼ ∼ (8. biological tissues) and synthetic (e. forming two interpenetrating continua (e.g.163 CHAPTER 8 *Poroelasticity and Thermoelasticity 8. These theories allow an expanded description of the material’s behavior under different conditions. A coupled theory is a theory in which field variables interact with each other. However. in poroelasticity. and the intervertebral and temporomandibular joint discs. In poroelasticity. rocks. As applied to poroelasticity. In thermoelasticity. A full fledged theory of poroelasticity was not given until 1941 by Biot [23].

k = k I .2. where k is a constant. we obtain the pressure contributing solely to flow.2) where the minus sign ensures that fluid flows “downhill.” i. as flow is not affected by the vertical pressure drop caused by gravity under hydrostatic conditions. . from high to low pressure. reflecting the fact that the permeability of the solid through ˜ which the fluid is flowing need not be the same in all directions. The fluid velocity is ν= � q f luid φ � (8. note that choosing T ¢ to be given by a hyperelastic or viscoelastic constitutive equation is also possible. 8.164 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where T is the total stress. The flux is divided by porosity to account for the Vtotal fact that only a fraction of the total volume is available for flow.. The tensor k is the anisotropic permeability tensor. Darcy’s law says that the fluid flux is directly proportional to the permeability of the material through which the fluid is flowing and the pressure gradient driving the flow. fluid velocity within the pores (i.e. Darcy’s law is valid for Newtonian liquids at low velocities. � q f luid = − 1 μf luid � � � κ ∇p − ρ g e 3 ∼ (8.e. gravitational effects can be neglected. It was empirically deduced by Darcy in the latter part of the 19th century during his experiments on the flow of water through sand. ˜ though the latter is called poroviscoelasticity. The theoretical analyses have shown that. It has since been theoretically derived [11].. T ¢ is the effective stress (i.4) Vfluid where f = is the porosity of the material.e.. while it is inversely proportional to the fluid’s viscosity. we see that when Ñp ~ DP / L >> rg. stress in the solid matrix).2 darcy’s Law Darcy’s law is a constitutive equation describing the flow of fluid through a porous medium. Furthermore. The matrix representation of k is ˜ ⎤ ⎡ κ11 κ12 κ13 ⎥ ⎢ [κ ] = ⎣ κ12 κ22 κ23 ⎦ (8. and p is the hy˜ ˜ drostatic pressure. Mathematically. Finally. though having limitations. By subtracting the gravitational pressure drop from the total pressure drop.3) ∼ κ13 κ23 κ33 ® If the material is isotropic with respect to permeability. Darcy’s law has often been employed in the study of the flow of biological fluids through tissues. The term rge 3 ˜ ˜ enters. Though we will restrict our attention in this chapter to linear poroelasticity. Thus. seepage velocity) is related to the flux by the porosity.

z. as it is the dimensionless ratio of two volumes. increment of fluid content is essentially the “fluid strain”. is the third. or Ñ × n = 0. There is no contribution of the pressure term to the shear stresses because.2) and (8. The velocity is ( ) x1 + ν= � 1 φ � P1 − P2 L � κ � e1 μf luid 8. the above expression becomes � � � � � � 1 1 1 κ κ ∇ · �f luid = ∇ · − q κ ∼ ∇p = − I ∇ · ∇p = − ∇2 p = 0 φ φ μf luid φ μf luid φ μf luid dp =0 dx1 P2 - P1 Solving the above equation subject to the pressure boundary conditions yields p = L P1. In this book. An incompressible fluid of viscosity μ is flowing through a porous medium of permeability k I and length L. and final.46). we present the mixed stiffness formulation. Increment of fluid content.2.6). What is the velocity of the fluid within the medium? Solution. in which case it would be a pure stiffness formulation) [24]. As the fluid is incompressible. Thus. as we have . The velocity comes from back substitution into Eqs. “Stiffness” refers to stress being one of the dependent variables.2) we get ∇·ν = � where the identity tensor was eliminated via Eq. one for each direction. From Eqs. Thus. (1. there are three independent ais. For stress in a linear anisotropic poroelastic continuum. It is equal to the volume of fluid added to or removed from a control volume. and mixed refers to using increment of fluid content as the other (as opposed to pressure. Let us align the direction of flow with the x1 axis and assume gravitational effects can be ® neglected. the continuity equation is given by Eq. new concept we will present with respect to poroelasticity. (8. while on the other ˜ side it is held at P2.5) Note: There is no summation on i in the last term. Tij = Cijkl Ekl − αi pδij (8.4) and (8.3 Constitutive equations and Material Constants Having discussed the principle of effective stress and Darcy’s law. (8. (5. Assuming the pressure depends only on x1. let us now turn to the constitutive equations of linear poroelasticity.PoRoeLASTICITy ANd TheRMoeLASTICITy 165 Demonstration.4). The pressure on one side of the medium is held at P1. There are several ways to formulate the constitutive equations of poroelasticity depending on the choice of independent variables one desires to use.

tending toward one for completely saturated materials (i. For reference. It is the ratio of fluid added to the change in bulk volume (recall Ekk is the dilatation).8) Skempton’s coefficient is a measure of load distribution between the solid and the fluid.e. It is given as � ∂ζ � K � α≡ = � ∂ Ekk p=constant H (8.. The latter is a measure of the change in increment of fluid content relative to a change in the average applied stress at constant pore pressure. as the solid framework must support the load.7) The coefficient a is known as the Biot–Willis coefficient. and 1/H is the poroelastic expansion coefficient. Alternatively. If we consider an isotropic material. In the mixed stiffness formulation. constrained means E = constant strain. highly compressible gases). linear poroelastic material is given by the familiar expression for stress in a linearly elastic solid. (8.. As it was for fluids. the increment of fluid content for isotropic materials is ζ = α Ekk + In these equations α p Ku B (8. undrained means z = no change in fluid content within a control volume. � � R ∂p � ∂ Ekk � � � B≡ − = = � � ∂ T ζ = constant ∂ ζ T = constant H (8. B. (8. load is shared by pores filled completely with fluid) and toward zero for void space pores (e.g. We will define a momentarily.166 INTRoduCTIoN To CoNTINuuM BIoMeChANICS learned. K is the drained bulk modulus (l + 2/3 μ). Skempton’s coefficient is the ratio of volumetric strain to increment of fluid content in unconstrained conditions.9) .6) and (8. and drained means p = pressure in the material equal to surrounding pressure at all times.1). we must also introduce the constitutive equation for the increment of fluid content. we had nine equations with nine unknowns (see Chapter 4).5) becomes Tij = 2μsolidEij + λ Ekk δij − α pδij (8. Furthermore. But. before we see how the system of equations comes together. we see that the effective stress in an isotropic. unconstrained means T = constant stress. fluids cannot support a shear stress. we have Skempton’s coefficient. the additional equation we will use is the continuity equation. In linear elasticity. Introduction of pressure into the constitutive equation leads to an additional unknown. It must be a number between zero and one. Next.6) Comparing Eqs. Eq.

the system passes through a sequence of equilibrium states.11) � ∂ p Ekk =constant M Ku B The specific storage coefficients measure the ratio of increment of fluid content to a change in pore pressure under different conditions. assume static equilibrium at each instant of time). (3. Combining a. a. Finally. Ku is the undrained bulk modulus. Ku. and descriptions thereof. there are four material coefficients needed to fully describe an isotropic poroelastic material. SE . and SE .21) and (8. There are many more poroelastic material coefficients. � ∂ζ � α 1 � SE ≡ = = (8. and B as done in Eq. however. (8. l. depending on the formulation of the constitutive equations used. � K ∂T� � (8.12) . m. Wave propagation is ignored in this quasi-static approximation. in general. Coupling of the equations comes from the pressure ( p) term in the equations of force equilibrium and from the volumetric strain (Ekk) in the fluid flow equation. � ∂ � 2μsolid Eij + λ Ekk δij − α pδij + ρ Bi = ρ ai ∂ xj (8. The system of equations is solved elastostatically (i. even in the absence of a traditional source). 1/R ® 0. and one increment of fluid content. Hence. Assuming the material properties have been measured. Thus. and one fluid continuity equation (uses Darcy’s Law and. Thus. Many also have micromechanical interpretations. this path is irreversible due to frictional drag loss as the fluid flows past the solid matrix. using a displacement–pressure formulation results in a system of four coupled equations (by substituting the strain–displacement. we need 11 equations: seven constitutive equations (six stress and one increment of fluid content). Darcy’s law. 8. as a large increase in fluid pressure will cause an insignificant increase in fluid volume.e.2.6). Using Eqs. three force equilibrium equations (one for each direction). and the constitutive relationships into the conservation equations) with the unknowns being three displacements and one pore pressure.PoRoeLASTICITy ANd TheRMoeLASTICITy 167 Furthermore.7) yields the constrained specific storage coefficient. one pore pressure. 1/R º ST is the unconstrained specific storage coefficient. For an incompressible fluid. becomes a nonhomogeneous “Heat” equation in pressure. As will be shown below. three displacement components (yielding six strains).4 u–p Formulation of Poroelastic governing equations Let us first look at the stress equilibrium equations.10) Ku ≡ = � ∂ E ζ =constant 1 − αB The bulk moduli have their familiar interpretation of the change in volume under hydrostatic pressure.. the unknowns in this system are 11: six stress components.

18) . In Eq. (8.14) Now. Eq. Ñ2. Eq. The ratio. adds 1/m2 to the dimensions such that the Laplacian of the pressure is N/m4.7) with constant material coefficients into Eq.15). Ñ. The gradient operator. The Laplacian. adds 1/m to the dimensions such that the divergence of the flux is 1/s.16) ∇p + Q = ∇ 2 p + Q = k∇ 2 p + Q = −∇ · − ∂t μf luid μf luid Recall that k is the intrinsic permeability of solid. the time rate of change in increment of fluid content in a control volume is equal to any source producing fluid in the control volume (Q) less the divergence of the fluid from the control volume.15) Note that a negative Q would correspond to a fluid sink. This expression shows the coupling between displacement and pressure. � � ∂ζ � � = −∇ · q f luid + Q x . time) and q is the flux vector (m/s). agreeing with the time derivative of the increment of fluid content. Taking the x3 direction as an example. neglecting gravity and with k and μfluid independent of spatial location. is the hydraulic conductivity and has units m4/N. Heuristically.17) This expression can be rearranged to � �� � Ku B ∂ Ekk ∂p Ku B 2 k∇ p + = Q−α ∂t α α ∂t (8. (8. into the fluid continuity equation to obtain � � ∂ζ κ κ (8.13) should remind you of Navier’s equations. We now substitute Darcy’s law. and μ is the fluid viscosity. we obtain μsolid ∂p ∂ 2 ui ∂ 2 uk + (λ + μsolid ) =α − ρ Bi 2 ∂ xk ∂ xi ∂ xk ∂ xi (8. (8. Q is a traditional source term ® having units of (volume fluid contributed/reference volume . we will examine the fluid continuity equation. Substitution of Eq.13) Recall that k is summed on in the expression ¶xk2. k / μfluid = k . (8.168 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Assuming quasi-static conditions.s.16) yields α α ∂p ∂ Ekk + = k∇ 2 p + Q ∂t Ku B ∂ t (8. Mathematically.13) is � � � � ∂p ∂ 2 u3 ∂ 2 u3 ∂ 2 u3 ∂ 2 u1 ∂ 2 u2 ∂ 2 u3 μsolid + + + + − ρ B3 + (λ + μsolid ) =α 2 2 2 2 ∂ x3 ∂ x1 ∂ x3 ∂ x2 ∂ x3 ∂ x1 ∂ x2 ∂ x3 ∂ x3 (8. (8. t ∂t (8.

Indeed. and Ku.17) becomes ∂ 2 ui ∂ 2 uk ∂p + (λ + μsolid ) = − ρ Bi 2 ∂ xi ∂ xk ∂ xi ∂ xk (8. 8. We refer the reader to reference [24] for more details.20) By Eq. (8. hence.e. (8.13) and (8. a. Our goal is to obtain the pressure and displacement fields within the finite layer when subjected to a load. The above equations. whereby the pore size changes as fluid is effluxed from the material due to pressure gradients. Under these assumptions. Eqs.. these assumptions mean a /KuB = SE = 0. (8. Eqs.18) should be recognized as the “Heat” equation in p with a source term. and p. for an infinitely incompressible fluid.19) ∂ Ekk = k∇ 2 p + Q ∂t (8. μsolid. Taken together. l. If we assume the solid matrix is intrinsically incompressible. compressive loading leads to relative motion between the solid and fluid. k. Here.5 Consolidation of a Finite Layer (Terzaghi’s Problem) As an example of poroelasticity. Eq. These assumptions leave only three material coefficients. Concomitantly. (8. u2. μsolid. We confine the finite layer of height L on five of six sides within a rectangular parallelepiped and apply a load through a permeable platen on the free surface. require six independent material coefficients.20) is the statement that the time rate of change of the dilatation is equal to the time rate of change of the increment of fluid content.16). This makes sense when considering both the solid and fluid incompressible. Eq. k. the solid network experiences a collapse in size as it comes to occupy the space previously occupied by the fluid.18). dilatation) of that space. (8.PoRoeLASTICITy ANd TheRMoeLASTICITy 169 Eq. are often employed in biomechanics.13) becomes μsolid and Eq. u3.2. . This completes the u–p formulation of poroelasticity. (8. B. u1. (8. They form a system of coupled PDEs that must be solved simultaneously subject to appropriate initial and boundary conditions on the displacements and pressure. These equations can be cast in many different forms. the source term consists of everything within the brackets. a = 1.18) represent four equations for the four unknowns. The assumptions of the incompressible nature of both the solid and fluid components are justified for most tissues and. Furthermore.13) and (8. as any movement of an incompressible fluid into a space must correspondingly increase the volume (i. we will investigate the consolidation of a finite layer consisting of a solid matrix inundated with a pore fluid. depending on the choice of independent variables. B = 1 and Ku = ¥. l. It is important to recognize that the assumption of incompressible solid and fluid components does not mean loading will not cause deformation of the tissue.

170

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

The above problem description amounts to the condition of uniaxial strain and, as we shall see, is a unique condition that uncouples the displacement and pressure variables. We will let x3 be the direction of nonzero strain. Thus, p and u3 are independent of x1 and x2 and u1 = u2 = 0. Given E11 = E22 = 0 and E33 = e, Eq. (8.6) yields

T33 = (λ + 2μ ) ε − α p

(8.21)

Where we have dropped the subscript ‘solid’ from m. Turning to Eq. (8.17) and assuming no source term (Q = 0),

SE

From Eq. (8.21),

∂ Ekk ∂ε ∂p ∂ 2p − k 2 = −α = −α ∂t ∂t ∂t ∂ x3

(8.22)

∂ε ∂ T33 α ∂p 1 = + ∂t (λ + 2μ ) ∂ t (λ + 2μ ) ∂ t

Substituting Eq. (8.23) into Eq. (8.22) and rearranging yields

(8.23)

SE or

∂p ∂ 2p α ∂ T33 α2 ∂p −k 2 =− − ∂t (λ + 2μ ) ∂ t (λ + 2μ ) ∂ t ∂ x3

(8.24)

**� �−1 � �−1 ∂p α2 ∂ 2p α α2 ∂ T33 − k SE + =− SE + 2 ∂t (λ + 2μ ) (λ + 2μ ) (λ + 2μ ) ∂t ∂ x3
**

Now, assuming no body forces and quasi-static loading, Eq. (3.21) for i = 3 becomes

∂ T31 ∂ T32 ∂ T33 ∂ T33 + + = =0 ∂ x1 ∂ x2 ∂ x3 ∂ x3

(8.25)

which shows that T33 is independent of x3 in this problem. Thus, we can replace the partial derivative of T33 with respect to time in Eq. (8.24) with the total time derivative. Thus, Eq. (8.24) can be written succinctly as

**∂p ∂ 2p dT33 − C1 2 = −C2 ∂t dt ∂ x3
**

where we have collected the material coefficients into the constants C1 and C2.

(8.26)

PoRoeLASTICITy ANd TheRMoeLASTICITy

171

Eq. (8.26) is a general result for the uniaxial consolidation problem. Let us consider the specific case when a step load is applied to the loading platen at t = 0, i.e., T33 = -(F/A)H(t). For all time dT greater than t = 0+, 33 = 0. Thus, Eq. (8.26) becomes dt

∂p ∂ 2p = C1 2 ∂t ∂ x3

The initial condition is

(8.27)

and the boundary conditions are

� � p x 3 , 0+ = p 0 p (0, t) = 0

(8.28)

and

(8.29)

∂p (L, t) = 0 ∂ x3

The first boundary condition reflects the fact that the platen is porous (i.e., free draining), and the second reflects the fact the permeability of the base is infinite (i.e., no flow; there is a zero pressure gradient). The solution to Eq. (8.27) subject to Eqs. (8.28) and (8.29) is

p (x3 ,t) = 2p0

** where ln = (2n - 1)p . 2 What about the displacement? Eqs. (8.25) and (8.21) imply
**

(λ + 2μ )

� x � C 1 −λ 2 1 t 3 e n L2 sin λn ∑ L n=1 λn

∞

(8.30)

∂p ∂ 2 u3 =α 2 ∂ x3 ∂ x3

(8.31)

Taking the derivative of Eq. (8.30) with respect to x3, substituting the result into Eq. (8.31), and integrating twice with respect to x3 yields

u3 (x3 ,t) = −2p0 L

�

α λ + 2μ

�

� x � −λ 2 C1 t 1 3 e n L2 + f (t )x3 + g(t) cos λn ∑ λ2 L n=1 n

∞

(8.32)

Thus, we need two boundary conditions to complete the solution. The first boundary condition is u3(L, t) = 0. This results in g(t) = - f (t)L. The second boundary condition comes from the stress condition and Eq. (8.21). We have

172

INTRoduCTIoN To CoNTINuuM BIoMeChANICS

T33 (0, t) =

−F ∂ u3 ∂ u3 −F (0, t) − α p(0, t) ⇒ (0, t) = = (λ + 2μ ) A ∂ x3 ∂ x3 A (λ + 2μ )

Applying this condition yields f (t) = -F . Thus, Eq. (8.32) becomes A(l + 2m)

u3 (x3 , t) =

� �∞ � x � C F α 1 −λ 2 1 t 3 e n L2 cos λn (L − x3 ) − 2p0 L ∑ λ2 A (λ + 2μ ) λ + 2μ n=1 n L

(8.33)

For an incompressible fluid and solid, a = 1 and SE = 0 so that C1 = k(l + 2μ). Also under these assumptions, p0 = F/A, as the initial stress is transferred immediately throughout the pore fluid. When the solid and fluid are considered incompressible, this problem is equivalent to confined compression of a biphasic material (see Chapter 9). For more about the equivalences between incompressible poroelasticity and the biphasic theory, see reference [25].

**8.3 TheRMoeLASTICITy 8.3.1 Introduction and Fourier’s Law
**

Thermoelasticity is concerned with how the temporal deformation of a body can induce temperature changes and vice versa. We present thermoelasticity in this chapter due to its mathematical equivalence to poroelasticity [24,26]. Thus, by becoming competent in the mathematics and principles of one theory, one becomes familiar with the other theory. However, note that the interpretation of the two theories is not equivalent. As we shall see shortly, stress in a thermoelastic material is borne, in part, by changes in a temperature field. For a poroelastic material, it is easy to visualize how the fluid can share the load. However, there is not as concrete a visualization for a temperature field. Without going too deep into the mechanisms behind thermoelastic coupling of stress and temperature, we offer the following greatly simplified explanation: Elementary physics allows us to idealize a solid as a lattice of atoms bonded together vibrating about an equilibrium position. Increased temperature increases the interatomic separation, thereby increasing the material’s size (i.e., inducing strain). This is the coupling of temperature and deformation (i.e., thermal strain), which, when a body is confined in some way (i.e., not able to freely expand), induces thermal stresses. Another fundamental difference between the two theories is that the temperature field can support a shear stress. Fourier’s law is the Darcy equivalent in thermoelasticity. Empirically we know that 1. heat does not “flow” under conditions of constant temperature, 2. heat flows from hot to cold, 3. the greater the difference in temperature, the greater the heat flow, and 4. heat flow depends on the material through which it is flowing. These observations were summarized mathematically (and demonstrated experimentally) by Fourier as

�

q heat = −κ ∇Θ = −κ ∇θ ∼ ∼

(8.34)

PoRoeLASTICITy ANd TheRMoeLASTICITy

173

where q is the heat flux vector (heat per time per unit area, i.e., heat/time = heat ), and k is area ˜ time ´ area the anisotropic thermal conductivity tensor. Q and q are defined below. The negative sign ensures heat “flows” from hot to cold.

**8.3.2 Constitutive and governing equations (“u–q” Formulation)
**

As it was for poroelasticity, there are several ways to formulate the constitutive equations of thermoelasticity depending on the choice of independent variables one desires to use. We will present the form that clarifies the analogy between the two theories. For further reading on thermoelasticity, see the references [27,28]. For stress in a linear anisotropic thermoelastic continuum,

**Tij = Cijkl Ekl − bij θ
**

If we consider an isotropic material with b = g I , Eq. (8.35) becomes ˜ ˜

(8.35)

Tij = 2μ Eij + λ Ekk δij − γθ δij

(8.36)

In thermoelasticity, the analogous concept to increment of fluid content, is entropy, S. For isotropic materials

S = γ Ekk +

cε θ θ0

(8.37)

where q0 is a constant reference temperature corresponding to the resting state of the material, i.e., undeformed and stress free. q is the deviation in the absolute temperature from the reference temperature, i.e., q = Q - q0. Furthermore, these equations refer to conditions where q does not deviate q far from the reference temperature, i.e., — << 1. q0 � ∂S � � γ≡ = (3λ + 2μ ) α = 3Kα (8.38) ∂ Ekk �θ =constant

The material coefficient g relates the change in entropy to the change in volume at constant temperature. The coefficient a is the coefficient of linear thermal expansion (see problem 5), and K is the bulk modulus of the solid (see chapter 4). The material coefficient ce is the specific heat at constant strain. Recall that the specific heat of a substance is the quantity of heat required to raise the temperature of a unit weight of a substance by one degree. It is related to the “normal” specific heat by ce = rc. There are other constants that can be described for thermoelasticity. Importantly, many of the constants have statistical mechanical foundations and interpretations.

(8. ∂h � � = −∇ · q heat + Q( x. for stress equilibrium.45) g q where h = 0 . Eqs. u1. Now. Eq.34) and Eq. � η ∂ Ekk Q − ∂t κ � = ∇2 θ − cε ∂ θ κ ∂t (8. Heuristically.43) Substituting Eqs.39) ∂ 2 ui ∂ 2 uk μ 2 + (λ + μ ) = ∂ xi ∂ xk ∂ xk (8. the source term consists k of everything within parentheses.174 INTRoduCTIoN To CoNTINuuM BIoMeChANICS Now.41) dS = Eq.42) into Eq.45) represent four equations for the four unknowns. Here.43) results in � � ∂ Ekk cε ∂ θ (8. This completes the “u–q ” formulation of thermoelasticity. and q.36) give Taking the derivative. the time rate of change in heat is equal to any heat source less the divergence of heat.t) ∂t (8.21) and (8. (8. (8. (8.44) θ0 γ + = κ ∇2 θ + Q ∂t θ0 ∂ t Rearranging. we will examine the heat (h) balance. Eqs. u3. u2. (8. They form a system of coupled PDEs that .42) θ0 ∂S � � = −∇ · q heat + Q( x. (8.45) is the Heat equation with a source term.37) with constant material coefficients into Eq. Mathematically.t) ∂t We now recall the thermodynamic relation (8. (3.40) This expression shows the coupling between displacement and temperature.41) gives dh ⇒ h = Sθ0 θ0 (8. we obtain � ∂ � 2μ Eij + λ Ekk δij − γθ δij + ρ Bi = ρ ai ∂ xj � � ∂θ ∂ 2 ui γ − ρ Bi + ρ 2 ∂ xi ∂t (8. Reference [12] contains a good discussion of heat balance and the heat equation.40) and (8. (8.

The second case leading to uncoupled equations occurs when we neglect the coupling term (equivalent to h = 0. From the displacements. (8. the solution can then be used to solve Eq. (8.40) and (8. i. i. we solve Eq. all quantities are independent of time). h small. is brought to a new temperature qc > qo (by contact with thermal baths of temperature qc placed at each end of the rod). We will look at two ways this may arise. In either way. Thus.47) is Poisson’s equation in q. We see that q can be solved for independently of the equations of motion. For the 1-D case with no sources. The temperature increase is an example of a “thermal load.3.. T11.47) Eq.48) Eqs.e. The lateral surface of the rod is insulated and rigidly confined. Finally. (8.e. As ® with stationary problems. the equation for temperature becomes ∇2 θ − cε ∂ θ Q =− κ ∂t κ (8. After solving Eq. Ekk = 0).47) is d2 θ =0 2 dx1 . and then for the axial stress.” What is the axial stress generated in the rod during this process? Solution. Steady-state cases are known as the stationary problems of thermoelasticity [27].36). stresses are calculated from the constitutive equation. A rod of length L is held fixed at its ends in an undeformed.. the strains can be calculated. initially at a uniform temperature q0.40) to obtain the displacements. Eq.3 Thermal Prestress/PreStrain If the equations describing the displacement and temperature can be uncoupled.e. then we have a special situation of thermal prestress/prestrain. t) and use the result in Eq. (8. Eqs. or an incompressible material. (8. (8. The problem described is stationary.48) constitute the governing equations for the theory of thermal stresses [27].48) for q(x .PoRoeLASTICITy ANd TheRMoeLASTICITy 175 must be solved simultaneously subject to appropriate initial and boundary conditions on the displacements and temperature. we will solve for q first.46) ∇2 θ = − (8. (8. In that case.47). unstressed state.40) and (8. (8. The first is steady-state problems (i. Demonstration..46) for the displacements (where it now acts as a known term). (8. 8. and steady-state conditions are allowed to be reached. The rod.45) reduce to μ and ∂ 2 ui ∂θ ∂ 2 uk + (λ + μ ) + ρ Bi = γ 2 ∂ xi ∂ xk ∂ xi ∂ xk Q κ (8. Eq.

and 2. Substituting these results into Eq. there is no deformation. Finally. stress and temperature are state variables. i.4 PRoBLeMS 1. determine the temperature distribution within the material. As the rod is held fixed at its ends and laterally confined. there are the complicated theories of plasticity. show that the governing equation for the dilatation of a poroelastic medium consisting of an incompressible solid matrix and infinitely incompressible fluid is given by ∂ Ekk = k (λ + 2μ )solid ∇2 Ekk ∂t Eq (8. In this chapter.e. Under quasi-static conditions in the absence of body forces and fluid sources.36) yields T11 = −3Kα (θc − θ0 ) The negative sign indicates the rod is in a “state of compression” even though there has been no deformation. for which we have piezoelasticity and derivatives thereof. One can see that a wealth of theories has been developed for describing material behavior. though loading and temperature change may be occurring simultaneously.11.176 INTRoduCTIoN To CoNTINuuM BIoMeChANICS The solution to the above equation when q(0) = q(L) = qc is trivially found to be q = qc. one may be wondering if there is such a thing as porothermoelasticity. the problems of thermal prestress/strain can be analyzed in two parts: 1. and even poro-thermo-visco-hyper-elasticity! Of course. In thermoelasticity. Second-order tensors obey the following transformation law T � = Qri QsjTrs ij (8. Finally. express k ˜ (a) for a material orthotropic with respect to permeability (b) for a material transversely isotropic with respect to permeability 2. for deformations beyond the elastic limit. The answer is a definite yes. these all neglect the interaction of electric forces with the material. we introduced the permeability tensor and thermal conductivity tensor.50) . use the temperature distribution in the equations of motion to find the resulting displacements. E11 = E22 = E33 = 0. (8.49) In the manner of Section 4. strains. thermohyperelasticity. 8. thermoviscoelasticity. porohyperelasticity.50) equation resembles the “Heat” equation for Ekk. Thus. (8. and stresses corresponding to the loads and temperature changes. There are also theories of poroviscoelasticity..

T = 0. After being heated from 25°C to 3. infinitesimal loading. Solve Eq.a heat for a homogeneous material? For an inhomogeneous material? 5. assume no body forces.36) to show that the dilatation during unconstrained thermal expansion is Ekk = g q. ˜ (a) Use Eq. Furthermore. Show that.e. when the initial temperature is q(x1. (c) A material initially has a volume of 1 m3. Is the initial stress greater or less than the equilibrium stress? What does this mean for the design of materials that may experience thermal loads? 8. (8. Also. where a is the coefficient of linear thermal expansion (i. At x1 = L.26) for incompressible solid and fluid components when the loading rate is constant.t).e.025°C. 4. the volume is 3 m3. (8. repeat the last demonstration in this chapter to find the transient axial stress. What is the axial strain in the rod at steady state? You may assume the material is isotropic and homogeneous. no stress is produced. no additional heat sources. a traction-free lateral surface.. T11(x1. show that the change in volume per unit volume is 3aq. Assuming no heat sources. unstressed rod of temperature q0 and length L is fixed at x1 = 0. (8. What is ® -Ñ.e. Öp An infinite poroelastic medium has an amount of fluid Vf added to the origin at time t = 0. In the absence of body forces. E11 = aq). we derived Eq. in the limit as t ® ¥. — K (b) Using Eq..0) = q0. the bath at x1 = L is pulled in the positive x1 direction. h.. Consider when we neglect the coupling term.e. the rod is brought into contact with a thermal bath of temperature q2. i. What is T12? ò • • • • . For unconstrained thermal expansion. 7. creating a state of uniaxial tension (see Chapter 4). (8.PoRoeLASTICITy ANd TheRMoeLASTICITy 177 3. 6. A thermal bath of temperature q1 is brought into contact with the bar at x1 = 0.26) as the governing equation for pressure during a uniaxial consolidation experiment. T11 = -3Ka(qc - qo ). and quasi-static. The solution for the shear stress in an infinite thermoelastic medium when a point source of heat equal to h is supplied to the origin at time t = 0 is [28] � � � � � −R2 R2 3 1 γ μh e 4(κ /cε )t 1+ T12 = 4 (x1 x2 ) � 2π cε (λ + 2μ ) 6 (κ /cε ) t R π (κ /cε ) t �� � 1 R − erf � R 2 (κ /cε ) t x -x 2 where erf(x) = 2 — 0 e dx..38). i. Write Fourier’s law for a material orthotropic with respect to thermal conductivity. What is the coefficient of linear thermal expansion? You may assume that the coefficient of thermal expansion is constant over this temperature range. T33 = -st. i. An undeformed.

.

as the solid matrix may.. 2) the balance laws for the mixture as a whole have the same form as the equations for a single phase mixture. Common to these tissues or tissue-engineered constructs is viscoelastic behavior that results from the flow of fluid through a solid network containing pores of small size. mixture theory describes a material as a continuum mixture of n phases. a fact we will describe later. However. From these principles. and energy equations. The biphasic theory has been identified as a suitable framework to describe the constitutive behavior of soft. In this chapter our objective will be to describe the rheological behavior of materials made of binary mixtures. each also has continuity. have intrinsic flow-independent viscoelasticity. then the equations should reduce to those for a material composed of n –1 phases. or exchange of mass or momentum. temporomandibular joint disc. fibrocartilages (e. between the particular constituent and all other mixture constituents. and thus. and the volume fraction of one of those phases is equal to zero. such that each spatial point in the mixture is occupied simultaneously by all the phases. Each of the phases has a density and displacement field. and intervertebral disc). and even cells.1 INTRoduCTIoN Biphasic theory is a simplification of mixture theory.g.179 CHAPTER 9 Biphasic Theory 9. It allows mathematical simplification. In general. This aspect of superimposed continua is extremely important. as we do not have to follow individual spatial points corresponding to separate phases to describe the tissue’s mechanical behavior. This is known as flow-dependent viscoelasticity. These . it can be concluded that separate conservation equations can be constructed for each constituent and then added together to obtain the conservation equation for the entire mixture [28. where there are only two phases or constituents. momentum. hydrated biological materials such as articular cartilage. It has also been applied in tissue engineering. Truesdell constructed three guiding principles for the theory of mixtures as follows: 1) every property of the mean motion of the mixture is a mathematical consequence of the properties of the motion of the constituents. by itself. the meniscus. this seemingly simple operation is complicated by the fact that a new terms must be introduced to each constituent is constitutive equation to describe the interaction.29]. equations will be developed to describe the macroscopic structure using mixture theory. Instead of using the exact geometry of the microstructure. and 3) if one begins with n phases.

materials also have small permeabilities.1). A porous platen. nonporous. slowing the flow of fluid through the solid network and allowing an interstitial fluid pressure to be generated. and indentation compressive testing. unconfined. and 3) indentation (see Figure 9. two densities can be defined. Finally. 2) unconfined compression. 9. which allows fluid to exit through the specimen’s surface. is where the center of a tissue specimen is indented by a frictionless cylindrical probe that can be either porous or non-permeable. is a test setup where a tissue specimen is confined between two rigid. See text for description. but is free to expand in the radial direction.2 deFINITIoNS Consider an infinitesimal volume. Unconfined compression. Let us briefly discuss these three testing setups. There are three common compressive testing modalities to which the biphasic theory has been applied: 1) confined compression. Later in the chapter. The true density of each constituent is f ρT = dmf dV f s and ρT = dms dV s (9. dV. another testing setup most conveniently described in cylindrical coordinates. frictionless platens. From these masses and volumes. most easily discussed in cylindrical coordinates. Confined compression is an idealized deformation configuration in which a tissue specimen is placed into a chamber that confines the specimen at the bottom and on the sides.1: Illustration of experimental setup for confined. consisting of a fluid of volume dV f and mass dmf. each of these modalities can be used for creep or stress relaxation. we will examine some of the mathematics used to model confined and unconfined compression. is used to compress the sample. An axial load or deformation is applied. Furthermore. indentation.180 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 9. and a solid of volume dV s and mass dms. and the radial deformation is monitored.1) .

40).3) f f is also known as the porosity of the material.4) true.2). If. the continuity equation for the fluid and solid becomes ρf and (9.e. the fluid and solid are intrinsically incompressible). for some reason. (9.3) into Eqs.BIPhASIC TheoRy 181 The apparent density is ρf = dmf dV and ρ s = dms dV (9. (9.3 CoNSeRVATIoN oF MASS � � Dρ f � div ν f + =0 Dt � � Dρ s � div ν s + =0 Dt Using Eq. volume fractions of the fluid and solid can be defined as φf = dV f ρf = f dV ρT and φ s = dV s ρs = s dV ρT (9.1) and (9.4) This equation constrains the mass to remain within the mixture.. 9. the other n – 1 f i ’s would have to change.. given n constituents in the mixture. increase.6) yields � � ∂φf � + ∇ · φ fν f = 0 ∂t (9.g. Note that. f i of one constituent was to change. decrease. to keep Eq.7) .5) ρs (9.5) and (9. the following must hold i=1 ∑ φi = 1 n (9. substitution of Eq.6) Assuming the true densities of the fluid and solid are constant (i. (2..2) From Eqs. e. and f s is known as the solidity. (9.g. e. (9.

∂t n3f is equal to the volumetric flow rate divided by the cross-sectional area of the surface available for f s Now.9) � � � � �� f f f s s s ∇ · φ f ν 1 �1 + ν 2 �2 + ν 3 �3 + φ s ν 1 �1 + ν 2 �2 + ν 3 �3 = 0 e e e e e e � � � � f� s� f s � ∇ · φ f ν3 e 3 + φ s ν3 e 3 = ∇ · φ f ν3 + φ s ν3 e 3 = 0 where the second step follows. as the only direction that matters is the direction coincident with the axis of loading. At the bottom surface. The confined compression set-up is a 1-D deformational configuration. ν3 = ν3 = � �� � φ fν3f + φ s ν3s � Top = �� � f s � φ f ν3 + φ s ν3 � Bottom (9.10) . at the top surface. we arrive at the continuity equation for a biphasic system containing incompressible fluid and solid constituents Demonstration. Consider a confined compression setup where a displacement of magnitude u3(t). Eq.7) and (9. Solution. Determine the volumetric fluid flux through the bottom platen.182 INTRoduCTIoN To CoNTINuuM BIoMeChANICS and � � ∂φs � + ∇ · φ sν s = 0 ∂t (9.4).1). and using Eq.8) Adding Eqs. we get � ∂ � f f s φ ν3 + φ s ν3 = 0 ∂ x3 Integrating with respect to x3. (9. is imposed on the top face of the tissue by a rigid. and at the bottom surface. rigid.9) is � � � � ∇ · φ fν f + φ sν s = 0 (9. At the bottom face of the tissue there is an immobile. f s φ f ν3 + φ s ν3 = constant This implies that ∂ u3 . (9. (9. A (see Figure 9.8). porous platen also of area A. ν3s = 0 . x3 in this case. Performing the dot product. impermeable platen of cross-sectional area. In long form.

Substituting these values into Eq. we can write for the fluid and solid phases of the mixture ∂ Tijf ∂ xj and � � + ρ fBif + Πf = ρ f af i i (9. The term in the equation for the solid phase is interpreted similarly. We also see that the . we see the total acting body force is the density weighted sum of the external body forces acting on the individual phases and the internal body force terms.11) and (9. in terms of total area. Summing Eqs. n. and comparing Eq. In this term. we get � � � � � � � ∂ � f Tij + Tijs + ρ f Bif + ρ s Bis + Πf + Πs = ρ faf + ρ s as i i ∂ xj (9.4 CoNSeRVATIoN oF MoMeNTuM From Eq.21). (3.12).13) to Eq.BIPhASIC TheoRy 183 Q . (9. n = Q/A.10) gives the final answer. we identify that � � � � ρ Bi = ρ f Bif + ρ s Bis + Πf + Πs i i � � f f s s ρ ai = ρ ai + ρ ai (9. B if represents the sum of external body forces per unit mass acting on the fluid. and Õfi is the term describing internal body forces reflecting the interaction between the fluid and solid.11) ∂ Tijs ∂ xj + (ρ s Bis + Πs ) = ρ s as i i (9. (9. Note that Darcy’s law for fluid flux through a porous medium describes Aφ f relative fluid velocity.14) this.21). (9. ν3f = velocity in terms of the actual permeation area. (3.13) Noting Truesdell’s second principle.14) From Eq. Q=A ∂ u3 ∂t 9. whereas the biphasic theory describes fluid permeation. (9. The term (rfB fi + Õfi ) reflects non-inertial and nondeformation-induced forces. n f = n /f f.12) The term (rfB if + Õfi ) is worth discussing.

subject to the restriction Õfi = -Õsi . we will model the interaction between the fluid and solid as a viscous drag force proportional to the relative velocity between the two phases as was originally done by Mow and coworkers [31]. by Newton’s third law. (9. for a mixture of n constituents.12) under quasi-static conditions with no external body forces yields ∂ Tijf ∂ xj and + Πf = 0 i (9. Õsi = Ci(q)).e. simplifying Eqs.) and many more. For example. the stress tensors describing the fluid and solid could be any of the constitutive equations we have already discussed (i. Of course. of course. We will see that this simple assumption captures the gross rheological behavior of articular . the level of sophistication required to “accurately” model the problem of interest (as more complicated choices can lead to intractable mathematics and numerical simulations) and.11) and (9. resulting in zero net total acting body force.184 INTRoduCTIoN To CoNTINuuM BIoMeChANICS overall acceleration within the body is a density weighted sum of the acceleration of the solid and fluid phases. we are free to choose whatever we would like for Õfi or Õsi and T fij and T sij. which suffices for the majority of problems in soft tissue biomechanics. for our discussion.16) ∂ Tijs ∂ xj + Πs = 0 i (9.e. Also. viscoelastic. thermoelastic.. (9. Newtonian fluid.. Let us now assume quasi-static equilibrium and no external body forces acting on the individual phases. in ac˜ cordance with Truesdell’s second principle. Õfi = -Õsi . which we will assume are nonzero.17).e.15) Finally. Õsi = ajE 3ij + bjE 2ij + cjEij + di) or be a constant that depends on the temperature (i. good old fashioned intuition.. So. the interaction body force could be defined as a cubic function of the current strain (i. hyperelastic. This leaves only Õ fi and Õsi . it is required that i=1 ∑ Πi = 0 n � (9.16) and (9.17) Note that the sum of Eqs. Then. In general. the choices made for both the internal body forces and constitutive equations should be based on experimental data. linearly elastic. is Ñ × T = 0. etc. However.

With Eq.BIPhASIC TheoRy 185 cartilage under various loading conditions.s/m.20) (9. k. (9. They are � � � � � Πf = K ν s − ν f and � � � � � � Πs = −Πf = K ν f − ν s where K is the coefficient of diffusive resistance and has units of N.5 CoNSTITuTIVe eQuATIoNS The next choices we have to make are what constitutive equations will be used to relate the stresses in the fluid and solid phases to the hydrostatic pressure. of Darcy’s law. This is because the development of constitutive equations is either an entirely empirical process or a complex process depending on principles such as the second law of thermodynamics.18) ∂ Tijf ∂ xj ∂ Tijs ∂ xj � � + K νis − νif = 0 � � + K νif − νis = 0 (9. which are beyond the scope of this book. Please see problem 4 for how K relates to the permeability.21) . and strain rate generated in the tissue upon loading. the conservation of momentum equations for our biphasic mixture become (9. the equations for Õfi and Õsi are those used to describe a viscous damper. Thus. 9. Mow and coworkers arrived at the following constitutive equations for the fluid and solid phases Tijf = −φ f pδij + λf Df δij + 2μf Df − QEkk δij + 2KC Γij kk ij and s s Tijs = −φ s pδij + λ Ekk δij + 2μ Eij + λs Dkk δij + 2μs Dij − 2KC Γij . In this text. [1]. (9. strain. material reference frame indifference.18). and considering infinitesimal deformations. Using the latter approach. we have simply quoted constitutive equations. etc.19) We will now turn to a discussion of the constitutive equations used in the classical linear biphasic theory [31].

and are not the intrinsic structural properties of the solid constituents. or matrix. let Q = KC = 0. we will treat the fluid phase as inviscid and neglect the viscous nature of the solid matrix. It is worth pointing out that the constants of the solid are apparent.17)). linearly elastic solid (Eq. The remainder of T s represents ˜ the viscous nature of the solid matrix should it need to be included. or drag coefficient) is constant ...20) and (9. and the first three terms in T s represent an isotropic. dij is the Kronecker delta. The remaining ˜ terms are material constants. Notice that the first three terms in T f represent an isotropic Newtonian ˜ fluid (Eq.e. there is no rate of strain or rate of dilatation-induced stress contributions. Furthermore. Furthermore.e. and not the properties of specific components (i.22) (9.e. properties (i.. the resistance to flow. at every point in the tissue. 9. tissue permeability (i. both the solid and fluid are present 2. where p is the hydrostatic pressure in the system. and therefore rTf and rTs are constant (requiring the total tissue volume to remain constant) 3. Eij are components of the infinitesimal strain tensor. the total hydrostatic pressure is p = f f p + f sp). we have made several assumptions in the development of the biphasic theory equations. These assumptions are the following: 1. (9. as well as dilatational. these properties are the properties of the solid matrix.e. and therefore f f and f s are constant 4. As is commonly. the solid “matrix” is not incompressible. but isotropic and linearly elastic (think of a truss with rigid membranes attached by elastic hinges) 5. collagens and proteoglycans).23) The fluid stress in classical biphasic theory is only due to an apparent hydrostatic pressure. properties of the solid continuum). in articular cartilage. and is proportioned to the solid and fluid phases according to their volume fractions (i.31)) with the contribution of hydrostatic pressure in the tissue. and Gij =W ij - W ij (recall W is the spin tensor). the fluid is inviscid 6.. the strains are infinitesimal.186 INTRoduCTIoN To CoNTINuuM BIoMeChANICS respectively. We shall now apply several simplifying assumptions to these constitutive equations. For example.21) then become and Tijf = −φ f pδij Tijs = −φ s pδij + λ Ekk δij + 2μ Eij (9. (5. the solid and fluid phases are intrinsically incompressible. The solid stress is the sum of the apparent hydrostatic pressure. ˜ (4.and strain-induced stresses. Dij are s f components of the strain-rate tensor.6 SuMMARy ANd eQuATIoNS oF MoTIoN So far. and was originally done. Eqs.

26) Inserting Eqs.27) and Note that these two equations each represent three equations. considering the x3 direction. no flow-independent viscoelasticity) Assuming that the fluid and solid are intrinsically incompressible. They are given as � � � � � � −∇ φ f p + K ν s − ν f = 0 (9.29) . we arrived at the following equations for the balance of momentum � � � � ∇ · T f + K νs − νf = 0 ∼ � � � � (9.28) −φ f � � ∂p s + K ν3 − ν3f = 0 ∂ x3 (9.26) into Eqs.. isotropic solid phase to describe the constituents. we derived the continuity equation for the biphasic mixture as � � � � ∇ · φ fν f + φ sν s = 0 (9.e. the equations of motion are � � � � � � −∇ φ s p + λ ∇Ekk + 2μ ∇ · E + K ν f − ν s = 0 ∼ (9.. inertial effects are negligible 8. choosing to model the interaction force between the fluid and solid phases as a viscous drag force and assuming quasi-static conditions.25) ∇ · T s + K νf − νs = 0 ∼ We then assumed infinitesimal deformations and chose an inviscid fluid phase and a linearly elastic.BIPhASIC TheoRy 187 7.. viscoelastic (i. time dependent) effects are caused by frictional drag associated with fluid flow through the solid matrix (i. (9.e. we get the equations for motion of the biphasic system. For example. described by the following constitutive equations T f = −φ f pI ∼ ∼ T s = −φ s pI + λ Ekk ∼ + 2μ E I ∼ ∼ ∼ (9. one for each of the principal coordinate directions. (9. i.25). taking the dot product of Eqs.24) Furthermore. (9.27) ® and (9.28) with e3.e.

x1 and x2 directions).30) Demonstration. Setting the x3 direction as the direction of loading. Furthermore. using Eqs. we are left with 2mÑ × E = 0 from Eq. Also. (9. Thus. Tij = -pdij. the strain tensor is � � ∂ 2 u3 � ∂ u3 e 3 and from which the dilatation. is calculated to be . confined and unconfined compression.26). equivalently. It follows that ∇Ekk = 2 ∂ x3 ∂ x3 � � � � ∂ u2 � � 3 . as both phases are incompressible.28). Furthermore. Consider a state of uniform hydrostatic pressure.30) can be combined (see problem 4) to yield ⎤ 0 0 0 ⎥ ⎢ [E ] = ⎣ 0 0 0 ⎦ ∼ � 0 0 ∂ u 3 ∂ x3 ⎡ (9. But. (9.31) . T f = -ffpI and T s = -fspI . Hence. Using these ∇·E = e3 e ∂ x2 ∂t 3 relationships. ˜ ˜ ˜ ˜ We will now examine two more complicated loading conditions. the displacement of the solid matrix is a function of x3 and t alone. so Ñ(ffp) = Ñ(f sp) = 0. the change in volume is zero. u1 = u2 = 0 u3 = u3(x3. which implies Ñ × E = 0.t) Thus. n f - n s = 0. ® ® Hence. Eqs. as force is balanced in all directions.29) and (9. 9. (9.7 CoNFINed CoMPReSSIoN The confined compression experiment is a 1-D loading situation because we assume lateral confinement prevents displacement and fluid flow in the radial direction (or.188 INTRoduCTIoN To CoNTINuuM BIoMeChANICS ∂ ∂p −φ +λ (E11 + E22 + E33 ) + 2μ ∂ x3 ∂ x3 � � f s + K ν3 − ν3 = 0 s � ∂ E31 ∂ E32 ∂ E33 + + ∂ x1 ∂ x2 ∂ x3 � (9. ˜ ˜ because of incompressibility.e. Ekk = 0. and no fluid flows into the tissue. Ekk. E = 0 under a state of hydrostatic pressure. there is no relative motion between the two phases. What are the stresses in the fluid and solid phase? Solution: There is no pressure gradient.. i. the velocity of the solid matrix is ν s = ∂ u3 �3 .

s). t) = − A where A is the surface area of the tissue (and loading platen). and (9. s T33 = λ ∂ u3 ∂ u3 ∂ u3 + 2μ = HA ∂ x3 ∂ x3 ∂ x3 ∂ u3 F (0. (9. Pa).7.e. The solution is � � ∞ 2 F x3 −λn HA k t h2 (h − x3 ) + ∑ cncos λn u3 (x3 . t) = 0 (9.37) constitute a nonhomogeneous PDE.36) so that − F ∂ u3 (0.1 Creep As described in Chapter 7.32) with the conditions (9. 9. The initial condition is u3 (x3 . Using Eqs. 0) = 0 (9. It can be solved by separation of variables with appropriate boundary and initial conditions [12].BIPhASIC TheoRy 189 HA k ∂ 2 u3 ∂ u3 = 2 ∂t ∂ x3 (9. i.35) T33 (0.31) and (9. then the stress on that surface is F s (9.34) The second boundary condition reflects the fact that the solid matrix at the tissue surface supports the applied load throughout the deformation.. (9. Eq. t) = − ∂ x3 HA A (9.33). HA is the aggregate modulus (units of stress.34). creep is deformation of a body under an applied constant load.23). (9. and noting that there is no hydrostatic pressure at the surface (for a free-draining platen). Eq.37) We have transformed the stress boundary condition into a displacement boundary condition. (9. If we designate the applied load as F. t) = e (9. t) = HA A ∂ x3 ⇒ (9.33) The first boundary condition is at the bottom of the confinement chamber where there is no displacement u3 (h.38) HA A h n=1 .32) is of the same form as the 1-D diffusion or heat transfer equation covered in a first course on PDEs. and k is the Darcy permeability (units m4/N.32) which is the governing equation for the confined compression configuration.

t) = H k ∞ F −λ 2 A t h + ∑ cn e n h2 HA A n=1 � � ∞ 1 2H k F −λn A t h2 = h 1−2 ∑ 2e HA A λn n=1 (9.000 s. (9. (9. we get F u3 (0. known as the “gel diffusion time (tD). At x3 = 0. k can be determined by nonlinear curve fitting Eq. each term in the series solution decays with a time constant equal to h 2/HA k. For example.42) where s is the stress in the solid matrix. HA is calculated by examining the curve at t → ∞ .2. Rewriting Eq. (9. First. The properties HA and k can be determined by curve fitting experimental data. an expression reminiscent of Hooke’s law. An intermediate step of problem 4 shows that ∂p 1 ∂ u3 = ∂ x3 k ∂t (9.40) A plot of this solution using the first 100 terms of the sum is shown in Figure 9.40). Let us make a few observations concerning Eq. The equation for finding HA is HA = hF u3 (0.41).40) with the calculated value of HA and a measured value of the tissue’s height. Eq. A typical value of the “gel diffusion time” for articular cartilage is ~1.39) Experimentally. one could import the creep data into MatlabÓ and use a nonlinear curve fitting algorithm (see problem 6).” The “gel diffusion time” increases with decreasing HA and k and increasing h. ∞) A (9. (9. and e is the bulk tissue strain.190 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where λn = (2n − 1) π 2 � �� � � �� � F 2h F 2h cn = (cos (λn) − 1) = − 2 2 HA A λn HA A λn (9. at which time the fluid pressure and fluid flow have gone to zero.38) becomes u3 (0. Let us now examine the pressure distribution in the tissue. and the entire load is supported by elastic stresses in the solid matrix.43) .41) Next. ∞) HA = h A → σ = HA ε (9. the most easily accessible displacement that can be measured is at the tissue’s surface.

Taking the time derivative of Eq.BIPhASIC TheoRy 191 FIguRe 9. t) = � n=1 ∑ ∞ � 2F Ah �� h λn � � � � 2 x3 −λn HA k t h2 + f (t) sin λn e h (9.38) term by term results in 1 ∂p = ∂ x3 k � � � 2 x3 −λn HA k t 2 h cn cos λn e ∑ h h n=1 �� � � � �� � � ∞ � 2 F 2h x3 −λn HA k t 2 HA h2 = ∑ −λn 2 − e cos λn 2 HA A λn h h n=1 � � � ∞ � 2 x3 −λn HA k t 2F h2 = ∑ cos λn e Ah h n=1 ∞ � � 2 HA k −λn 2 � (9.44) Integrating Eq.44) term by term yields p (x3 . (9.45) . (9.2: Strain and pressure resulting from a confined compression creep experiment.

we see that f(t) = 0.. As shown in Figures 9. except at x3 = 0 where the boundary condition requires p = 0 (see Figure 9. t = 0.47) Let us briefly discuss the interpretation of Eqs. Furthermore.1tD (+). t) = −u0 (9.2 Stress Relaxation For stress relaxation. such that at equilibrium the entire stress is borne by the solid matrix. t = 0.3. i. Also note that load is shared differently depending on vertical location within the tissue.t) = 0. t = tD (–). proceed downward. t) = 0 Applying Eq. the pressure gradient and fluid velocity become more uniform throughout the tissue. Accordingly. the total stress is the sum of the stress in the solid phase and the stress in the fluid phase. (9. the pressure gradient. This reflects the fact that fluid cannot be instantaneously squeezed from the tissue.45).2). At t = 0+. t = 0.8tD (D).192 INTRoduCTIoN To CoNTINuuM BIoMeChANICS where f (t).. Concomitantly. (9.2). 9. t = 0. x3/h = 0. The external load is initially supported entirely by a hydrostatic pressure equal to f fp in the fluid and f sp in the solid.47).1) where the solid has a larger load bearing role. ¶p/¶x3. Notice that the total normalized stress is always one. With increasing time. nf and u3 are initially greatest at the surface region (~top 10%. the boundary condition at the tissue surface is u3 (0. and increasing solid stress.e. during fluid flow is greatest at the top surface and declines with depth in the tissue.001tD (*). is determined from the boundary condition p(0. t = 0. (9.2tD (–).2 and 9. a “constant” of integration with respect to x3. u3 (0.1 correspond to different time points: t = 0 (solid line).4tD (x).40) and (9.7. Accordingly.46) to Eq. t = 5tD (°).48) . Hence. fluid pressure declines over time to zero. t) = n=1 ∑ ∞ � � � � 2 x3 −λn HA k t 2F h2 sin λn e λn A h (9. there is no instantaneous deformation. t = 0. Compaction of the solid matrix. i.3 shows how the load is shared between the fluid and solid phases during confined compression.01tD (solid line). the fluid pressure declines. an initial displacement is applied and held constant (see Figure 7. Figure 9. The several curves in the graph on the right of Figure 9.46) p (x3 . (9.e.

ln = np. t) h h n=1 � � ∞ H k u0 −λ 2 A t 1 + 2 ∑ e n h2 = HA h n=1 (9. respectively. t) = (x3 − h) + 2u ∑ sin λn exp −λn 2 t h h h n=1 λn (9. The equilibrium stress at the tissue surface is also of interest. s T33 (0.3: Load sharing between fluid and solid during a confined compression creep experiment.50) Note that the infinite stress at t = 0 is not a physical reality.51) . It is given by Eq. the stress on the tissue’s surface is experimentally accessible through the use of a standard materials testing system.36). The boundary condition at the bottom of the confinement chamber and the initial condition are still given by Eqs. Similar to displacement at the surface in a creep test. (9. (9. (9. given by s T33 (0. The solution to Eq.33). Figure 9. ∞) = HA u0 = HA ε h (9.32) under these conditions is � � � � ∞ u0 1 x3 0 2 HA k u3 (x3 . t) � 2 ∂ u3 � u0 2u0 ∞ −λn HA k t � = HA = HA + HA ∑ e h2 ∂ x3 �(0.BIPhASIC TheoRy 193 FIguRe 9.49) Note: Here.34) and (9.4 shows the stress at the tissue’s surface.

. let us now turn to another common method for testing biological tissues in compression. Note that the time constant governing stress relaxation is the same as that governing creep.e. and we refer the interested reader to reference [31] for the solution. namely. At equilibrium. The above solution applies only to an instantaneously applied displacement. with HA as the “equivalent modulus of elasticity.194 INTRoduCTIoN To CoNTINuuM BIoMeChANICS FIguRe 9. it is experimentally impossible to apply a step displacement. when performing a creep or stress-relaxation test. unconfined compression.4: Stress at tissue’s surface during a confined compression stress relaxation experiment. Typically. the “gel diffusion time”. we use e to denote the applied strain. The solution for stress-relaxation subject to ramp loading is beyond the scope of this text.” Similar to creep. Hence. equal amounts of time should be given. the stress-relaxation test is divided into a compressive phase and relaxation phase. Here.52) u3 (0. t) = ν0 t0 t ≥ t0 So. The unconfined .8 uNCoNFINed CoMPReSSIoN Having analyzed the confined compression problem. In reality. the result is similar to Hooke’s law. a ramp displacement is applied as follows � ν0 t 0 ≤ t ≤ t0 (9. HA is determined by the equilibrium response. i. and k is determined by curve fitting the experimental data. 9.

where the radius of the specimen is a. total Trr (r = a.56) both of which must be transformed to displacement boundary conditions. one obtains the governing equation for radial displacement of the solid matrix during unconfined compression (see problem 8) ∂ 2 ur 1 ∂ ur ur 1 ∂ ur 1 r ∂ Ezz + − 2− = ∂r2 r ∂r r HA k ∂ t HA k 2 ∂ t (9. these assumptions are ur = ur (r. combining that result with Eq. Furthermore.57) . t) = 0. which means that the problem is independent of q. Assuming that the tissue specimen is compressed between two impermeable. (9. and z. t)= nrs (r = 0.. q. frictionless platens. t) = 0 p (r = a. Unconfined compression is most easily analyzed using cylindrical coordinates. Mathematically.54) φs where α = f .BIPhASIC TheoRy 195 compression setup is shown in Figure 9. as done before.54). the geometry of the testing setup suggests axisymmetric displacement and pressure fields.e. p = p (r. Heaviside) or a ramp displacement. pressure can be eliminated from the equations for the φ stress in the radial direction of the fluid and solid. and 2) the axial strain is independent of z [31]. this equation can now be solved using Laplace transforms under conditions of creep and stress relaxation provided appropriate boundary and initial conditions and the axial strain rate.1. t) = 0 (9. Furthermore. Finally. For boundary conditions we have the following. r. t) and uz = z Ezz (t ) (9. due to symmetry. where for creep we have � a(t) 0 total Tzz 2π r dr = F (t ) (9. nr f (r = 0. Though complex.55) Note that this is a nonhomogeneous PDE in cylindrical coordinates. we get the following expression for the velocity of the fluid in the radial direction (see problem 8) r ∂ εzz νrf = − (1 + α ) − ανrs 2 ∂t (9. For stress relaxation the axial strain rate is prescribed as either a step (i. t) . 1) the radial displacement and pressure are independent of z.53) From the continuity equation and the fact that.

Let the x3 axis be coincident with the loading axis. show that ν3 = − ν3 = − ν3. For the biphasic theory. explain how the solid portion of the matrix is compressible while the biphasic material is incompressible. (iii) Use the last result from part (a) and the fact that f s + f f = 1 to eliminate p between the conservation of momentum equations. φf φf (Hint: Consider the boundary conditions). determining the axial strain rate. Let HA = l + ~ ~ 2μ (HA is known as the aggregate modulus). For the interested reader.9 PRoBLeMS 1. (i) Using Eq. where (φ f )2 . What is the approximate force recorded by the sensor? How could the actual (not approx. � � 4.32). (9.) force be calculated? Explain why this force develops. 1−φ f φs s f s (a) Using the continuity equation. The initial condition is ur (r.33]. (ii) Write the conservation of momentum equations in the x3 direction for the fluid and solid phases under static conditions. (9. A biphasic material has a porosity of 96%. as the initial and final configurations of the tissue are approximately the same. Eq. The result is Eq. The constitutive relationship for the solid portion of a biphasic material is s Tij = −φ s pδij + λ Ekk δij + 2μ Eij . 3. F(t) = -F0H(t). This problem will examine confined compression. 0) = 0 (9. the radius does not change over time. For infinitesimal deformations.e.32). write the solid and fluid stress tensors. i.. Lastly. we would be remiss to not mention creep indentation. A force sensor of small area A is placed on the side wall inside of the confined compression chamber at a distance midway between the tissue surface and the bottom of the chamber. more information on creep indentation can be found in the references [32.31).58) Solutions for each of these situations were first worked out by Armstrong and colleagues in 1984 [32]. 9. Describe in words what each term on the right hand side represents. what is the apparent density of the fluid? 2.10). k= . T s and T f. There is no motion allowed in the x1 or x2 directions. K 5.196 INTRoduCTIoN To CoNTINuuM BIoMeChANICS a Heaviside load application. (9. (9. following the steps below. Eq. What is n3f - n3s ? (b) Derive the governing equation for confined compression. What is the solidity? If the true density of the fluid is 1 g/mL.

Eqs. 5.40) to fit the curve using n = 1.0135 0.34).0128 0. Use Eq.02 0.0183 0.009 0. Find H A and k .BIPhASIC TheoRy 197 6. A step load of F = 0.007 N was applied to a porous platen of A = 0.0082 0.0071 0.0107 0.0142 0.0052 0. Also. (9.018 0. (9.0201 7. (9. evaluate Eq. . You may find that the data and fit do not match. If so.017 0.49) at t = ¥ to obtain the equilibrium response. and (9.0199 0.0116 0.49) satisfies the boundary and initial conditions of stress relaxation.0192 0. The latter requires nonlinear curve fitting. (9. why may this be? TIMe(s) dISPLACeMeNT (mm) 0 1 2 3 4 7 10 15 20 25 40 60 80 120 140 240 340 390 450 470 500 (» ¥) 0 0.015 0.016 0.5 mm.503 mm2.33).48).0201 0. (9. and 10 terms of the sum.0201 0. MatlabÓ’s “cftool” command is one possible method. Consider the following data from a confined compression creep experiment of a tissueengineered articular cartilage construct with thickness h = 0. Verify that Eq.

In cylindrical coordinates. appropriate boundt s ary conditions. (9. h � ⎢ ⎢ ⎢ � � ⎢ 1 1 ∂ ur ∂ uθ uθ [E ] = ⎢ + − ∼ ⎢ 2 r ∂θ ∂r r ⎢ ⎢ � � ⎣ 1 ∂ ur ∂ uz + 2 ∂z ∂r ∂ ur ∂r 1 2 � 1 ∂ ur ∂ uθ uθ + − r ∂θ ∂r r 1 ∂ uθ ur + r ∂θ r � � 1 2 ∂ uθ 1 ∂ uz + ∂z r ∂θ � � � ⎤ 1 ∂ ur ∂ uz + ⎥ 2 ∂z ∂r ⎥ � �⎥ 1 ∂ uθ 1 ∂ uz ⎥ ⎥ + 2 ∂z r ∂θ ⎥ ⎥ ⎥ ⎦ ∂u z ∂z Furthermore. • • • • . (6. will be helpful. Eq. (9. eliminate the pressure term from the two equations and use the result from part (a) to arrive at Eq.42). (b) Write the radial equilibrium equations for the fluid and the solid constituents in terms of displacements. This problem will examine unconfined compression. and the fact that n z = n z (because there is no relative motion between the fluid and solid in the z direction) to arrive at Eq. (9.9). in cylindrical coordinates. Recall that unconfined compression is an axissymmetric situation.198 INTRoduCTIoN To CoNTINuuM BIoMeChANICS 8. n=1 � ∞ 0 � nπ x f (x) sin dx . recall Eq. (c) Now. cn = � � h ∞ nπ x f (x) = ∑ cn sin where h . E has the following matrix representation ˜ ⎡ 2 The formula for the coefficients of a Fourier sine series. (a) Use the continuity equation.55).54).41) and (6.

We would also like to thank prior teaching assistants of the Introduction to Continuum Biomechanics course that has been taught at Rice University for some time now. Finally. Adrian Shieh.199 Acknowledgments We would like to extend wholehearted gratitude to the late Professor R. and C. Graduate student Najmuddin Gunja helped create the first soft copy of the course notes. thanks to the many students who have sweated through the course. Michael Lai. Corey Scott were instrumental in editing and adding to the course notes. Much of the first half of this text is modeled off of that excellent treatise. We would also like to acknowledge the book Introduction to Continuum Mechanics coauthored in part by Professor Lai. whose biomechanics classes were taken by Professor Athanasiou during his time as a graduate student. Michael Detamore. Drs. . Skalak and to Professor W. Specifically. which is reflected in the present text. which taught Professor Athanasiou continuum mechanics many years ago. both of Columbia University.

.

The subject of continuum biomechanics could fill up volumes. we solved several problems. as with all subjects. We began the text with a solid foundation in Cartesian tensors. Finally. appropriate boundary and initial conditions are applied to obtain a solution to the governing equations. A common numerical approach in . typically yielding displacement or velocity fields. While closedform solutions are desirable. Along the way. even at an elementary level. there is a relatively straightforward recipe for tackling continuum mechanics problems. as not every problem has a closed form solution). However. and energy). much of the enrichment of your study of biomechanics is contained in the problems we have included at the end of each chapter. but depth on only a few. If you take a moment to reflect. As this is an introductory text. Note that curvilinear coordinates and a presentation of general tensor algebra and calculus are the subjects of intermediate and advanced texts. there is much more to learn. linear and angular momentum. numerical approaches are often taken (and are sometimes the only recourse.e. We then spent a good deal of effort studying strain (kinematics) and stress separate from one another. and demonstration of important points by solving some simple. the mathematical aspects of the solution of continuum biomechanics problems can be quite complex. Finally. we showed how stress and strain can be related to each other through the many constitutive theories we discussed. In fact. stress is determined from substituting the strains or strain rates back into the constitutive equations. We have chosen to curtail our discussion to salient features of the theories we have addressed. However.g. Though the physical concepts and solution approach are straightforward.201 Afterword We hope you have learned much about the subject of continuum biomechanics from reading this book. One then substitutes the constitutive equation(s) describing the body of interest and the kinematic relations (i. classical problems. Lastly. compatibility equations) into the conservation laws to arrive at governing equations for the loading configuration under investigation. we have presented a breadth of topics.. as they are distinct concepts. One begins with conservation laws which must be true for all systems (e. strains and strain rates can be calculated. conservation of mass. reading is not enough! There is no substitute for solving continuum mechanics problems. presentation and explanation of the constitutive equations.. Reaction forces can be obtained by integrating the stress over its area of application. From these fields.

. Similarly. the references contain many texts that expand upon the material presented here at an intermediate or advanced level. we showed how biphasic theory and poroelasticity are related. In fact. though “mixtures of one component” certainly reduce to the theories of fluids or solids. We believe that once these principles and theories have been studied and understood. However. homework problems. We would like to point out.e. A fantastic introductory text on finite element analyses can be found in the references [35]. “to err is human. For example. We presented theories that are germane to establishing descriptions of and solutions for the behavior of biological tissues and materials. as the latter is a subset of the former. our emphasis was the foundations of continuum mechanics. it is with relative ease that one can transition to biomechanics problems. as well as bringing to our attention any errors of commission or omission that you may come across. While we have made every effort to ensure the content of the text. but fluid and thermal effects as well). without being exhaustive in the “bio” aspects. and are hence limited by the extent to which they sufficiently describe certain aspects of material behavior. and homework solutions is correct. including the mathematics necessary to solve more complicated problems.” We would appreciate any and all constructive feedback as to how we can improve the text. KA2 and RMN . we have attempted to integrate the vertical and horizontal layers of complexity among the theories we have presented.202 INTRoduCTIoN To CoNTINuuM BIoMeChANICS continuum mechanics is the finite element method. we view hyperelasticity as vertically above linear elasticity. in that they are able to accurately describe a vast array of material behavior. Though the subject of this book is biomechanics. that while these theories provide great mathematical models of physical reality. not solely stress or strain. we view these as more horizontally related to elasticity through the inclusion of different aspects of physical reality (i. we showed how linear viscoelasticity is vertically above linear elasticity and Newtonian viscous fluids. Mixture theory is somewhat in a class of its own. Thank you. We have made every effort to cite texts that one should be able to pick up and follow after reading this book (including the fact that many of the references share our notation). they are only models of physical reality. Finally. While elasticity is a subset of poroelasticity or thermoelasticity.

[14] Fung YC. [5] Holmes MH. River Edge. [7] Pioletti DP. New York. Leyvraz PF. [10] Leipholz H. J Biomech 1998. Rakotomanana LR. Upper Saddle River.3:111–24. [8] Miller K. Burlington. Mineola. 2004. World Scientific Publishing Co. River Edge. Biofluid Mechanics. Benvenuti JF. Applied Partial Differential Equations with Fourier Series and Boundary Value Problems.23:1145–56. Pte. Inc. Upper Saddle River. Biomechanical Models for Soft Tissue Simulation. 4th ed. Cambridge University Press 2008. Ltd. 2nd ed. 153–58. Krempl E. Weichert D. Mow VC. Mechanical characterization of skin-finite deformations. Springer Verlag 2000.1016/0021-9290(70)90055-2. [6] Veronda DR. Theory of Elasticity. Transport Phenomena in Biological Systems. The nonlinear characteristics of soft gels and hydrated connective tissues in ultrafiltration. [3] Spencer AJM. 2004. Nonlinear Continuum Mechanics of Solids—Fundamental Mathematical and Physical Concepts. J Biomech 1990. Leyden. [13] Reddy JN. An Introduction to Continuum Mechanics with Applications. [15] Mazumdar JN. 1992. Method of testing very soft biological tissues in compression. Wu Y. J Biomech 1970. World Scientific Publishing Co. Pearson Prentice Hall 2004. Katz DF. [11] Truskey GA. doi:10. Pte. Continuum Mechanics. [9] Maurel W.203 Bibliography [1] Taber LA. Ltd. Springer Verlag 1993. Viscoelastic constitutive law in large deformations: Application to human knee ligaments and tendons. Dover Publications. Westmann RA. Nonlinear Theory of Elasticity Applications to Biomechanics. Noordhoff International Publishing 1974. New York. Yuan F. J Biomech 2005. 3rd ed. Butterworth Heinmann 1993. Biomechanics Mechanical Properties of Living Tissues. New York. Thalmann NM. [12] Haberman R. Thalmann D. [2] Lai MW. [4] Basar Y.1016/S0021-9290(98)00077-3. New York.31:753–7. doi:10. . Springer 1998. Rubin D. Introduction to Continuum Mechanics. Pearson Prentice Hall 2004.

[25] Simon BR. Waltham.cmpb. doi:10. Viscoelasticity of Engineering Materials. Armstrong CG. Multiphase poroelastic finite element models for soft tissue structures. An Introduction to Biomechanics: Solids and Fluids.2002. [33] Mak AF. Delange SL. Phan AV. [31] Mow VC. Madanagopal S.29:209–44.1060. Zhu WB. New York.1016/0021-9290(87)90036-4. Springer Science+Business Media. doi:10. Biphasic indentation of articular cartilage—I. J Appl Phys 1941.83:29–33.2.351278. Craine RE. [23] Biot MA. [30] Humphrey JD.1093/qjmam/29. Viscoelasticity. Princeton. Kuei SC. 2nd ed. Princeton University Press 2000. [17] Machiraju C. An analysis of the unconfined compression of articular cartilage.05.1572517. [26] Norris A. Shieh AC. Theoretical analysis. Biphasic indentation of articular cartilage—II. Analysis and Design. Continuum biomechanics of soft biological tissues. Pergamon Press Inc. [20] Haddad YM. Blaisdell Publishing 1968. Lai WM. Proc R Soc Lond A 2003. Mow VC. Creep indentation of single cells. Viscoelastic studies of human subscapularis tendon: Relaxation test and a Wiechert model.1098/rspa. Biphasic creep and stress relaxation of articular cartilage in compression: Theory and experiments.1115/1. Gibbs MC. [19] Flugge W. Comput Methods Programs Biomed 2006. 1986. Athanasiou KA. Boca Raton. J Biomech 1987. [29] Atkin RJ. Lai WM.459:3–46. Athanasiou KA. Thermoelasticity. Lai WM. Pearsall AW. Springer Verlag 1987. CRC Press LLC 1999.1712886.106:165–73. J Biomech Eng 2003.2006. Chapman & Hall 1995. [32] Armstrong CG.1063/1. Pearson Education Limited 1992. Petersson H. Introduction to the Finite Element Method.20:703–14. General theory of three-dimensional consolidation. . On the correspondence between poroelasticity and thermoelasticity. Continuum theories of mixtures: Basic theory and historical development. Thermoelasticity. New York. [27] Nowacki W.1063/1. J Biomech Eng 1984.22: 853–61. Spencer AJM. Lai WM. Blaisdell Publishing Co.102:73–84. Mow VC. doi:10. doi:10. [21] Lakes RS. doi:10. New York.45:191–218.1016/0021-9290(89)90069-9. Harlow. doi:10. [28] Parkus H.12:155–64. [35] Ottosen NS. doi:10. [24] Wang HF. A numerical algorithm and an experimental study. [18] Boehler JP. [34] Mow VC. Applications of Tensor Functions in Solid Mechanics. J Biomech 1989.209. Elmsford.125:334–41. Betten J. Appl Mech Rev 1992. [22] Koay EJ.71:1138–41. Waltham. J Appl Phys 1992. Q JMAM 1976. Inc. 2004. Viscoelastic Solids.004. J Biomech Eng 1980. 1967.204 INTRoduCTIoN To CoNTINuuM BIoMeChANICS [16] Humphrey JD.1016/j. doi:10. Theory of Linear Poroelasticity with Applications to Geomechanics and Hydrogeology.

205 Author Biography K. He also heads the Musculoskeletal Bioengineering Laboratory at Rice University. Athanasiou is the Karl F. Hasselmann professor of bioengineering at Rice University and an adjunct professor of orthopaedics and oral and maxillofacial surgery at the University of Texas. in chemistry from the University of Chicago. He holds a B.D.S.A. in biological chemistry and A. .M.B. His research project focuses on impact loading of articular cartilage and mechanobiological aspects of articular cartilage tissue engineering. in mechanical engineering (bioengineering) from Columbia University. He is also a student in the Baylor College of Medicine MSTP program. Natoli is a graduate student in Professor Athanasiou’s laboratory. R. He holds a Ph.

- Continuum Mechanics
- Solutions Manual Continuum Mechanics Lai 4th Edittion
- All Problems
- Atmospheric and Oceanic Fluid Dynamics
- Continuum mechanics problems
- Finite Element Analysis - From Biomedical Applications to Industrial Developments1
- Numerical Geodynamics - An Introduction fo computational methods with focus on solid Earth applications of continuum mechanics
- Biomechanics
- Biomechonics
- Introduction to Biomechanics
- Measurement of Man Locomotion
- Theoretical Biomechanics
- Introductory Biomechanics
- HAAKE Practical Approach to Rheology and Rheometry - Schramm
- Introduction to BIomechanics
- Shoulder Anatomy
- Physical Therapy Protocols(2)
- 34689757 Bove and Davis Diving Medicine 4th Ed
- A Multi-scale Model for Coupled Heat Conduction and Deformations of Viscoelastic Functionally Graded Materials
- CIV 1303-JN Presentation 2015-10-06 With Video
- 8178356813 Fitness
- Exercise Physiology
- Physical Therapy
- Schaum Continuum Mechanics Mase
- Occupational Therapy
- Drilling swab effect
- 1556432283
- ACSM's Exercise is Medicine. A Clinician’s Guide to Exercise Prescription [2009]
- Assessement in Occupational Therapy and Physical Therapy
- Physical Therapy Protocols- Knee Conditions

Sign up to vote on this title

UsefulNot usefulRead Free for 30 Days

Cancel anytime.

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Loading