Principles and
Selected Applications

Principles and
Selected Applications

Clement Kleinstreuer

Boca Raton London New York

CRC is an imprint of the Taylor & Francis Group,
an informa business

Cover design: Christin C. Kleinstreuer

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2006 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works
Version Date: 20110713

International Standard Book Number-13: 978-1-4200-0397-0 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.

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

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
and the CRC Press Web site at

To my family,
Christin, Nicole, and Joshua

Contents vii



1.1.1 Micro-to Macro-scale Systems 2
1.1.2 Solute Transport 7

1.2.1 Continuum Mechanics Axioms 18
1.2.2 Flow Field Descriptions 19 Lagrangian Description 20 Eulerian Description 21
1.2.3 Derivation Approaches 22

1.3.1 Mass Conservation 26
1.3.2 Momentum Conservation (Integral Approach) 27 Stress Tensors and Stress Vectors 30 Equation of Motion and its Special 34
Cases Force Balance Derivation 36
1.3.3 Energy Conservation 42 Heat and Mass Transfer Equations 43 Basic Heat and Mass Transfer 44
1.3.4 Turbulent Flow Equations 49 Aspects of Turbulence 49 Turbulence Scales 54 Summary of Turbulence Modeling 55
1.3.5 Solution Techniques 64 Solution Methods for Differential 67

viii Biofluid Dynamics

Equations Solution Procedures for the Navier- 67
Stokes Equations Similarity Theory 71 Integral Methods 72 Dimensional Analysis and Scaling 76

1.4.1 Modeling Approaches 79 Definitions 81 Phase Coupling 83
1.4.2 Mixture Models 88 Homogeneous and Non-Newtonian 88
Flow Models Drift-Flux Model 98
1.4.3 Separated Flow Models 99 Particle Trajectory Models 99 Species Mass Transfer 108
1.4.4 Porous Media Flow 109

1.5.1 Introduction 120
1.5.2 Principal Stresses 120
1.5.3 Equilibrium Conditions 126
1.5.4 Deformation Analysis and Stress-Strain 127
1.5.5 Simplifications 131



References 155


2.1.1 Biofluid-compartment Models 163
2.1.2 Tissue Heat and Mass Transfer 173
2.1.3 Joint Lubrication 186

Contents ix

2.1.4 Cell Transport and Microvascular Beds 192
2.2.1 Cardiovascular Transport Dynamics 197
2.2.2 The Heart 199
2.2.3 The Blood Vessels 209


References 237


3.1.1 Atherosclerotic Plaque Formation 242 A Particle-Hemodynamics Model 244 A Pathway Model for Atherogenesis 244
3.1.2 Intimal Hyperplasia Development 245
3.1.3 Thrombogenesis 246
3.1.4 Particle-Hemodynamics 247 Equations of Particle Motion 251 Near-Wall Forces 254 Hemodynamic Wall Parameters 257
3.1.5 Treatment Option: Femoral End-to-Side Graft 265
Bypass Computational Fluid-Particle 266
Dynamics Solution Model Validation 271 Results for a Distal End-to-Side 272
Femoral Bypass Novel System Design and Discussion 276

3.2.1 Aortic Aneurysms 279 Mechanisms of AAA Development 280 AAA-Wall Stress and Rupture 282
3.2.2 Treatment Option: Stent-graft Implants 283

x Biofluid Dynamics

3.2.3 Stented AAA-model Analysis 284 Basic Structure Equations 287 Numerical Method 287 Model Validations 289 Results and Discussion 290 Conclusions 295

3.3.1 Introduction 296
3.3.2 Image File Conversion Steps 297
3.3.3 A Stenosed Artery Model for Surgical Bypass 303
3.3.4 AAA-Rupture Prediction 306


References 313


4.1 THE LUNGS 322
4.1.1 Respiratory Tract Geometry 328
4.1.2 Pulmonary Disorders and Treatment Options 330

4.2.1 Kidney Structure and Functions 340
4.2.2 Fluid Flow and Mass Transfer in an Artificial 342
Kidney Model

4.3 THE LIVER 349
4.3.1 Liver Structure and Functions 351
4.3.2 Fluid Flow and Mass Transfer in a Liver Model 351


References 361

Contents xi


5.1.1 Problem Recognition and System 366
5.1.2 Types of Models and Modeling Approaches 367
5.1.3 Mathematical Representation and System 371

5.2.1 Flow in Microchannels 377 Numerical Solution Techniques 378 Microchannel Flow Effects 383
5.2.2 Controlled Nanodrug Delivery in 392

5.3.1 Nanoparticle and Microparticle Depositions in a 399
Human Upper Airway Model
5.3.2 Modeling Approach and Results 399 Numerical Method 404 Model Validations 405 Results and Discussion 407 Conclusions 418
5.3.3 Micro-drug Aerosol Targeting in Lung Airways 419

5.4.1 Aneurysms and Their Possible Repairs 422
5.4.2 A Stented Abdominal Aortic Aneurysm Model 426 Introduction 426 Theory 428 Results 434 Discussion 441


References 445

xii Biofluid Dynamics


A Review of Tensor Calculus, Differential Operations, 452
Integral Transformations, and ODE Solutions
B Single-Phase Field Equations 468
C Suitable CFD Solvers 470
D Physical Properties 475

References 478


and Fournier (1999) are typic al examples. neural network s. automatic in-vivo drug delivery. and/or medical/regulatory applications. they omitted heat transfer altogether. Many such themes a re emerging from interfacing the life s ciences and phys ical sciences. lung airflow dynamics. a t le ast fre shman-level biology (see Glossary). Example s inc lude genetic manipulations. Fung (1997). O nce th e re ader ha s ac quired a fu ndamental knowledge base and sufficient computational skills. info rmation p rocessing. r egulatory. formation of drug co mpounds. as well as the standard engineering math series (see Section 1. and industrial applications. Chandran (1992). solid mechanics an d h eat t ransfer. in-d epth monitoring of com plex bio logical processes. However . The books by Ped ley (1 980). Existing biofluids books follow a more traditional approach. In ord er to ac hieve t hese m odern objectives. Employing an inductive. Nich ols & O’Ro urke (1 990). often i n con junction with “Bio mechanics” and “B iomaterials. Prere quisites includ e junior -level course s in fluid mechanics.3. (2004) nicely stresse s ma ss transfer processes related to single and multiple cells as well as tissue and entire organ systems. usually excluding fluid-particle dynamics. non-invasive sur gical in terventions. Mazu mdar (1 992). “Biofluid Dynamics” has evolved as an importa nt res earch a nd teaching area. but. compre hensive computer simulations of the dynamics of mo lecules. cell interactions. and. of cours e. where a basic biology background is as sumed. variably repeating approach for maximum pedagogical effects. etc. a balanced phys ical- mathematical approach has to be taken. the overall goals are to provide th e r eader with a tho rough biofluids ba ckground an d go od problem-solving skills for medical. advanced themes not directly addressed in the text can be readily explored . Then there xiii . focusing mainly on flow in large arteries. The undergraduate text by T ruskey e t al. knowing the underlying mechanisms of singular events is not enough – an understanding of the multi-level intera ction processes is also a requirement for the development of new therapies and medical devices. This senior/graduate level text pr ovides a unified treatment of the fundamental principles of biofluid dynamics and shows how to apply the principles towards solutions of transf er problems in the human body and medical devices. PREFACE Due to its inherent complexity and increasing industrial a pplications.5 and Appendix A).” The focus is on “computational fluid -particle mechanics plus heat & ma ss transfer” as an integral pa rt of biome dical engineering and related fields.

and liver. the book may also serve as a text for a tw o-course sequence in “Biofluid Dynamics. discusses modeling/design aspec ts of case studie s ranging from drug delive ry via mic rochannels and lung airways to stent-graft implantation to protect aortic aneurysms..” as all books. as well as in applied mathematics and the physical/life sciences. Advancing and applying th e fu ndamentals d iscussed i n Chapter I. derivations and applica tions of the conservation laws plus closure models in a step-by-step. (ii) a list of n ecessary partia l differential equations. Special thanks go to Dr P. tissues. collections of proceedings.e.e. relying on the material presented in the previous chapters. and first-year grad uate students in the engine ering sciences. scientists. selec ting topics from Chapter I is a “pick-a nd-choose” affair. Chapter I introduces and re views basic b iological an d physical transport ph enomena as well as two-p hase flow and fluid -structure interaction aspects. g raduate cen ters. Dr Ju nemo Ko o. and biomedical engineering departments as well as the formation of n ew b ioengineering add-on pro grams.. Although this is the first rather co mprehensive tex tbook in “b iofluid dynamics. it relied on numerous sources. 2. the book will be of interest to medical students.” Thus. related books. Chapter IV focuses on biofluid dynamic aspects of the human lungs. Dr Zh onghua Li . insightful fashion. lecture series. depending up on the su bject-reviews needed for: (a) the learn ing students and (b) the subsequent chapter/section selections by the instructor. and Dr Zhe Zhang f or their . seniors . a s w ell as the author ’s colleague s. medical device companies. and microvascular beds (se e Sec t.2). Specifically. Chapter V then. Clearly. Considering the number of established medic al schools. commercial software packages.xiv Biofluid Dynamics is a number of e dited books on the market.” Chapter III deals with physical factors of two critical arterial disea se proce sses as we ll a s sur gical remedies based on computational analyses. 2. e..g. Chapters I and II fo rm the “b asic concept part. and research MDs. It will also be a useful reference for practicing engineers. res earch associates.” while Chapters III to V constitute the “applied part. (iii) an overview of suitable numerical equation solvers. Four ap pendices pr ovide: (i) revi ew mat erial o f th e en gineering mathematics needed. Chapter II provides analy ses and basic applica tions of exemplary biofluid flows in lumped-para meter compartments. This well-balanced i ntroductory text will fit nicely into exi sting and newly developed BME programs as well as engineering departments with bioengineering minors. knee joints. and invited chapters on specific topics. Worth Longest.6 outlines a study guide for the serious student. stressing c oncepts. i. this text should serve important market sectors. journal article s and reviews. i . being updated every five years or so. and start-up companies. and former graduate students. Section 1. kidneys. and (iv) tables of material property data.1) a s we ll as in compartme nts of the cardiovascular system (s ee Sect.

Ms He lena Redshaw a nd Ms Suzanne Lass andro. Danny Bluestein (SUNYSB. and contributed to several problem solutions. Sa mir Ghadiali (Lehigh Univ. CA). S tanley Rittgers (UoA. The text could not have been produced without the expertise and help of R esearch A ssociate Professor Dr Zhe Zha ng who reforma tted the manuscript. Gee rt Sc hmid-Schönbein (UC-SD.ncsu. and Zahir W arsi (MSU . MA). Also very valuabl e was the page -by-page critique from a student perspective by Robert Richter.7968). NY). Chris Bertram (UNSW . Clement Kleinstreuer Raleigh 2005 . For technical correspondence. Michael Slaughter . Dalin T ang (WPI. Jimmy Moore (Texas A & M. Dr Z hang is th e author of the S olutions Manual accompanying this book. generated the figu res and g raphs. NY). especially new homework ass ignments and future text cha nges ba sed on teac hing experience.515. as well as to the professionals in the Mechanical Engineering Department and several Libraries at Stanford University.. especially Mr . a BME Se nior at N CSU. OH). Mrs J oyce Sorensen and Mrs J oanne Self expertly typed the first draft of the manuscript. Ashim Datta (Cornell Univ . Australia). Many thanks for their support go also to the editori al staf f at CRC Press. MS) are gratefully ac knowledged as well.Preface xv contributions to Chapter V . CA). The critical comments and helpful suggestions p rovided by th e expert revi ewers S tan Ber ger (UC- or fax (919. TX). ple ase conta ct the author via email (ck@eos. PA).

and large arteries in which raised capillary The smallest vessel in the areas. and all smooth muscle cells. bile pigments. xvii . carbohydrate An organic molecule ATP (Adenosine triphosphate): “The containing carbon. prevent large changes in pH by arteriovenous anastomosis Direct either combining with H+ or by connection between an artery and releasing H+ into solution. Autoregulation cells. ADH (Antidiuretic Hormone): It acts cholesterol. of its small arteries and arterioles cardiac muscle Muscle of the heart. and degree of constriction or dilation polysaccharides. Glossary A B absorption Transport of molecules bile Fluid produced by the liver and across (epithelial) membranes into stored in the gallbladder that the body fluids. and other molecules. cholesterol. or “plaques. bronchiole The smallest of the air arteriosclerosis Any group of passages in the lungs.” carbohydrate class of molecules is autoregulation The ability of an subdivided into monosaccharides. containing diseases characterized by smooth muscle and cuboidal thickening and hardening of the epithelial cells. C atherosclerosis A common type of calyx A cup-shaped cavity in an arteriosclerosis found in medium organ (pl. the urine volume. artery wall and narrowing of its buffer A molecule that serves to lumen. and to thus regulate the rate of its consisting of striated muscle own blood flow. particular molecules in the plasma arteriole Small terminal branch of an from entering the central nervous artery that typically connects with system. and universal energy carrier of the oxygen in a ratio of 1:2:1. for the formation of thrombi. The cell. Capillary walls tunica intima are formed from are only one cell thick. contains bile salts. blood-brain barrier The structures adrenals Glands near the kidneys and cells that selectively prevent secreting adrenaline. These plaques the blood and tissue fluid occur occlude arteries and serve as sites across the capillary wall. thereby decreasing intestine. on the kidneys to promote water The bile is secreted into the small reabsorption. the myocardium. exchanges of molecules between and other lipids. a capillary. organ to intrinsically modify the disaccharides. a vein that bypasses the capillary bed. hydrogen. calices).” within the vascular system. These cells are may occur through myogenic or interconnected into a mass called metabolic mechanisms.

precursor of steroid hormones. as well as primary tissues. since reproduction is transports an ion (e. zone of the lungs. The outer layer of an cilia Tiny hairlike processes extending internal organ or body structure. The convoluted layer of gray cirrhosis Liver disease characterized matter that covers the surface of by the loss of normal microscopic the cerebral hemispheres. characterized by RNA. where gas chemoreceptor A neural receptor that exchange occurs. coordinated fashion. nucleus. from the cell surface that beat in a as of the kidney or adrenal gland. transport or secondary active clone 1. 2. identical. adequate blood flow due to heart cholesterol A twenty-seven-carbon disease or hypertension. the trachea. such as a congestive heart failure The leukocyte. DNA. It CNS (Central nervous system): That includes both facilitated diffusion part of the nervous system and active transport. Carrier-mediated a single parent cell by mitotic cell transport in which a single carrier division. sound waves. and chromosome A structure in the cell edema. toward a chemical inability of the heart to deliver an stimulus. the descendants of the pumped by either the right or the parent cell are genetically left ventricle each minute. NA+) down . 2. structure. consisting of the brain and spinal catalyst A substance that increases the cord.g. salt and water retention. which is replaced by cotransport Also called coupled fibrosis and nodular regeneration. The conducting is sensitive to chemical changes zone includes such structures as in blood and other body fluids. A term used to refer carrier-mediated transport The to cells as separate individuals (as transport of molecules or ions in white blood cells) rather than across a cell membrane by means as part of a growing organ. containing DNA and connective tissue One of the four associated proteins. It is steroid that serves as the associated with breathlessness. of specific protein carriers. cortex 1. rate of a chemical reaction cochlea The organ of hearing in the without changing the nature of inner ear where nerve impulses the reaction or being changed by are generated in response to the reaction. A group of cells derived from transport. and larger chemotaxis The movement of an bronchioles.xviii Biofluid Dynamics cardiac output The volume of blood asexual. cellular respiration The energy.. conducting zone The structures and releasing metabolic pathways in a airways that transmit inspired air cell that oxidize organic into the respiratory zone transmit molecules such as glucose and inspired air into the respiratory fatty acids. that is made according to an abundance of extracellular the genetic instructions in the material. bronchi. organism or a cell.

interaction results in the diastolic blood pressure The multiplication of the solute minimum pressure in the arteries concentration in the interstitial that is produced during the phase fluid of the renal medulla. This the resting phase of the ventricles. also contains many enzymes and is the precursor of norepinephrine. and the nucleus. . molecules or ions from regions of cytoplasm The semifluid part of the higher to regions of lower cell between the cell membrane concentration..Cell.g. diastole refers to Henle in the kidney. required for cotransport because it deductive approach Explaining is needed to maintain the steep topics or solving problems starting concentration gradient of the ion. of diastole of the heart. ECG (Electrocardiogram) (also abbreviated EKG): A recording D of electrical currents produced by Dalton’s law The statement that the the heart. It is crenation A notched or scalloped indicated by the last sound of appearance of the red blood cell Korotkoff when taking a blood membrane caused by the osmotic pressure measurement. with general theories. glucose) against its part contributed by each gas is concentration gradient. loss of water from these cells. exclusive of dopamine A type of neurotrans-mitter membrane-bound organelles. structural proteins. approach This allows sodium chloride to be dialysis A method of removing trapped in the interstitial fluid unwanted elements from the blood while water is carried away from by selective diffusion through a the kidneys. Opposite: inductive blood flows in U-shaped loops. It in the central nervous system.Glossary xix its concentration gradient while individual gas in the mixture transporting a specific molecule would exert independently. diffusion The net movement of cyto. another neurotransmitter cytoskeleton A latticework of molecule. total pressure of a gas mixture is edema Swelling due to an increase in equal to the sum that each tissue fluid. countercurrent multiplier system diastole The phase of relaxation in The interaction that occurs which the heart fills with blood. semipermeable membrane. The (e. The known as the partial pressure of hydrolysis of ATP is indirectly the gas. structural proteins in the cytoplasm arranged in the form of E microfilaments and microtubules. concepts countercurrent exchange The and/or equations to arrive at a process that occurs in the vasa specific statement or problem recta of the renal medulla in which solution. between the descending limb and Unless accompanied by the the ascending limb of the loop of modifier atrial.

Efferent nerve fibers equal signs : =ˆ means “equivalent to” conduct impulses away from the := “equal to computed value” or central nervous system. HCO3-. membrane. process of blood clot formation. for “equal to derived expression. fibrin The insoluble protein formed endothelium The simple squamous from fibrinogen by the enzymatic epithelium that lines blood vessels action of thrombin during the and the heart. within the cytoplasm. and efferent arterioles erythrocyte A red blood cell. The most discharges its secretion through a common electrolytes in the duct to the outside of an epithelial plasma are Na+. heart will pump all the blood epi. and K+. also called muscle characterized electrically ductless glands. over. able to ionize and thus carry an exocrine gland A gland that electric current. entropy statement describing the increases whenever energy is relationship between end-diastolic transformed. outer. This occurs inability of the myocardium to by invagination of the cell contract as a unit and pump blood. membrane until a membrane- It can be fatal if it occurs in the enclosed vesicle is pinched off ventricles. which returns to it. by random and continuously endocytosis The cellular uptake of changing patterns of electrical particles that are too large to cross activity and resulting in the the cell membrane. the direction of their concentration endo. facilitated diffusion The carrier- It results in decreased vital mediated transport of molecules capacity and increased airway through the cell membrane along resistance. It does not require the endocrine glands Glands that secrete expenditure of metabolic energy. volume and stroke volume of the enzyme A protein catalyst that heart. inner.xx Biofluid Dynamics efferent Conveying or transporting and forms exocrine and endocrine something away from a central glands. emphysema A lung disease in which alveoli are destroyed and the F remaining alveoli become larger. entropy A measure of the degree of Frank-Starling law of the heart The disorder in a system. transport blood away from the Erythrocytes are the formed glomerulus. hormones into the circulation fibrillation A condition of cardiac rather than into a duct. the type of tissue that covers and lines the body surfaces . the chemical reactions. elements of blood that contain electrolyte An ion or molecule that is hemoglobin and transport oxygen. location. epithelium One of the four primary tissue types.Upon.Within. it is the principle increases the rate of specific that within physiology limits.” example. gradients. Basically.

structures within the cell hematocrit The ratio of packed red membrane that regulate the blood cells to total blood volume passage of ions through in a centrifuged sample of blood. and skeletal muscles. Gates may expressed as a percentage. forms hemoglobin. bottom of a test tube during gonads Sexual glands. heparin A mucopolysaccharide found in many tissues. high-density lipoproteins (HDLs) glycogen contains more highly Combinations of lipids and branched chains of glucose proteins that migrate rapidly to the subunits than does plant starch. and carbon dioxide down their heart sounds The sounds produced by concentration gradients that closing of the AV valves of the occurs between pulmonary heart during systole (the first capillaries and alveoli. – also called animal starch – that is used medically as an produced primarily in the liver anticoagulant. but in greatest glycogen A polysaccharide of glucose abundance in the lungs and liver. membrane channels.. fluid is directly proportional to the glycocalyx Glycoproteins on the partial pressure of that gas. HDLs are carrier testis proteins that are believed to transport cholesterol away from H blood vessels to the liver. in response to a threshold level of hemoglobin The combination of depolarization).Glossary xxi G structural defects. pulmonary trunk during diastole gates A term used to describe (the second sound). ovary or centrifugation. Similar to hepatitis Inflammation of the liver. usually of the gas exchange The diffusion of oxygen valves or septum. and sound) and by closing of the between systemic capillaries and semilunar valves of the aorta and the surrounding tissue cells. (Bowman’s) capsule of the kidney Hemoglobin also serves as a weak tubules. heme pigment and protein within glomerular ultrafiltrate Fluid red blood cells that acts to filtered through the glomerular transport oxygen and (to a lesser capillaries into glomerular degree) carbon dioxide.g. plant starch in composition. e. together with the regulated (in which case they open protein globin. surface of cells. buffer within red blood cells. be chemically regulated (by heme The iron-containing red neurotransmitters) or voltage pigment that. glomeruli The tufts of capillaries in Henry’s law The statement that the the kidneys that filter fluid into the concentration of gas dissolved in a kidney tubules. and thus heart murmur An abnormal heart to offer some protection from sound caused by an abnormal flow atherosclerosis. of blood in the heart due to histamine A compound secreted by tissue mast cells and other .

organism.Under. the that is defended against by maintenance of which is the shivering and other physiological principal function of physiological mechanisms that generate body regulatory mechanisms. hypothermia A low body homeostasis The dynamic constancy temperature. or homologous antibody in order to essential. literally.xxii Biofluid Dynamics connective tissue cells that decrease in the blood carbon stimulates vasodilation and dioxide concentration to below increases capillary permeability. In active immunity a readily absorbs water. less. normal. that employ specific bonding hypertension High blood pressure. hypertension of identify and quantify a substance unknown cause or secondary in a sample. lymphocytes. concept of homeostasis provides a hypovolemic A rapid fall in blood framework for understanding pressure as a result of diminished most physiological processes. pancreas that promotes the hyperventilation A high rate and anabolism of carbohydrates. homeo Same. fat.” that stimulate the development of hydrophobic Denoting a substance clones of specific B or T that repels. in passive immunity water. between an antigen and its Classified as either primary. symptoms of inflammation and hypotension Abnormally low blood allergy. This is a condition of the internal environment. person is injected with antigens “water loving. hypertension that develops as a infarct An area of necrotic (dead) result of other. “water fearing. blood volume. Insulin acts to . It is responsible for many of the hypo. hypertonic A solution with a greater insulin A polypeptide hormone solute concentration and thus a secreted by the beta cells of the greater osmotic pressure than islets of Langerhans in the plasma. that is secreted into the blood and carried to target cells that respond I to the hormone by an alteration in immunization The process of their metabolism. pressure. depth of breathing that results in a and protein. literally. and that is repelled by. hormone A regulatory chemical hypoxemia A low oxygen produced in an endocrine gland concentration of the arterial blood.” a person is injected with hyperemia Excessive blood flow to a antibodies made by another part of the body. increasing one’s resistance to hydrophilic Denoting a substance that pathogens. hyperplasia An increase in organ size immunoassay Any of a number of due to an increase in cell numbers laboratory or clinical techniques as a result of mitotic cell division. below. known disease tissue as a result of inadequate processes. blood flow (ischemia). The heat.

Equal. a lumen The cavity of a tube or hollow solution with the same solute organ. steroids. and that is derived from tissue fluid. It is the organ responsible blood glucose and. ligand A smaller molecule that intra. phospholipids. same. tissue that contains many parallel interferons Small proteins that inhibit arrangements of collagen fibers. chemically binds to a larger intrapulmonary The space within the molecule. and hormones or environment. therefore. concentration. is in vitro Occurring outside the body. neurotransmitters can be the in vivo Occurring within the body. to for voice production. lipid An organic molecule that is ischemia A rate of blood flow to an nonpolar and thus insoluble in organ that is inadequate to supply water. low-density lipoproteins (LDLs) isotonic solution A solution having Plasma proteins that transport the same total solute triglycerides and cholesterol to the concentration. the multiplication of viruses inside It connects bones or cartilages and host cells and that also have serves to strengthen joints. to form ions. lipoproteins (containing phospholipids) secreted by type II L alveolar cells into the alveoli of Laplace.Within. protein. with which it is compared. iso. Lipids include sufficient oxygen and maintain triglycerides. It lowers surface pressure within an alveolus is tension and prevents collapse of directly proportional to its surface the lungs.Between. larynx A structure consisting of lymph The fluid in lymphatic vessels epithelial tissue. inside. osmolality. and arteries and that are believed to osmotic pressure as the solution contribute to ateriosclerosis. in the ligand for the heme in a test tube or other artificial hemoglobin. cartilage that serves as a sphincter lymphatic system The lymphatic guarding the entrance of the vessels and lymph nodes. . law of The statement that the the lungs. band of dense regular connective inter. antitumor properties. as occurs in hyaline tension and inversely proportional membrane disease when to its radius. insulin deficiency leukocyte A white blood cell. which is usually a air sacs and airways of the lungs. lower the blood glucose lesion A wounded or damaged area. for example. surfactant is absent. among. concentration and osmotic lung surfactant A mixture of pressure as plasma. ligands for specific membrane ionization The dissociation of a solute proteins.Glossary xxiii promote the cellular uptake of trachea. muscle. and aerobic respiration in that organ. produces hyperglycemia and ligament A tough cord or fibrous diabetes mellitus. Oxygen.

membrane potential The potential mucous membrane The layers of difference of voltage that exists visceral organs that include the between the inner and outer sides lining epithelium. this term is used to describe . travel to a different site.xxiv Biofluid Dynamics lymphocyte A type of mononuclear in the liberation of energy leukocyte. the cell responsible for (catabolism). They occur M on the apical (luminal) surface of macromolecule A large molecule. moles of solute per liter of tending to metastasize. a the cells of the small intestine and term commonly used to refer to in the renal tubules. mitosis Cell division in which the two macrophage A large phagocytic cell daughter cells receive the same in connective tissue that number of chromosomes as the contributes to both specific and parent cell (both daughters and nonspecific immunity. Of a tumor. submucosal of a cell membrane. RNA. solution. mmHg in the pulmonary monoclonal antibodies Identical circulation. myocardial infraction An area of meta. and DNA. but is capable of being cases) a thin layer of smooth changed by excitable cells muscle (the muscularis mucosa). those that result in energy storage myogenic Originating within muscle (anabolism) and those that result cells. humoral and cell-mediated metastasis A process whereby cells of immunity. immunity. stem that contains neural centers monocyte A mononuclear. necrotic tissue in the myocardium metabolism All of the chemical that is filled in by scar reactions in the body. transformed into a macrophage. it includes (connective) tissue. for the control of breathing and for nongranular leukocyte that is regulation of the cardiovascular phagocytic and that can be system via autonomic nerves. microvilli Tiny fingerlike projections of a cell membrane. a malignant tumor separate from lymphokine Any of a group of the tumor. parent are diploid).Change. and (in some cells. It formula weight (atomic weight for averages about 100 mmHg in the an element or molecular weight systemic circulation and 10 for a compound). It exists in all connective tissue. (neurons and muscle). mean arterial pressure An adjusted mole The number of grams of a average of the systolic and chemical that is equal to its diastolic blood pressures. antibodies derived from a clone of medulla oblongata A part of the brain genetically identical plasma cells. protein. malignant A structure or process that molar Pertaining to the number of is life threatening. chemicals released from T cells and divide to produce a new that contribute to cell-mediated tumor.

membrane potential for given ions surrounded by a double saclike when the concentrations of those membrane called the nuclear ions on each side of the membrane envelope. actions of the effectors serve to nucleotide The subunit of DNA and counteract these changes and RNA macromolecules.Glossary xxv self-excitation by cardiac and neurotransmitter A chemical smooth muscle cells. performs specialized tasks. cell The organelle. called dendrites. and the nucleus. and thymine or nephron The functional unit of the uracil). renal tubules and a vascular nucleus brain An aggregation of component that includes neuron cell bodies within the capillaries of the glomerulus and brain. O neuron A nerve cell. differs histologically from the nexus A bond between members of a physiological cell death of group. nucleotide is composed of a neoplasm A new. and genetic information of the nerve A collection of motor axons and cell. calculate the equilibrium nucleus. tissue. and a phosphate group. cytosine. nuclei. away from the cell body. endoplasmic reticulum. . and lysosomes. the type of intercellular apoptosis. as in a tumor. and a single fiber. Effectors are ribosomal RNA is produced. Nuclei within the brain are the peritubular capillaries. that function. This postsynaptic potentials. where it necrosis Cellular death within tissues stimulates the production of either and organs as a result of excitatory or inhibitory pathological conditions. consisting of a system of ribose). the site where homeostasis. consisting of a organ A structure in the body cell body that contains the composed of two or more primary nucleus. Golgi apparatus. contained in synaptic vesicles in nerve endings that is released into N the synaptic cleft. surrounded by white matter and Nernst equation The equation used to are deep to the cerebral cortex. The that conducts nerve impulses term includes microchondria. or a cell nucleus. that contains the DNA are known. Each maintain a state of balance. a sugar (deoxyribose or kidneys. or axon. sensory dendrites in the peripheral nervous system. connection found in single-unit negative feedback A mechanism of smooth muscles. activated by changes in the nucleoplasm The protoplasm of a internal environment. carry electrical charges to the cell organelle A structure within cells that body. guanine. abnormal growth of nitrogenous base (adenine. response that serves to maintain a nucleolus A dark-staining area within state of internal constancy. short branching tissues that performs a specific processes.

bone. of the esophagus. The anterior pituitary microorganism or substance. reduction reaction. the tubular digestive tract that Directly related to the osmolality involve circular and longitudinal of the solution. Peripheral resistance is more concentrated solution largely a function of the radius of through a membrane that is more small arteries and arterioles. Solutions with a pH of 7 are neutral. loss of stature. It may be accompanied pH The symbol (short for potential of by pain. stomach. those with a pH lower P than 7 are acidic. usually in the hypophysis. paralyzed. The atrium serve as the main pituitary gland is functionally pacemaker of the heart. kilogram of solvent. These gland joined to the hypothalamus autorythmic cells in the right at the base of the brain. either Dacron or Teflon patients who are inactive or based. locations along the tract. secretes ACTH. FSH. peripheral resistance The resistance osmosis The passage of solvent to blood flow through the arterial (water) from a more dilute to a system. it is the pressure muscle fibers at successive required to just prevent osmosis. divided into anterior and posterior pathogen Any disease-producing portions. the membranous and bony labyrinths number of moles of solute per of the inner ear. surrounding. seen most commonly in PET. peptic ulcer An injury to the mucosa growth hormone. such an organ. to propel the contents of the tract osteoporosis Demineralization of in one direction. osmolality A measure of the total perilymph The fluid between the concentration of a solution. as ribosomes and centrioles.Around. proportional to the fourth power osmotic pressure A measure of the of the radius of the vessel. peri. and those with a higher pH are basic. or The posterior pituitary secretes small intestine caused by acidic gastric juice.Pertaining to bone. . A small endocrine sinoatrial node (SA node). hydrogen ion (H+) concentration oxidizing agent An atom that accepts of a solution. ePTFE Synthetic graft postmenopausal women and materials.xxvi Biofluid Dynamics it is also used for some structures perfusion The flow of blood through not enclosed by a membrane. pacemaker of the heart The fastest pituitary gland Also called the depolarizing cell. tendency for a solution to gain peristalsis Waves of smooth muscle water by osmosis when separated contraction in smooth muscles of by a membrane from pure water. common use ranges from 0 to 14. LH. The permeable to water than to the resistance to blood flow is solute. and other hydrogen) used to describe the deformities and fractures. It serves osteo. PTFE. TSH. The pH scale in electrons in an oxidation. and prolactin.

The two major portal milliliters of plasma that are systems in the body are the cleared of a particular solute each hepatic portal system and the minute by the excretion of that hypothalamo-hypophyseal portal solute in the urine. dorsal surface. when stimulated by a sensory series. the production of the LH surge by respiratory zone The region of the the stimulatory effect of estrogen. derived from bone marrow cells protoplasm A general term for the called megakaryocytes. reabsorption The transport of a polymer A large molecule formed by substance from the lumen of the the combination of smaller renal nephron into the peritubular subunits. organ. The potential plasma cells Cells derived from B difference is measured in voltage. and the terminal . drained by veins that return blood renal plasma clearance The to the heart. If there is no system. They are responsible molecules composed of large for humoral immunity. Positive feedback results in repolarization The reestablishment of avalanche-like effects. Platelets colloidal complex of protein that circulate in the blood and includes cytoplasm and participate (together with fibrin) nucleoplasm. plasma is capable of difference in charge between two forming insoluble fibrin threads solutions separated by a when in contact with test tubes. the plasma The fluid portion of the blood. capillaries. which in turn is renal Pertaining to the kidneys. the fibrinogen). portal system A system of vessels receptive field An area of the body consisting of two capillary beds in that. reabsorption or secretion of that positive feedback A mechanism of solute by the nephron tubules. in forming blood clots. where blood from the first stimulus. R including nerves and ganglia. 2 to hundred amino acids are bonded 4 micrometers in diameter. as occur in the resting membrane potential the formation of a blood clot or in after depolarization has occurred. secrete large amounts of protein The class of organic antibodies. capillary bed. or part. lymphocytes that produce and prophylaxis Prevention of protection. organism. in which individual alveoli are found. that is together by pepticle bonds. Unlike serum (which lacks potential difference In biology. activates a particular is drained by veins into a second sensory receptor. lungs including the respiratory bronchioles. PNS The peripheral nervous system. the response that results in the renal plasma clearance is equal to amplification of an initial change. in which over a platelet A disc-shaped structure. or monomers. polypeptides.Glossary xxvii oxytocin and antidiuretic hormone posterior At or toward the back of an (ADH). the glomerular filtration rate. membrane.

phospholipids and proteins sinus A cavity. thus sensory neuron An afferent neuron maintaining gradients for these that conducts impulses from ions across the cell membrane. with ATPase the passage of other solute enzymatic activity that acts to molecules.g. becomes irreversible and leads to surfactant In the lungs. nucleated fibers. the supernatant when fluid. other clotting proteins have been steroid A lipid. As the cube) of differential size for the muscles contract. produced by alveolar cells that sinusoid A modified capillary with a reduces the surface tension of the relatively large diameter that alveoli and contributes to the connects the arterioles and lungs’ elastic properties. blood occurs in the respiratory skeletal muscle pump A term used zone of the lungs. and in this way help move the blood toward the heart. pressures and the colloid osmotic serum The fluid squeezed out of a clot pressures of the blood and tissue as it retracts. S smooth muscle A specialized type of semilunar valves The valve flaps of nonstriated muscle tissue the aorta and pulmonary artery at composed of fusiform. derived from removed as a result of clotting. In the liver. synergistic Pertaining to regulatory lymphoid tissues. These form the uncontrolled fall in blood steroid hormones of the adrenal pressure. veins.. It contracts in an semipermeable membrane A involuntary. permits the passage of solvent and sodium/potassium pump An active some solute molecules but restricts transport carrier. and some processes or molecules (such as endocrine organs. venules in the liver. rhythmic fashion in membrane with pores of a size that the walls of visceral organs. a mixture of death. Serum is movement of fluid out of or into plasma from which fibrinogen and blood capillaries. they squeeze the derivation of balance equations. a rapid.xxviii Biofluid Dynamics alveoli. accumulate K* within cells and extrude Na+ from cells. The balance of the Starling a sample of blood clots in a test forces determines the net tube and is centrifuged. sided carbon ring. single- their juncture with the ventricles. a flow of blood in veins. bone marrow. which in some cases cortex and gonads. Gas exchange between phagocytic cells of the the inspired air and pulmonary reticuloendothelial system. with reference to the effect of REV (Representative Elementary skeletal muscle contraction on the Volume): Open system (e. that has three six- shock As it relates to the sided carbon rings and one five- cardiovascular system. cholesterol. hormones) that have sinusoids are partially lined by . peripheral sensory organs into the Starling forces The hydrostatic central nervous system.

ventilation Breathing. and vasoconstriction A narrowing of the suppressor. thymus A lymphoid organ located in the superior portion of the anterior . Used alone. It processes T effects. vaso. as compared transplantation The grafting of tissue to the pulmonary circulation. equal to the difference between intrapulmonary pressure T and intrapleural pressure. helper. of the lung. lymphocytes and secretes systemic circulation The circulation hormones that regulate the that carries oxygenated blood immune system. or from a donor to a cardiac cycle. thrombosis The development or presence of a thrombus. the process of thorax The part of the body cavity moving air into and out of the above the diaphragm. bones of its origin and insertion. from the left ventricle via arteries total minute volume The product of to the tissue cells and that carries tidal volume (ml per breath) and blood depleted in oxygen via ventilation rate (breaths per veins to the right atrium. lumen of blood vessels as a result tendon The dense regular connective of contraction of the smooth tissue that attaches a muscle to the muscles in their walls. T cell A type of lymphocyte that provides cell-mediated immunity V (in contrast to B lymphocytes. threshold The minimum stimulus that villi Fingerlike folds of the mucosa of just produces a response. from one part of the body to systole The phase of contraction in the another part. the term atrial systole pressure difference across the wall refers to contraction of the atria. term refers to contraction of the transpulmonary pressure The ventricles. the small intestine.Pertaining to blood which provide humoral immunity vessels. the chest. lungs. cytotoxic (killer).Glossary xxix complementary or additive mediastrinum. vasa-. general circulation. thrombus A blood clot produced by the formation of fibrin threads around a platelet plug. protein fibrinogen into insoluble fibrin. through the secretion of vasa vasora Blood vessels that supply antibodies). this recipient. the minute). thrombin A protein formed in blood vital capacity The maximum amount plasma during clotting that of air that can be forcibly expired enzymatically converts the soluble after a maximal inspiration. There are three blood to the walls of large blood subpopulations of T cells: vessels.

1 . e. the triple pedagogical goals of understanding.” i. the objective is to learn to develop mathematical models. from-the-simple-to-the-complex. submodels. plotting and visualizing results. For example. it should be augmented later on by more independent work. deriving governing equations. interdisciplinary applications. requiring molecular models solved via statistical mechanics or molecular dynamics simulations. and boundary conditions with their appropriate solution techniques from available sources.. and design can be achieved only via independent practice. finding new. Traditionally. interpreting observations in a more generalized form. approach is adopted throughout this text.e. most biofluid dynamics systems are governed by continuum mechanics laws with notable micro- and nano-scale exceptions. here. and creative thinking. a “bottom-up. however. the answer to a given (flow) problem is obtained by copying suitable equations. To reach these lofty goals. Chapter I ELEMENTS OF CONTINUUM MECHANICS A sound understanding of the physics of fluid flow. exploring new concepts. improving basic submodels. or even pushing the envelope of existing solution techniques. This is called “matching” and may result in a good first-step learning experience. hard work. heat/mass transfer. In any case.g. two-phase flow and stress-strain theory as well as a mastery of basic solution techniques is an important prerequisite for studying and applying biofluid dynamics systems. skills. As always. where more general biofluid dynamics principles are recognized and assembled from special cases. approximate representations of actual biofluid flow phenomena in terms of differential or integral equations. Fortunately. The (numerical) solutions to the describing equations should produce testable predictions and allow for the analysis of biofluid system variations leading to a deeper understanding and possibly to new or improved medical procedures or devices by design..

rigorous nano. Especially on the micro-scale level. 1.. For more quantitative results. proteins.2 Biofluid Dynamics in this chapter basic transport phenomena needed in biofluid dynamics modeling (see Chapter II) are reviewed. and organelles moving in fluids and interacting/binding with pico-scale analyses will be necessary. In Chapter V the case studies are presented again in an inductive fashion. at least qualitatively. molecules. In Macro-scale Systems Of the many micro-size particles in any biological system. biochemical transport and reaction models have to be modified or new ones developed to describe. rely on such fluid-particle transport (and fluid-structure interactions) as well. cells. as for all engineering subjects. and in organs may be found in the book by Truskey et al. the reader should strive for an equal balance between physical insight and mathematical application skills (see App. and cell-cell contact forces. while intramural cells (smooth muscle cells and fibroblasts) are controlled by the pressure-induced circumferential (or hoop) stress. For example. across a cell.1 Micro. tissue. biomedical engineering and the life sciences are becoming more and more subsets of physics. such as lumen pressure. complex life forms require well-regulated transport processes. say. which in turn change the shape of the cells and trigger the production of different molecules as well as changes in gene expression. employing (future) computational fluid-particle dynamics or fluid-particle-structure interaction simulations. bio-transport processes go beyond conventional convection and diffusion. drug delivery. Repetition of some key concepts or topics from different angles is another pedagogical device used throughout the text. solute carrying. building on simple examples to arrive at more general methods or conclusions. In general. or endocytosis. 1. A nice description of diffusive mass transport within a cell. protein motoring. inside tissues). for kidney dialysis. proteins are the most fascinating ones.e.g. ions. A). Some biomechanical processes are even more complex than convection/diffusion. Biomedical devices. as already noted by René Descartes (1596-1650). ion channeling.e. binding interaction..1. After all. cell signaling. As mentioned. monolayers of cells lining blood vessels or lung airways are exposed to fluid mechanical loads. which is like a tree trunk where its branches are all the other sciences.1 BIOLOGICAL TRANSPORT PROCESSES After sufficient uptake of air. wall shear. between cells (i. while in Chapters III and IV major systems are introduced in increasing order of complexity. Thus. or blood-vessel stenting. endothelial cell shape and metabolic activity is greatly influenced by wall shear stress magnitude and temporal/spatial changes. and organs. water. i. (2004). e. biological transport can also occur via blood vessel contraction/dilation. Their multiple functions include binding of . minerals.. such exotic events. and nutrients.

Chapter I: Elements of Continuum Mechanics 3
molecules from the extracellaur fluid, nutrient transport across cells,
regulation of cell attachment, and signal transmission, to name a few. Of
the many micro-size particles most important for biofluid dynamics
considerations are platelets (for blood clotting after injury), red blood cells
(as oxygen carriers), monocytes (members of the white blood cell group),
endothelial cells (regulators lining all blood vessels), epithelial cells
(regulators lining inner organ walls, such as the lungs, stomach, intestines,
kidneys, etc.) and, of course, bacteria and viruses. Figure 1.1.1 provides a
sketch of a cell with key elements and their functions (see Alberts et al.,

Cell (Plasma) Membrane
consisting of proteins & lipids,
Cytoplasm is including cholesterol
intracellular fluid
consisting of cytosol Transport of gases and
+ organelles small molecules across
cell membrane
Vesicles for large
Lyosomes molecule transport
etc. Nucleus from extracellular fluid
Adhesion Molecules
Organelles for protein
is a filamentour
synthesis, degradation,
protein network
and export as well as
providing structural
cell division and energy
cell support as well
generation from glucose
as movement of
and oxygen (Golgi
organelles, proteins,
Fig. 1.1.1 Schematic of a mammalian (eukaryotic) cell with basic elements.
Tissue, such as muscle, nerve or connective tissue, is always
composed of different cell types. As a first approximation, fluid and
particle transport through tissue, i.e., cell junctions, could be modeled as
“flow through porous media” (see Sect. 1.4.4). The microstructure, i.e., the
capillary network of tissue is labeled the extracellular matrix (ECM). It
gives tissue strength, shape and form, regulates cells, including metabolic
activities, and it provides an aqueous environment for cell migration,
macromolecule diffusion, and protein/cell anchoring. Tissues form organs
and organ groups which execute special functions, such as cardio-vascular,
digestive, pulmonary, reproductive, or cleansing ones. For example, the
cardiovascular system (heart, blood plus vessels) distributes oxygen,
nutrients, and hormones throughout the body. In addition, waste products

4 Biofluid Dynamics

Vena Cava Pulmonary Veins

Right Atrium Left Atrium
Tricuspid A four-chamber Mitral
Valve muscle/pump under Valve
neural and hor-
Right Left Aorta
P-valve monal control
Ventricle Ventricle
Coronary Arteries valve

Pulmonary Artery (CO2)

Oxygenation of blood Pulmonary
and removal of CO2 Veins (O2)

Superior/Infe- Carotid Arteries
rior Vena Cava

• Digestion and absoption of nutrients
• Waste disposal
Synthesis and metabolism of
lipoprotein, cholesterol,
plasma, toxins, etc.

Renal Arteries
Filtration and removal of
urea and waste products

Iliac Arteries
LEGS, etc.

Fig. 1.1.2 Schematic of key organs with arterial and venous systems.

Chapter I: Elements of Continuum Mechanics 5
are carried to the liver and kidneys for synthesis and removal; when
needed, blood clots are produced and antibodies as well as white blood
cells are provided at sites of injury and infection, respectively. Figure 1.1.2
depicts major organ groups and associated blood flow conduits.
Blood vessels are large arteries, small arteries (arterioles), capillaries
and venules, as well as large and small veins (Fig.1.1.2). Arteries,
1mm ≤ D ≤ 8mm , are three-layer composites, made of the intima (i.e.,
layers of endothelial cells and collagen matrix), the media (i.e.,
extracellular matrix, elastic lamina, and smooth muscle cells), and the
adventitia (i.e., connective tissue, smooth muscle cells, and fibroblast with
capillaries and lymphatic vessels embedded). Capillaries,
1µm ≤ D ≤ 12µm , consist of an endothelium layer plus a basement
membrane. Venules are reservoirs receiving blood from where it flows via
the heart to the lungs for oxygenation (see Fig. 1.1.2). Blood itself consists
mainly of plasma (i.e., water, proteins, ions, sugars, and amino acids), red
blood cells (with hemoglobin, an oxygen-carrying protein), white blood
cells (fighting pathogens, etc.), and platelets (to mend injuries via clotting).
In contrast to the circulatory blood flow, tissue fluid (i.e., lymph) drains
unidiectionally into lymphatic vessels (i.e., lymphatics) that deliver lymph
into the venous system.

Inhalation (O2) Exhalation (CO2)

Upper Airways
(Nose/mouth, pharynx,
larynx, trachea)

(Bronchi, bronchioles,
terminal bronchioles)

O2 (RBCs) Blood
Alveoli w./CO2
Oxygenated from
blood to
pulmonary CO2
Fig. 1.1.3 Respiratory system with O2-CO2 exchange.

6 Biofluid Dynamics
While the heart (see Fig.1.1.2) is a finely tuned pump, the respiratory
system is a very complex, highly efficient O2-CO2 gas exchange apparatus
(see Fig. 1.1.3). Specifically, air inhalation convects oxygen into the
alveolar region where the millions of alveoli (i.e., tiny sacs with a surface
area of 130m2) are surrounded by pulmonary capillaries (with 115m2 in
surface area). Thus, oxygen diffuses rapidly from the alveoli through
several surface barriers into the capillary bed. The reverse occurs with the
CO2 in the blood arriving in the pulmonary artery (see Fig. 1.1.2).
Surfactants and associated surface tension forces play important roles in
governing lung mechanics.
The liver and kidneys are highly perfused because of their multitude of
vital functions. The liver is responsible for detoxification, metabolism of
cholesterol, lipoprotein and various ions, synthesis of bile, vitamins and
retinoids, as well as energy metabolism and the processing of amino acids,
fatty acids, and sugars. The kidneys are responsible for waste removal via
blood plasma filtration and water reabsorption as well as blood pressure
Additional biofluid dynamics applications, e.g., fluid motion and
traveling waves in the cochlea for hearing or fluid motion in the ear’s
vestibular system for motion sensing, are discussed in more specialized
In summary, of interest in this book are:
• fluids, i.e., air, O2, and CO2 as well as water, solvents, solutions, sus-
pensions, serum, lymph, and blood;
• particles, including molecules, proteins, cells, clots, toxins, and drugs
in the nanometer and micrometer ranges, assuming in most cases a
quasi-spherical shape; and
• structures, such as membranes, blood vessels, organ walls, and tissue
as well as implants, grafts, and medical devices.
Material transport inside and across cells is largely driven by
concentration gradients (i.e., diffusion), while fluid flow and particle
transport in lymphatic vessels, blood conduits, and airways is mainly due to
pressure gradients. Under nonpathological conditions fluid flow is laminar,
i.e., 0.1 < Re D ≤ 1800 , throughout the body. However, local obstructions,
e.g., in the upper airways or in large arteries, may cause transition to low-
level turbulence. Thus, at such low speeds and pressure differentials, all flu-
ids are incompressible. Most molecular mass and energy transfer across
natural barriers (e.g., membranes, endothelium, epithelium, etc.) are greatly
aided by very large surface areas; as, for example, in the gastrointestinal
tract, the alveolar region of the lung, the liver, and the kidneys. Thus, large
near-wall gradients and surface areas generate high transfer rates of fluids,
molecules, particles, and energy.

Chapter I: Elements of Continuum Mechanics 7
1.1.2 Solute Transport
Clearly, the proper transport of dissolved species ranging from gases
(e.g., CO2 and O2) and liquids (e.g., lymphate) to electrolytes (i.e., charged
biological molecules), nutrients, and waste products is most important for
homeostasis. Such solute transport occurs across membranes, e.g., walls of
capillaries, driven by fluid and/or osmotic pressure differentials and
concentration gradients, aided by large membrane surface areas and
selective membrane pores or channels. While these transport processes are
very slow when compared to convection, the micro-distances such
molecules have to travel and the huge (membrane) surface areas available
make diffusion, osmosis and ultrafiltration locally very effective.
Membranes are semi-permeable, i.e., only certain species or fluids may
permeate. For example, cells need proteins, nucleotides, and other
molecules for proper cell structure and function; hence, those species
cannot cross the plasma membranes.
Membrane processes can be divided into passive and active transport.
Passive transport, such as diffusion and osmosis, is driven by concentration
and/or pressure gradients. In contrast, active transport goes against
gradients and hence requires expenditure of metabolic energy from
adenosine tri-phosphate (ATP), involving specific carrier proteins.

Diffusion across a membrane. Known as Brownian motion, molecules in
gases as well as dissolved molecules and ions in a liquid (solvent) are
moving constantly around in a random fashion, as a result of their thermal
energy. If any differences in solute concentration exist anywhere in the
solution, there will be a net movement, i.e., net diffusion, along the
concentration gradient until all differences are eliminated (see Fig. 1.1.4a).
Thus, the unstable state of a species mass gradient, having a low entropy
level because of the highly ordered molecule arrangement, resolves into a
stable, disordered state (i.e., high entropy) with uniform species
concentration throughout the solution.
Larger, polar molecules (e.g., glucose) and charged inorganic ions
(e.g., sodium Na+ and potassium K+) cannot permeate the membrane and
hence need carrier proteins and ion channels formed by integral proteins,
respectively, to migrate across the cell membrane. Carrier-mediated
transport, occurring against a concentration gradient and therefore
requiring expenditure of cellular energy obtained from ATP, is a type of
active transport. A basic example are the kidney tubules (see Sect. 4.2),
where glucose is moved from the lumen (i.e., site of lower concentration) to
the blood (i.e., site of higher concentration). Similarly, all cells extrude
Ca++ into the calcium-ion-rich intracellular environment via a hinge-like
motion of the integral protein subunit (see Fig. 1.1.4b). Such active
transport carriers are referred to as “pumps”; the most important one is the

8 Biofluid Dynamics

sodium-potassium (Na+/K+) pump. Each cell has numerous Na+/K+ pumps;
for example, several million per cell as part of the tubules of the kidneys.

(a) Passive Diffusion (c) Osmosis
Plasma H2O
membrane •ٛ
net low O2


Extracellular dilute dense
environment: high O2 solute
(b) Active Transport
low high
ion conc. ion conc.
Na+ •ٛ
channel Ca++

Ion Two-step ion
Channeling transfer via
carrier proteins
Fig. 1.1.4 Solute transport modes.

As a result of a membrane’s semi-permeable function and the Na+/K+
pump actions, a cell inside is negatively charged compared to the outside.
This transverse difference in voltage is known as the membrane potential.
At equilibrium, the difference in electric charge across a membrane is
balanced by a difference in ion concentrations. Thus, the equilibrium
potential, say, ±90 milli-Volts (mV), counteracts ion diffusion, resulting in
zero net transport. Thus, as expressed in the Nernst equation for a specific

Chapter I: Elements of Continuum Mechanics 9
ion, a membrane’s equilibrium potential depends on the ratio of the ion
concentrations inside and outside a cell, i.e., in simple form:
61 C 
∆ψ k = log  o  (1.1.1)
n  Ci 
where ∆ψ k is the equilibrium potential in mV for ion k, n is the ion
valence, and C o / i are the outside/inside ion concentrations.
The gradient of the electrical potential, ∇ψ , leads to a solute flux,
where C k is the k-ion concentration and v k is the ion migration velocity
vector, i.e.,
r r
jk = v k Ck (1.1.2a)
For dilute, i.e., non-interacting, suspensions,
r DC z F
jk = − k k k ∆ψ (1.1.2b)
where Dk is the diffusion coefficient, z k is the net charge of the molecule,
F = 96,487 Coulombs/mole is Faraday’s constant, R is the universal gas
constant, and T is the temperature.
For example in 1-D, assuming equilibrium between the membrane
potential and concentration gradient, Eq. (1.1.2b) can be rewritten with
K=constant as
dc dψ
= KC (1.1.2c)
dx dx
which integrated yields Eq. (1.1.1).

Osmosis. When a membrane is more permeable to water molecules than to
solute, there will be a net movement of water by diffusion across the
membrane from the more dilute solution side, where more H2O-molecules
are concentrated, to the less dilute side (see Fig.1.1.4c). The driving force is
the osmotic pressure difference, ∆π , of the two solutions, where π depends
on the molar concentration of the solute, i.e.,
π = C M RT (1.1.3a)
where C M = C / M (i.e., C =ˆ mass concentration and M =ˆ molecular
weight of solute), R is the universal gas constant, and T is the temperature.
Clearly, for pure water π ≡ 0, which implies in view of Fig. 1.1.5 that
an osmotic pressure results:
π =γ h (1.1.3b)
where γ is the specific weight of the solution. This static pressure is needed
to keep its solvent in equilibrium with the pure water at a given

10 Biofluid Dynamics
temperature. The fluid flow rate across a barrier, i.e., membrane or porous
wall, can be estimated with Starling’s law of filtration:
jv = = L p (∆p − σ p ∆ π ) (1.1.4)

patm patm


glucose water
solution d

Fig. 1.1.5 Equilibrium between a solution and pure water due to osmotic pressure.

where j v in [L3 /(L2 T)] is the solution flux, S is the surface of the
membrane and L p its hydraulic conductivity, ∆p is the hydrostatic pressure
difference, ∆π is the osmotic pressure difference, and σ s is the osmotic
reflection coefficient, 0≤ σ s ≤1.0, i.e., σ s =1.0 implies that the membrane
(or vessel wall) is impermeable to the solvent, say, glucose molecules.
For solute diffusion across a membrane (or microvessel wall), the
Kedem & Katchalsky (1958) approach has been typically employed, i.e.,
j s = j v (1 − σ f )C s + Ps (C i − C 0 ) (1.1.5)
where jv is the solution flux of Eq. (1.1.4), σ f is the filtration reflection
coefficient, 0≤ σ f ≤1, C s ≈ (C i − C o ) / 2 is the average solute
concentration inside the membrane, P s = D eff ⁄ d [L/T] is the membrane
permeability, i.e., the solute effective diffusivity divided by the membrane
thickness, and C i , C o are the inside/outside solute concentrations. An
alternative approach, given by Patlek et al. (1964), is discussed by Truskey
et al. (2002), p. 437.

Multi-component mixtures. Solute fluxes in mixtures are described by
ji = ρ i v i (1.1.6a)

Chapter I: Elements of Continuum Mechanics 11
where the velocity of each component is
r r r
vi = v D + v (1.1.6b)
r r
with v D being the species diffusive velocity and v being the bulk, mixture
or mass-averaged velocity, i.e.,
r 1 h
∑ρ v i =1
i i (1.1.6c)
Then, the diffusive mass flux, j , can be defined as:
rD r r r
ji = ρ i ( v i − v ) = ρ i v D (1.1.7)
When additional forces act on the solute, the resulting total flux may be due
to ordinary diffusion (see Eq. (1.1.9)), an applied electric field (see Eq.
(1.1.2a), pressure differential (see Eq. (1.1.5), and thermal diffusion, i.e.,
r total r diff . r electr . r press r thermal
ji = ji + ji + ji + ji (1.1.8)
As mentioned, most frequently encountered is ordinary diffusion,
specifically in a dilute binary mixture, for which Fick’s law holds:
ji = − Dij ∇C i (1.1.9)
where D ij is the diffusion coefficient of solute i in solvent j. Equation
(1.1.9) plays a key role in the mass transfer equation derived and discussed
in Section 1.3.

The diffusion coefficient. Key in ordinary diffusion is the binary diffusion
coefficient, typically picked for a given solute, solvent, and temperature
from a table in a biochemical engineering handbook. Here we consider a
dilute binary mixture of spherical particles (i.e., solute). The approach is to
combine a 1-D force balance for a representative, randomly moving
particle with 1-D kinetic theory and Einstein’s random-walk behavior of a
diffusing particle (i.e., “translational Brownian motion”) which is based on
Fick’s law of diffusion (see Eq. (1.1.9)) and probabilistic arguments
(Probstein, 1994). Specifically,
m p 〈 v 2x 〉 = 32 k BT
42 43 123 (1.1.10)
kinetic energy thermal energy

〈 x 2 〉 = 6 Dij t (1.1.11)
where m p is the particle mass, v x is its velocity in the x-direction,
k B = 1.38 × 10 −23 JK −1 is the Boltzmann constant, T is the temperature,
<x> is the mean random particle displacement, and D is the binary
diffusion coefficient of interest. Now, with a 1-D force balance for Stokes
flow (Re≤1) we obtain:

12 Biofluid Dynamics

d 2x
mp = − FD + FC (1.1.12)
dt 2
where FD = fv x is the drag force with f being the friction coefficient,
f = 6πµ r 0 , and Fc is a random collision force which is negligible in
dilute suspensions.
In order to generate an m p 〈 v 2x 〉 -term and to use Eq. (1.1.10), we
multiply Eq. (1.1.12) through by x and expand, i.e.,
d 2 x d  dx   dx 
x 2 :=  x  −  
dt dt  dt   dt 
which yields after ensemble-averaging:
f d〈x2 〉 mp d 2 〈x2 〉
m p 〈v 〉 = 2
x + (1.1.13)
2 dt 2 dt 2
Employing Eq. (1.1.10) and solving for 〈 x 2 〉 subject to 〈 x 2 〉 =
d 〈 x 2 〉 / dt = 0 at t=0 yields the mean-squared “random walk”

6 k BT 6 m p k BT   f  
〈 x2 〉 = t− − − t
  m  
1 exp (1.1.14)
f f 2   p  

For relatively large observation times, say, t >> = ο (10 −7 s ) for
particles with 1 µm radius,
6 k BT
〈 x2 〉 ≈ t or 〈 x 2 〉 = 6 Dij t
based on Eqs. (1.2.19). Hence, for spherical particles of radius r ,
k BT k T
Dij = := B (1.1.15a, b)
f 6πµr0
Equation (1.1.15) is known as the Stokes-Einstein equation which also
holds for random motion, i.e., solute diffusion in three dimensions. For
nonspherical particles, f → f (see Happel & Brenner, 1983). Examples 1.8
and 1.21 illustrate 1-D diffusive mass transfer while Sects.5.3.1 and
discuss the use of Eq. (1.1.15).

λ. Pnueli & Gutfinger. (2002). λ.g.. Potter & Wiggert. 1992. White.54×107 molecules with a mean free path of λ=0.2). as expressed with the Knudsen number Kn = λ/L (1. 1. (2002).1. Kays & Crawford (1993). Naturally. and Arpaci et al.. Kleinstreuer. for normal devices or systems where 0.2 BASIC MOMENTUM. fluids deform continuously under (shear) stress. continuum mechanics encompasses both solid and fluid materials. AND MASS TRANSFER CONCEPTS There are a number of basic fluid mechanics texts available on an introductory level (e. as well as the inequality of entropy (cf. Gebhart et al. Second Law of Thermodynamics). may be replaced by the intermolecular length. For example. ignoring molecular effects which may be important in microfluidics (cf. 1.g. 2003. L=ρ/|∂ρ/∂y| to calculate the local Kn-number. Brenner & Edwards (1993). HEAT. which may be fixed or moving (see Sect. Panton. 1999.065 mm. They can be readily defined for an identifiable constant mass (Lagrange’s closed system) and also derived for an open system (Euler’s control volume). this review of transport phenomena relies mainly on the author’s previous text (Kleinstreuer.2. Actually. Warsi. are distributed continuously. 1999) when the continuum mechanics assumption (see Sect. 1. or Schlichting & Gersten.e. 1999). 2003) as well as Bird et al.2. Gad-el-Hak.g. or Smits. between fluid (gas) molecules is much less than an appropriate minimum length-scale L of the flow system. 2004. 1996.0×10-10 m. where λIM<<l. the continuum approach is valid as long as Kn < 0. which are treated in separate mechanics disciplines solely for historical (and feasibility) reasons.0034 mm. or L=ldevice for the global Kn-number..2. 1997. 1. and graduate level (e. actually material points. In any case.001 m ≤ . Supplementary information on heat and mass transfer may be found in Bird et al. and energy. By definition.1 as depicted in Fig. The fundamental principles of mechanics are the conservation laws of mass.1) For example. while the intermolecular spacing is only about 0. intermediate level (e. Alternatively. White. or Wilkes. 1998. which also lists modeling approaches for different flows. a 1- mm cube of standard air contains 2..Chapter I: Elements of Continuum Mechanics 13 1. Naterer (2003). Fox et al. 3. which is the average distance traveled by (air) molecules between collisions. 1991. 2000). for example. Clearly.1) is violated.1). Middleman. for liquids the mean-free-path. (1999).. i.2. The Knudsen number is appropriately used for gases because in a liquid the molecules are predominantly in vibrational modes rather than in random motion. It implies that the mean-free-path. 2000). The continuum hypothesis postulates that fluid elements. The underlying hypothesis for the conservation equations is that all materials (or media) form a continuum (see Sect. linear/angular momentum.2. (1988). λIM for water is about 3 Å. 1997.

” Clearly. ∆t gas = 10 s . a collection of micro-particles. have the implicit assumption that fluxes result instantaneously from the imposition of the corresponding gradients. Specifically.. Another potential restriction is inherent in the type of fluid and its limiting time scales. 1..2.2. that is. i. For example. not containing time as a variable.e. 2003). the time for molecule displacement (e.e.14 Biofluid Dynamics L ≤ 10 m. in Sect. and application limits for gas flows (after Kleinstreuer. 1. Sect. There are two “fluid time scales” that characterize (instantaneous) signal −10 propagation.1. Thus. in the conventional sense.g. Knudsen number regimes. modeling approaches. regions of matter are continua of properties that can be represented as differentiable functions (see Sect. i. 1. lim p = pi (1. Kn << 1. e. not only the fluid but also the dispersed phase.3). Fig.. constitutive equations. a fluid is a continuum when for all properties p convergence occurs for all points i.2) V →V ' where V is a volume around point i in the flow and V ' is traditionally “a very small volume containing a sufficient number/mass of molecules or particles.1).g.. a “fluid” as well (cf. is considered to be a continuum.

the left column represents the easier set of fluid flow problems. potential) flows where v = ∇φ . λ )) Incompressible Fluids Compressible Fluids r  ρ = ¢ . Newtonian Fluids Non-Newtonian Fluids r (air.2 Special cases of viscous fluid flows.2. t)       for liquids p > p vapor    high speed.2.e.   ρ = ρ (x. Thus. ∆t gas = 10 s and ∆t liquid = 10 s ). single- phase flow under the continuum mechanics assumption 10-4<Kn<10-1(see Fig. structures       Re < 5 × 10 5 for flat plate   l   Re >> 1. 1987).. Obviously. l s .. water or oil: µ = ¢ @ T & c) (µ→η( ∇v .2. for all practical processes.2..1) can be grouped as shown in Fig. high pressure flows Internal Flows External Flows & = ¢) (m (boundary-layer theory) Laminar Flows Turbulent Flows  Re D < 2300 for pipe flow   Random fluctuations. ω . 1. Bird et al.g. . K. n. τ turb = fct(k.1.t s )  Isothermal Flows Nonisothermal Flows r T=¢ T = T( x . 1. Notes : For gases Ma < 0.3. ε . These time intervals are so small that there is no need for basic and generalized Newtonian fluids to be concerned with relaxation effects in the constitutive equations (cf.Chapter I: Elements of Continuum Mechanics 15 −15 and ∆t liquid = 10 s ) and the elapsed time between collisions among −12 −13 molecules (e.t) (fluid properties Gr < 1 → forced convection evaluated at T) ≈ 1 → mixed convection Re 2 > 1 → free convection Fig. In order to connect Viscous Fluid Flow r Note: Inviscid (µ ≈ 0) and irrotational ( ω = 0 ) flows are idealized r (i.

1 with Fig.or ensemble-averaging methods have been employed to capture and express some microtransport phenomena on the macrotransport level. stresses as new conductivity as a boundary conditions. 1. concepts new concept etc..1 Level of Resolution in Computer Simulations Illustrated for Flow through Porous Media.. 1987). etc. or rods. permeability..e. etc. boundaries of porous BCs. etc.. gravity/ buoyancy. simulation approaches for “flow through porous media” may serve as a self-explanatory example (cf. • Molecular dynamics of spheres. back to the objective at hand: the overall goal is to find and analyze the interactions between fluid forces. material medium is a stack friction factors..2.16 Biofluid Dynamics Fig. for example. the porous molecules and particle • Property values for medium trajectories ρ . pressure.2. i. Results: Results: Results: • Fluid flow is the • Point-wise solution • Mean velocities interactions of billions of the extended from solving the of molecules as well as Navier-Stokes extended Darcy with the porous equations equation material • Pressure and shear • Hydraulic • Transport properties.1).2. drag/friction. describing equations for fluid etc. e. In turn. Molecular Dynamics Microtransport Macrotransport Needed: Needed: Needed: • Exact location and • Approximated • Porosity.2.2. Table 1. ν .2. and fluid motion. Thus micro-continuum transport laws could be derived from underlying sub-continuum molecular processes. 1. Table 1. statistical averaging schemes and volume. It should be noted that classical statistical mechanics endeavors to obtain continuum transport laws from the knowledge of discrete molecular-dynamics laws (Chandler. are not required Now.3): .g. inertia. 1. Brownian motion of submicron particles in a suspension was basic to the development of kinetic theory. the velocity vector field from which everything else can be directly obtained or derived (see Fig.or time. represented by the Boltzmann equation. For example.

1.3 Dynamics and kinematics of fluid flow: (a) force-motion interactions. γ& ~ ∇v x Fig. especially in industrial settings. Exact flow problem identification. 1. In turn. After obtaining some basic information and reliable data. Once a given fluid dynamics problem has been categorized (Fig.2. some justifiable assumptions have to be considered in order to simplify the general transfer equations. ς ~ ∇× v r (t ) rr r r r • Deformation→ ε . is one of the more important and sometimes the most difficult first task. a r r r r • Rotation→ ω. it helps to think and speculate about the physics of the fluid flow. and (b) 2-D fluid kinematics. scalar transport equations can be solved based on the velocity field to obtain critical magnitudes and gradients (or fluxes) of mass and energy.2.2).2) does the given flow system fall into. 1. and how does it respond to normal as well as extreme changes in operating conditions? (ii) What variables and system parameters play an important role in the observed transport phenomena? (iii)What are the key dimensionless groups and what are their expected ranges? Answers to these questions assist in grouping the flow problem at hand and determining the main assumptions listed below. as exemplified here: . asking: (i) What category (see Fig.2.Chapter I: Elements of Continuum Mechanics 17 (a) Cause-and-effect dynamics Surface Translation Unbalanced  and FORCES MOTION Rotation  Body Deformation (b) Kinematics of a 2-D fluid element (Lagrangian frame) Rotation At time t+∆t: y P′ Translation Deformation r At time t: P r (t + ∆ t ) Notes: r r • Translation→ v .

18 Biofluid Dynamics

Flow Assumption Consequence
∂ r r
• Time-dependence → = 0 : steady ; v = v (t ) : transient
• Dimensionality → Required number of space coordinates
• Directionality → Required number of velocity components

• Development phase → = 0 : fully developed
(s =
ˆ axial coordinate)

• Symmetry → = 0 : midplane
(n =ˆ normal coordinate)

= 0 : axisymmetry

Information on a given flow problem, in terms of the viscous flow
grouping in conjunction with a set of proper assumptions, allows for the
selection of a suitable solution technique (see Sect. 1.3.5 and App. A). That
decision, however, requires first a brief review of possible flow field
descriptions, i.e., Lagrangian vs. Eulerian framework in continuum
mechanics, and in case of probabilistic events, the type of probability
density function in the framework of statistical mechanics.

1.2.1 Continuum Mechanics Axioms
Newton’s second law of motion holds for both molecular dynamics,
i.e., interacting molecules, and continua, like blood, air, vessel walls,
tissues, bones, and implants. Such solid and fluid structures are assumed
continua, because for a fluid Kn « 1 [see Eq.(1.2.1)] and, in general, it
would be more than cumbersome to model them on a molecular (or multi-
particle) dynamics level. Focusing on solids, a body (or structure) with a
closed surface forms a continuum. When a piece breaks or tissue grows, the
surfaces of the fracture or new cells form new (external) surfaces. Thus,
the key continuum mechanics assumptions are:
(i) When a force is applied, any two neighboring body particles at one
time remain neighbors at all times.
(ii) The applied tangential or normal forces result in body stresses,
material deformations (strains or strain rates), and/or material
element motion (fluid flow); these stresses, strains, pressures (part
of the normal stress), and velocities can be defined at all points in
the body/structure.

Chapter I: Elements of Continuum Mechanics 19
(iii) Relationships exist for stress, strain, and strain rates in terms of
constitutive equations which may also take into account
temperature and electrical charge effects.
Taking the lung (or the cardiovascular circulation) as an example, it is a
matter of desired resolution and the availability of reliable data sets if such
a system is represented as a single continuum or numerous interacting
For fluids, when Kn>0.1 (see Fig.1.2.1), the behavior of representative
molecules have to be described, using statistical means or molecular
dynamics. Problems where the continuum hypothesis breaks down include
rocket-air interactions in the lower atmosphere (continuum assumption)
and later in the upper thin atmosphere (statistical description) as well as
blood-particle flows in medium-to-large arteries (c.a.) and certain flows in
micro-vessels (s.d.).

1.2.2 Flow Field Descriptions
Any flow field can be described at either the microscopic or the
macroscopic level. The microscopic or molecular models consider the
position, velocity, and state of every molecule of the single fluid or multiple
‘fluids’ at all times. In gas flow, deterministic or probabilistic molecular
dynamics models (i.e., the three conservation laws or the Boltzmann
equation plus suitable probability density functions) are employed for
direct numerical simulation. The goal of the statistical approach is to
compute the probability of finding a molecule at a particular position,
velocity, and state. Occasionally, both approaches are employed for (two-
phase) flow problem solutions. Averaging discrete-particle information
(i.e., position, velocity, and state) over a local fluid volume yields
r r
macroscopic quantities, e.g., the velocity field v (x, t ) , at any location in the
flow. The advantages of the molecular approach include general
applicability (i.e., all Knudsen numbers, see Fig. 1.2.1), no need for
submodels (e.g., for the stress tensor, heat flux, turbulence, wall conditions,
etc.), and an absence of numerical instabilities (e.g., due to steep flow field
gradients). However, considering myriads of molecules, atoms, and
nanoparticles requires enormous computer resources, and hence only
simple channel or stratified flows with a finite number of interacting
molecules (i.e., solid spheres) can be presently evaluated. For example, in a
1-mm cube there are about 34 billion water molecules, which makes
molecular dynamics simulation prohibitive, but on the other hand,
intuitively validates the continuum assumption and hence the use of the
Navier-Stokes equations. Further discussions of molecular models and
solution procedures, especially the direct simulation Monte Carlo (DSMC)
method for dilute gases, can be found in Bird (1994) as well as in the
review by Oran et al. (1998).

20 Biofluid Dynamics
Within the continuum mechanics framework, two basic flow field
descriptions are of interest, i.e., the Lagrangian viewpoint and the Eulerian
approach. Lagrangian Description
Consider particle A moving on a pathline with respect to a fixed
Cartesian coordinate system. Initially, the position of A is at
r r r r r r
ro = ro (x o , t o ) and a moment later at rA = rA (ro , t o + ∆t ) as depicted in
r r r
Fig. 1.2.4, where rA = ro + ∆r .
Considering all distinct points and following their motion for t > to,
fluid (particle) motion, steady or unsteady, can be described with the
position vector
r r r
r = r (ro , t ) (1.2.3)
where in the limit,
dr r
=v (1.2.4)
r r
d 2r dv r
= =a (1.2.5)
dt 2 dt

y Particle A r r
v A (rA , t )
rA (t o + ∆t )

Fig. 1.2.4 Incremental fluid particle motion.

Now, the material-point concept is extended to a material volume with
constant identifiable mass, forming a “closed system” that moves and
deforms with the flow but no mass crosses the material volume surface
ever. Again, the system is tracked through space and as time expires, it is
of interest to know what the changes in system mass, momentum, and
energy are. This can be expressed in terms of the system’s extensive

Chapter I: Elements of Continuum Mechanics 21
property N s which is either mass m , momentum mv , or energy E. Thus,
the key question in mathematical shorthand is: “How can we express the
fate of N s , i.e., DN s Dt = ? ” Clearly, the material or Stokes derivative
D Dt ≡ ∂ ∂t + v ⋅ ∇ follows the closed system and records the total time-
rate-of-change of whatever is being tracked (see Sect.1.3).
Obviously, we cannot use d/dt from “particle dynamics” for this task.
For example, focusing on the fluid density, dρ/dt is ambivalent, implying
either a density change with time of a specific fluid element at a point
(x,y,z) in the flow field or a rate-of-change of density at the point because of
new fluid elements traveling during dt past point (x,y,z). Thus, to avoid
misconceptions, we briefly revisit the various time derivatives, i.e., ∂ ∂t
(local), d dt (total of a material point or solid particle), and D Dt (total
of a fluid element). Their differences can be illustrated using acceleration:
• a x ,local = , where u is the fluid element velocity;
r dv
• a particle = as employed in solid particle dynamics; and
r r
r Dv ∂v r r
• fluid = := + ( v ⋅ ∇) v
element Dt ∂t
r r
alocal a convective
The derivation of Dv Dt is left as a HW problem in Sect. 1.7. Eulerian Description
In the Eulerian framework an “open system” is considered where
mass, momentum, and energy may readily cross, i.e., being convected
across the control volume surface and local fluid flow changes may occur
within the control volume over time. The fixed or moving control volume
may be a large system/device with inlet and outlet ports, it may be small
finite volumes generated by a computational mesh, or it may be in the limit
a “point” in the flow field. In general, the Eulerian observer fixed to an
inertial reference frame records temporal and spatial changes of the flow
field at all “points,” or in case of a control volume, the transients inside and
fluxes across its control surfaces.
In contrast, the Lagrangian observer stays with each fluid element or
material volume and records its basic changes while moving through space.
While Sect. applies the Lagrangian approach directly to evaluate
particle trajectories, the Sect. 1.3 employs both viewpoints to describe
mass, momentum, and heat transfer in integral form, known as the
Reynolds Transport Theorem (RTT). A surface-to-volume integral

22 Biofluid Dynamics
transformation then yields the conservation laws in differential form in the
Eulerian framework.

1.2.3 Derivation Approaches
There are basically four ways of obtaining specific transport equations
reflecting the conservation laws. The points of departure for each of the
four methods are either given (e.g., Boltzmann equation or Newton’s
Second Law) or derived based on differential (REV) balances.
(i) Molecular Dynamics Approach: Fluid properties and transport
equations can be obtained from kinetic theory and the Boltzmann
equation, respectively, employing statistical means. Alterna-
r r
tively, ∑ F = ma is solved for each molecule using direct
numerical simulation.
(ii) Integral Approach: Starting with the Reynolds Transport Theorem
(RTT) for a fixed open control volume (Euler), specific transport
equations in integral form can be obtained (see Sect.1.3).
(iii) Differential Approach: Starting with the Generalized Transport
Equation (Sect. 1.3), the mass, momentum, and energy transfer
equations in differential form can be formulated. Alternatively,
the RTT is transformed where in the limit the field equations in
differential form are obtained. Also shown in Sect.1.3 is a third
approach where differential balance equations for a representa-
tive elementary volume (REV) are derived.
(iv) Phenomenological Approach: Starting with balance equations for
an open system, transport phenomena in complex transitional,
turbulent, or multiphase flows representing biomedical systems
are derived, largely based on empirical correlations and
dimensional analysis considerations.
These approaches were already alluded in Fig. 1.2.1. After a brief
illustrative Lagrange vs. Euler example, approaches (ii) and (iii) are
employed in Sect. 1.3.

Example 1.1: Lagrangian vs. Eulerian Flow Description

In the Eulerian fixed coordinate frame, river flow is approximated
as steady 1-D, i.e.,
( )
v(x ) = v 0 + ∆v 1 − e − ax
which implies that at x=0, say, the water surface moves at v0 and
then accelerates downstream to v ( x → ∞ ) = v 0 + ∆ v . Derive an
expression for v = v(v 0 , t ) in the Lagrangian frame.

Chapter I: Elements of Continuum Mechanics 23

r r
Recall: v = dr dt or in our 1-D case
= v(x ) = v 0 + ∆v 1 − e − ax ( )
Solution: Separation of variables and integration yield
x dx t
∫ = ∫ dt
0 (v 0 + ∆v ) − ∆ve −ax 0

so that

1  ∆v
ln 1 +
a  v0

1 − e − ax  = (v 0 + ∆v ) t )

Now, replacing the two x-terms with expressions from the v(x)-
1  v − v0 
equation, i.e., x = − ln1 −  and e −ax = 1 − v − v 0 , we
a  ∆v  ∆v
can express the Lagrangian velocity as

v 0 (v 0 + ∆v )
v (t ) =
v 0 + ∆v exp[− a (v 0 + ∆v )t ]

Given typical parameter values, the velocities v(x) and v(t) in the
Eulerian and Lagrangian frames are graphed below.


2 2
1.9 1.9
1.8 1.8
1.7 1.7
v(t) [m/sec]
v(x) [m/sec]

1.6 1.6
v0 = 1.0 m/s 1.5 v0 = 1.0 m/s
∆ v = 1.0m/s ∆ v =1.0 m/s
1.4 -1
1.4 -1
a =1.0 m ( ) or a = 1.0 m ( ) or
1.3 -1 1.3 -1
0.5 m ( ) 0.5 m ( )
1.2 1.2
1.1 1.1
1 1
0 2 4 6 8 10 0 2 4 6 8 10
x [m] t [sec]

24 Biofluid Dynamics
A Transport Theorem in Differential Form. Conservation of mass,
momentum, and energy can be expressed as one equation in differential
form, labeled by Kleinstreuer (1997) as the generalized transport equation
(GTE) (see Sect. 1.7 HW problem for derivation):
∂ ( ρψ ) r
• + ∇ ⋅ ( ρψ v) + ∇ ⋅ J − ρφ = 0 (1.3.1)
where ψ is the specific quantity being conserved. Specifically, ψ ≡ 1 (for
r 2
mass), v (for momentum), or ( u˜ + v ⁄ 2 ) (for internal plus kinetic
rr r rr r
energies); J is the (molecular) flux, i.e., J = 0, − T, or q − T⋅ v ,
r r r
respectively; and φ is the source term, i.e., φ = 0, g, or g ⋅ v ,
respectively. The total or Cauchy stress tensor T and the heat flux vector
q are discussed below. An alternative derivation of the field equations in
differential forms is outlined in Sects. 1.3.1-1.3.3.

A transport theorem in integral form. Consider N to be an arbitrary
extensive quantity of a closed system or a moving material volume (see
Sect. Examples include a piston-cylinder assemblage with
constant gas mass or a (yellow) vapor cloud of fixed mass moving through
the air; N could be the system’s mass, momentum, or energy. The first task
is to express in the Lagrangian sense the material derivative, i.e., the total
time-rate-of-change of N. Clearly, with m system ≡ m = constant , we can
• N ≡ m system → = 0 (conservation of mass)
r r r
r Dv r
• N ≡ (mv) system → m = ma = ∑ Fsurface + ∑ Fbody
(conservation of momentum or Newton’s Second Law)
DE & &
• N ≡ E system → = Q −W
(conservation of energy or First Law of Thermodynamics)

Now, the conservation laws in terms of DN/Dt are related to an open
system (Eulerian frame) in terms of a fixed control volume (C.V.) with
material streams flowing in and out, i.e., across the control surfaces (C.S.),
and possibly accumulating inside. Descriptively,

Chapter I: Elements of Continuum Mechanics 25

Total time - rate - of  Local time - rate - of   Net efflux of ( ρη ), i.e., 
     
change of system  = change of N/V ≡ ρη  + material convection across
property N  within the C.V.  control surfaces (C.S.) 
     
or mathematically,
DN ∂ r r
• = ∫∫∫ ρηdV + ∫∫ ρηv ⋅ dA (1.3.2)
Dt closed ∂t C .V . C .S .
which is the Reynolds Transport Theorem (RTT), for a control volume at
N 1r
rest, where the specific property η = :=  v
m  e
(i) For a moving control volume, v is replaced by
r r r
v relative = v fluid − v C.V. . The operator ∂ ∂t acting on the first term
on the R.H.S. has to be replaced by d dt when the control volume
is deformable.

(ii) For a non-inertial coordinate system, e.g., when tracking an
accelerating system, e.g., a rocket, ∑ F is expressed as
r r r r
ma abs = m(dv dt + a rel ) , where ma rel accounts for noninertial
r r r
effects, so that ∑ F − m a rel = m d v dt , and finally the extended
momentum equation reads (i.e., N ≡ (mv ) system )
r r d  r  r r
∑ F − ∫ a rel dm =  ∫ vρdV  + ∫ vρ (v rel ⋅ nˆ ) dA (1.3.3)
C .V . dt  C .V .  C .S .
Applications of the RTT are given in Sects. 1.3.1 -1.3.3 and as
Examples 1.2 and 1.3.

(iii) There are a few sequential steps necessary for tailoring the general
RTT [cf. Eq. (1.3.2)] toward a specific flow system description and
solving the resulting integral equations (cf. Examples 1.2 and 1.3 in
Sect. 1.3.2).
(1) Identify the extensive quantity Nsystem, e.g., the mass of the
identifiable material, linear or angular momentum, or total
energy. As a result, the specific property of the closed system,
η = N msystem , is known.

26 Biofluid Dynamics

(2) Determine DN Dt for each conservation case, i.e., mass,
momentum, or energy.
(3) Select a “smart” control volume and determine if:
• The control volume is fixed, or moving at VC.V.=¢, or
accelerating arel=¢, or accelerating and rotating;
• the flow problem is steady or transient;
• the fluid properties are constant or variable;
• the inflow/outflow velocity fields are constant, i.e., uniform,
or a function of inlet/exit space variables; and
• the resulting integral balance equations are decoupled.
(4) Set up the momentum equation, i.e., force balance, for each
coordinate direction.
(5) Solve the volume and/or surface integrals (cf. Integration
Tables), watching the inflow/outflow sign convention, i.e.,
ˆ “-” and OUT =
IN = ˆ “+”, and check the results for correctness.
Numerous sample problem solutions further illustrating the use of the
RTT may be found in Fox et al. (2004) and other undergraduate texts.

1.3.1 Mass Conservation
Taking N ≡ m and hence η =1, the RTT [Eq. (1.3.2)] reads
∂ r r
0= ∫ ρdV + ∫ ρv ⋅ dA (1.3.4)
∂t C .V . C .S .
Employing the Gauss divergence theorem
r r
∫∫S0 ρv ⋅ dS = ∫∫∫V0 ∇ ⋅ (ρv ) dV (1.3.5)
r r r r r
where ∇ ⋅ (ρv ) ≡ div(ρv ) and v ⋅ dS = v ⋅ nˆ dS , Eq. (1.3.4) can be rewritten
 ∂ρ r
0 = ∫∫∫  + ∇ ⋅ (ρv ) dV (1.3.6)
C .V .  ∂t 
Since dV ≠ 0 , the continuity equation follows as
∂ρ r
• + ∇ ⋅ (ρv ) = 0 (1.3.7a)
Note, for steady flow,
∇ ⋅ ( ρv ) = 0 (1.3.7b)
and for incompressible flow,
r r
• ∇⋅v=0 or div v = 0 (1.3.8a, b)

Chapter I: Elements of Continuum Mechanics 27
Mass balance derivation. Alternatively, Eq. (1.3.7a) can be derived with a
more physical approach, based on a fluid mass balance over an open
system, say, a cube (cf. Fig. 1.3.1).
• Global mass balance : ∑ m& in − ∑ m& out =
∂t ∆V
• In a 1-D differential form:


ρu x ∂ρ ρu x + ∆x
Fig. 1.3.1 One-dimensional fluid mass balance.

[(ρu ) x
− (ρu ) x + ∆x ∆y∆z =

• With Taylor’s truncated series expansion, f
x + ∆x
= f + ∆x ,
x ∂x
we obtain
∂ (ρu ) ∂ρ
− ∆x∆y∆z = ∆x∆y∆z
∂x ∂t
• Adding the other two net fluxes ( ρv ) and (ρw) in the y- and z-
direction, respectively, and dividing by the arbitrary volume ∆V ,
∂( ρu ) ∂ (ρv ) ∂ (ρw) ∂ρ
− − − =
∂x ∂x ∂x ∂t
∂ρ r
• + ∇ ⋅ (ρv ) = 0 (1.3.7a)

1.3.2 Momentum Conservation (Integral Approach)
r r
Taking N≡ mv and hence η ≡ v , the RTT reads

V . stress tensor (see Eq.V. sketch: control volume: t w C. where vjet.2: Force on Disk Moving Axially into an Axisymmetric Jet A steady uniform round jet hits an approaching conical disk as shown.(1.28 Biofluid Dynamics r Dv r r r ∂ r r r r ∂t C∫. Approach: RTT (mass balance and 1-D force balance) Solution: r v (a) Mass Conservation: 0 = ∫ ρv ⋅ dA = − ∫ ρv C .9) m = ma = Fbody + Fsurface = ρ v dV + v Dt r r r rr r rr F where body ≡ ρ ∫ bf dV F and surface ≡ ∫ ⋅ dA . C .3.3.S . Ajet.S .3. ρv ⋅ dA (1. A jet rel dA + ∫ ρwdA Aexits where v rel ≡ v fluid . and fluid layer thickness t are given. vdisk. = v jet − (− v disk ) = v j + v d and Aexit ≈ πD t .V. C . .V. diameter D. Note: For a rotating material volume.S .S . Thus.V. ∫C .V . w Aj Note: • vd=vC. T being the total T C . C . moving C.S . the fluid angular momentum per unit r r volume η ≡ ρ (r × v ) has to be considered. Example 1.V . we have ∂ r r r r rr r r ∫ ρ v dV + ∫ v ρv ⋅ d A = ∫ ⋅ dA + ∫ ρf b dV T (1.S. angle θ .10) ∂t C . w θ Assumptions: Constant averaged velocities and properties.13).V . C .v C. The law of conservation of angular momentum states that the rate-of-change of angular momentum of a material volume is equal to the resultant moment on the volume. = ˆ t “streamline” vrel = x w D/2 vj vd D vjet+vdisk Rx C. Find the force exerted on the disk.

e.e.. Thus.S. 0 ≤ y ≤ 3h 4  h Sketch: Approach: u∞ m& b p0 • Integral mass & momentum y h 3h balances (RTT) FD u( y ) x C.S . i. p0 m& b . C . η = v . based on a measured velocity profile downstream from the body. If the disk would move away with v d = v j . i.3: Force on a Submerged Body Find the drag per unit length on a submerged elliptic rod of characteristic thickness h.V. DN Dt : = Fsurf = − Rx (resulting force) r r r r x-component ∑ Fsurf = ∫ vρv rel ⋅ dA − R x = ∫ uρv rel dA C . u∞  y u( y ) = 1 +  . Example 1. vrel=0 and hence Rx=0. − Rx = − v rel ρv rel Ajet + w 2ρAexit sin θ v rel A jet where w = so that Aexit A j sin θ  ( • R x = ρA j v j + v d )2 1 − πtD    Comment: The resulting fluid-structure force R x = (mv x )out − (mv x )in ..Chapter I: Elements of Continuum Mechanics 29 (v j + v d )A j ( ) ∴ -ρA j v j + v d + ρ (πD t )w = 0 f w= πD t (b) Momentum Conservation: r r r N = (m v )s . R x is the result of a change in fluid flow & & momentum.

i. In expanded form rr rr rr T = − pI + τ (1.10) is a new unknown that rr constitutes a closure problem. rr r the stress vector τ relates to the symmetric second-order tensor T as r rr rr τ = nˆ ⋅ T = T ⋅ nˆ .11a) r where p is the thermodynamic pressure (or the static pressure for ∆v = 0 ).V. T has to be related to the principal r variable v or its derivatives. For any coordinate system. fixed C. The RTT treats it as a “black box” and elegantly obtains FD=Fpressure + Ffriction via “velocity defect” measurements. symmetry. and τ is the stress tensor.e. rr rr I is the unit tensor.2. 2-D.S .30 Biofluid Dynamics Assumptions: steady state. (ρg ) ..V.3. is very complex. constant properties. the rr Cauchy or total stress tensor T in Eq.3. 4 ∫ 1 + h dy 0 15 9 ∴ m& b = 6 ρhu ∞ − ρhu ∞ = ρhu ∞ 4 4 (b) Momentum Conservation (x-momentum): r ( ) − FD = ∫ v x ρv ⋅ dA := m& v x exit + m& v x b − m& v x inlet C . (1.1 Stress Tensors and Stress Vectors r r While in most cases fb is simply gravity per unit volume. especially behind the submerged body. 3h 9  − FD = −u ∞ ρu ∞ (6h ) + u ∞  ρhu ∞  + 2 ρ [u ( y )] dy ∫ 2 4  0 9 • FD = ρhu ∞2 in [N/m] 8 Comment: The fluid flow structure inside the C. Solution: (a) Mass Conservation: m& in = m& b + m& out 3h r v u∞  y 0= ∫ ρv ⋅ dA = − ρu ∞ 2(3h) + m& b + 2 ρ C ..3. 1.S ..

Without tensor symmetry.. i. i.. internal friction.7) that: .3. In tensor notation the total stress tensor is written for a Cartesian coordinate system as: Tij = − pδ ij + τ ij (1. j + v j . Expanding the total derivative in Cartesian coordinates  ∂u ∂u ∂u   ∂x ∂y ∂z   r  ∂v ∂v ∂v  r r r dv = dsi = ∇vds (1.g.3.e. an infinitesimal fluid element ( ∆∀ → 0 ) would spin at r ω → ∞ .e..11b) where − pδ ij is the isotropic part (e. This insight leads for fluids to the postulate ( ) τ ij = fct ε ij where ε ij = 1 2 ( ) v i .3. 1. fluid statics and inviscid flow) and τ ij is the deviatoric part for which a constitutive equation has to be found. the fluid element is experiencing during dt can be expressed as the “rate-of-deformation” tensor. starting with a fluid element displacement from point P (with v at r r t) to point P′ ( v + dv at t + dt ) a distance ds apart.. Physically. or deformations. The relation between τ ij and ε ij (plus vorticity tensor ζ ij ) can be more formally r derived.3.e.13) ∂x j It can be readily shown (see Sect. i.12a)  ∂x ∂y ∂z   ∂w ∂w ∂w     ∂x ∂y ∂z  the spatial changes.Chapter I: Elements of Continuum Mechanics 31 where nˆ is the normal (unit) vector. T ij = T ji .12b) ds ∂x j It can be decomposed into strain-rate tensor ε ij <symmetrical part> and the vorticity (or rotation tensor) ζ ij : ∂v i = ε ij + ζ ij (1. τ ij =ˆ τ ( i = force )( j = direction ) represents a force field per unit area as a result of the resistance to the rate of deformation of fluid elements.i is the rate-of-deformation tensor. r dv r ∂v i r = ∇v := (1.

5 provides further discussions of stresses. ∇ ⋅ v = 0 (see Eq.. and some fluid flows in microscale devices. rarefied gases.4: Shear Stress in Simple Couette Flow Provide a physical explanation and a mathematical derivation for τ xy in unidirectional flow. (a) Tangential force.16)  ∂x j ∂x i    and γ& ij ≡ 2ε ij is called the shear-rate tensor.15) where  ∂v ∂v j  2µε ij ≡ τ ij = µ  i +  := µγ&ij (1.14) where the viscosity coefficients λ and µ depend only on the r thermodynamic state of the fluid. ζ xz = −ζ zx = ω y . for Newtonian fluids rr r rr rr τ = λ (∇ ⋅ v )I + 2µε (1.3.13)..32 Biofluid Dynamics • ζ yx = −ζ xy = ω z . e. represent element distortion ∂v i ∂v j • The shear-rate tensor γ&ij ≡ + and hence the 1 in ∂x j ∂xi 2 ε ij = 12 γ&ij is mathematically necessary in order to match Eq. which is not the case for non-Newtonian fluids. keeping a viscous fluid in motion: . Specifically. thus.3.3. 2ω = ∇ × vr = ζ 2  ∂x ∂y  • ε ii := γ&ii ≡ ∂v i ∂xi indicate volume change (dilation) • ε ij := 12 γ&ij . and ζ zy = −ζ yz = ω x . where 1  ∂v ∂u  r r ω z =  −  . etc.(1. Section 1.3.3. For incompressible flow. Stokes suggested that τ is a linear function of ε . Fpull. (1. and deformations in (solid) continua. strains.8)) and the total stress tensor reduces to Tij = − pδ ij + 2 µε ij (1. i ≠ j .g. Example 1. MEMS. rr rr Now.

Chapter I: Elements of Continuum Mechanics 33 u0=¢ . µ) Asurface y x y or anywhere in the fluid dFx dAy dFx τ yx = dA y (b) Resistance to fluid element deformation: dAsurface dFviscous • Physics: δFi surface force ∆y τ ij = lim δ =ˆ ∆θ δAi → 0 A j unit area y dFv ∆θ τ= ~ dAs ∆t ∆x x • Geometry: ∆s ∆u ⋅ ∆t tan ∆θ ≈ ∆θ = = ∆y ∆y u+∆u ∆s=∆u⋅∆t y • Combining both: ∆u τ~ ∆θ ∆y ∆y and in the limit with the proportionality factor µ. ∆x du • τ yx = µ u dy in unidirectional flows. Fpull =– Fdrag  dp  τ wall We observe:  = 0 Asurf Fdrag  dx  u(y) τ wall = (ρ. .

3.17) ∂t C . (1.3.11). Transforming the surface integrals. leads to the equation of motion. the axial momentum diffusion.3.3. is negligible when compared to all other . For steady two-dimensional (2-D) flows.5). which is valid for any fluid r rr ∂ ( ρv ) rr r • + ∇ ⋅ ( ρvv ) = −∇p + ∇ ⋅ τ + ρg (1.S .34 Biofluid Dynamics Comments: • u(y) is linear as in simple Couette-type flows.V . the N-S equations read in rectangular coordinates: ∂u ∂v + =0 ∂x ∂y ∂u ∂u 1 ∂p  ∂ 2u ∂ 2u  u +v =− +ν  2 + 2  + g x ∂x ∂y ρ ∂x  ∂x ∂y  (1. ∂t Dt r Equation (1. (1. using Eq.3.2 Equation of Motion and its Special Cases Returning now to Eq. form the Navier-Stokes (N-S) equations.19) together with the continuity equation. 1. τ yx = τ wall = ¢.3. C .16): r ∂v r r r r + (v ⋅ ∇ )v = − ∇p + ν∇ 2 v + g 1 • (1. ∇ ⋅ v = 0 .3.18) ∂t (i) Navier-Stokes Equations: For constant fluid properties and using Eq. the RTT for linear momentum transfer can be written as rr rr ∂ r r r r   r r ∫ ρvdV + ∫ vρv ⋅ dA = ∫  − pI + τ  ⋅ dA + ∫ ρgdV (1.3.V .3. • This result for τ yx is a very special case of Eq.10) and including Eq.S .3.3.16).19) ∂t ρ r r ∂v r r Dv r where + (v ⋅ ∇ )v ≡ = a total and ν = µ ρ is the kinematic viscosity. (1.20a-c) ∂v ∂v 1 ∂p  ∂2v ∂2v  u +v =− +ν  2 + 2  + g y ∂x ∂y ρ ∂y  ∂x ∂y  (ii) Prandtl’s Boundary-Layer Equations: For thin wall boundary layers where the ratio of boundary layer thickness to flat-plate length δ ⁄ l « 1 and v « u because Re l » 1 . (1.   C . (1. ν ∂ 2 u ∂x 2 .2. C .

25) 2 ρ where v and p are locally averaged quantities along a representative streamline and the z-coordinate extends against the direction of gravity.7). Eq. for two points on a streamline the total energy per unit mass is balanced.3.3.23) applied in 2-D to a representative streamline along s yields ∂v s ∂v 1 ∂p + vs s + + g sin θ = 0 ∂t ∂s ρ ∂s 123 (1. Thus. i. the steady 2-D boundary-layer equations read (cf.3.3.22) ∂t (iv) Euler’s Inviscid Flow Equation: For frictionless flow.23) Dt which is the Euler equation. (iii) Stokes Equation: r r When the viscous forces are dominant.3. (1. r ∂v r • ρ = −∇p + µ∇ 2 v (1.19) reduces to r Dv r • ρ = −∇p + ρg (1.21a-c) ∂x ∂y ρ ∂x ∂y 1 ∂p • 0=− ρ ∂y The last equation implies that the pressure normal to the wall is constant.24) through by ∂s and integrating yields for steady incompressible inviscid flow the Bernoulli equation: v2 p • + + gz = C (1.. the (v ⋅ ∇ )v term in Eq.. p(x) of the outer flow is imposed onto the thin shear layer.3.20b).3.6 as well as Schlichting & Gersten.e. (1.19) is negligible and the Stokes equation is obtained. Equation (1.4 and 1. (1.3. only ∂p ∂y is relatively significant (see HWA in Sect. 1.3.3. Figs. Re<1.3. 1.3.e. . i. 2000 or Kleinstreuer.Chapter I: Elements of Continuum Mechanics 35 terms in the x-momentum equation (1..e.24) = ∂z / ∂s Multiplying Eq. Thus. i. In the y-momentum equation. 1997): ∂u ∂v • + =0 ∂x ∂y ∂u ∂u 1 ∂p ∂ 2u • u +v =− +ν 2 (1.

(1.3.9)): r r r Dv ρ = ∑ f surface + ∑ f body (1. One application of 2 Eq. Fig.3.2 Closed system.3.28) Dt From the Eulerian point of view.29a)  ∂t  pressure viscous buoyancy For example. external surface and body forces accelerate an REV of mass m. viz. the net pressure force per unit volume ∆∀ in the x-direction is z REV=∆x∆y∆z r r a. viscous.e. Fig. (1.: r  ∂v r r r r r ρ  + (v ⋅ ∇ )v  = f net + f net + f net (1. Employing rectangular coordinates and an incompressible fluid. i.3. for a given system where v2=0 (e.2.3.. . 1. p1 is the static pressure at point 1.3.3). considering a 1-D force balance (cf.3.2 and Eq. we have ρv12 p 2 = p1 + (1.36 Biofluid Dynamics v12 p1 v2 p + + gz1 = 2 + 2 + gz 2 (1.27) is the Pitot tube where v1 = 2( p2 − p1 ) ρ . and ρv1 2 is the dynamic pressure at point 1. v r f pressure r f drag y r f gravity x Figure 1.3.g.27) 2 where p2 is the total or stagnation point pressure. and gravitational forces. a wall) and g∆z ≈ 0 . 1.26) 2 ρ 2 ρ For example.3. so that we can write per unit volume in the Lagrangian framework (cf.3. the REV turns into a control volume and we record local and convective momentum changes due to net pressure.3 Force Balance Derivation A more physical approach for deriving the (linear) momentum equation starts with a force balance for a representative elementary volume (REV). 1.. accelerating material volume.

Eq. f net . ∆V pressure ∂x ∆x∆y∆z ∂x or in 3-D r  ∂p r ∂p r ∂p r  f net = −∇p = − i + j + k  (1. (1.3.Chapter I: Elements of Continuum Mechanics 37 ∂f p f net = fp − fp =− ∆x pressure x x + ∆x ∂x p∆A =− ∂p ∆y∆z ∆x = − ∂p and with f p ≡ . 1.29b) pressure  ∂x ∂y ∂z  Similarly. ∆V viscous ∂z ∆x∆y∆z ∂z or in 3-D  ∂τ xx ∂τ yx ∂τ zx  r  + +  i +  ∂x ∂y ∂z  r rr  ∂τ xy ∂τ yy ∂τ zy  r f net = ∇ ⋅τ =  + + j + (1.29c) viscous  ∂x ∂y ∂z   ∂τ xz ∂τ yz ∂τ zz  r  + + k  ∂x ∂y ∂z  Now.14) reduces to rr τ = µ (∇v + ∇v tr ) r r • so that Eq. f net =− ∆z = − .3.3. the net viscous force per unit volume in the x-direction reads: fv z + ∆z fp fp x x + ∆x fg z x fv z Fig.3.3. (1. ∂f v f net = fv − fv z + ∆z =− ∆z viscous z ∂z τ∆A ∂τ ∆x∆y ∂τ and with f v ≡ . with Stokes’ hypothesis for incompressible Newtonian fluids.3 Control volume for 1-D force balances.29a) reads (cf. N-S equation) .

Middleman.3. Also. etc. The boundary conditions are u ( y = 0) = U and u[ y = h( x )] = 0 . 1997. for example. etc.5: Parallel and Nearly Parallel Flows The assumptions of steady laminar fully-developed or unidirectional flow lead to analytic solutions of the Navier-Stokes equations. Problem: Load-Carrying Capacity of a Planar Bearing (l . v ≈ 0 h1 y h(x) h0 • p = p (x ) only x ρ. W . 1997). The trick for quasi-unidirectional flows is to solve the parallel flow equations and then incorporate the nonparallel wall effects via (the) boundary condition(s). flow in a slider bearing. flow in a slightly tapered tube. with v ≈ 0 ∂u ∂v ∂u ∂p ∂ 2u + =0 reduces to ≈ 0 . 2001. Stokes I and II flows. levitated disk flows. Kleinstreuer.38 Biofluid Dynamics r  ∂v r r r r • ρ  + (v ⋅ ∇ )v  = −∇p + µ∇ 2 v + ρg (1.21a-c). i. There is another group of analytical solutions where the wall flow is nearly parallel. Bird et al.e. 1998. However. (cf. (1. coating flows. (cf.19)  ∂t  Example 1.. hence 0 = − + µ 2 ∂x ∂y ∂x ∂x ∂y Clearly. With the nondimensionalizations .3. 0 = − because the normal velocity component is ∂y small and its second derivatives are negligible. θ ) Sketch: Assumptions: fixed plate • steady laminar 2-D flow • Re ( = Uh1 / ν ) << 1 and θ v << u . the starting equations are Eqs. Kleinstreuer. Examples include Poiseuille flow.. µ U • constant fluid properties l • no end effects Solution: Based on the assumptions. the viscous forces are dominant and axial diffusion is ∂p negligible. among others). Couette flows.

Example 1. ˆ y )  = −  1  −    −       η  2  dˆx  η  η   The pressure gradient dˆp / dˆx can be obtained from the mass balance h( x ) η m& lub e = ρW ∫ u (x. blood flow in arteries and air flow in the lung. h0. where æ ≡ is typically known. uˆ ( ˆy = η ) = 0 . 1990 or Zamir. l. e..Chapter I: Elements of Continuum Mechanics 39 xˆ = x / h1 . pˆ = ph1 / (µU ) and with η = h( x ) / h1 (h1 − h0 ) where h(x ) = h1 − x .. Pulsatile flow in a tube has been analyzed by Womersley (cf. we have after double integration and invoking the boundary conditions uˆ ( ˆy = 0 ) = 1. “start- up” due to a suddenly applied pressure gradient is discussed below. and m· can be postulated. ˆ   Q • dp = 12  1 − æ  .g.. pˆ (xˆ = 0 ) = p 0 . 2000). vˆ = v / U . Nichols & O’Rourke. pressure waves. in turn. shut-down. such as start- up. y )dy = ρWUh1 ∫ uˆdyˆ = ρQ 0 0 where W is the width and Q is the volumetric flow rate. inserted into the simplified x- l momentum equation.  2η η 3  UWh1 dˆx 2 Now various pressure distributions p(x) based on design parameters h1.e. yˆ = y / h1 . i. uˆ = u / U . Problem: Sudden Pressure Applied to Horizontal Flow in a Pipe . p(x) determines the lubricant velocity field and the load-carrying capacity l L = ∫ p(x ) cos θdx 0 within the stated restrictions. have to be considered for industrial flows. etc.6: Transient Flow in a Tube Most naturally occurring flows are unsteady.  l  and pˆ  xˆ =  = p0 :  h1   ˆy  η 2  dˆp   ˆy  ˆy   2 • ˆ u ( ˆ x . or transient effects.

t ) = 0 and ∂u ∂r r = 0 = 0 for all t. ∆p   r  2 = 4 µu max 1 −     L   r0       r  2 where max 1 −  u    = u ss is the Poiseuille solution.0. Now there are various techniques for solving the PDE available: direct numerical solution. t ). t ) transient With the assumption that − ∂ p ∂ z = ∆ p L = ¢ at all times. B) reduces to ∂u 1  ∂p  ν ∂  ∂u  = −  + r  ∂t ρ  ∂z  r ∂r  ∂r  subject to u (r .40 Biofluid Dynamics Sketch: Assumptions: • transient laminar 1-D flow valve is • constant fluid p1 suddenly opened properties Reservoir • ∂p/∂z=¢ for t > 0 p(z) r p0 z L Solution: r With the postulates v = [u (r . t = 0 ) = 0. − ∂p ∂z = ( p1 − p0 ) L =¢ and ∂ ∂θ = 0 . continuity states ∂u =0 ∂z Thus the z-momentum equation (see App. product solution leading to two ODEs.0].   r0     . or superposition with the postulate u (r . u (r = r0 . t ) = u (r ) ss + u (r .

5.0.0 u(r.75 0.25 0.5 10.0. uˆ (rˆ = 1) = 0. 8.5 0. Schlichting & Gersten (2000). and = 0.41.t) 1. 1993) ∞ J (λ r ˆ) uˆ = −8 ∑ 30 n e −λnt 2ˆ n =1λ n J 1 (λ n ) The eigenvalues λ n =2. and tˆ = t ρr0 ∂uˆ 1 ∂  ∂uˆ  =  rˆ  ∂tˆ rˆ ∂rˆ  ∂rˆ  ∂uˆ ( ) subject to uˆ tˆ = 0 = rˆ − 1.Chapter I: Elements of Continuum Mechanics 41 Knowing u (r ) . 2. among others.52. Graph: ∧ ∧ t = 0. we obtain for the transient part with ss µ uˆ = u tran / u max .0  umax Additional exact and approximate solutions based on the reduced Navier-Stokes equations may be found in Kleinstreuer (1997). and Panton (2005). t ) / u max .01 .65 for n=1.05 0. 3 and λ n +1 = λ n +p for n ≥ 3 are the zeros of the Bessel function J0. Bird et al. u (r . . The profiles shown illustrate the evolution of the axial velocity. (2002).0 r= 1 ∧ r= 0 0. ∂rˆ rˆ = 0 The solution to this Sturm-Liouville problem is (see Özisik. rˆ = r / r0 .9 1. Nichols & O’Rourke (1998). throughout the pipe.1 .

+ q conv . a . ∫ T : ε dV . was added as part of the internal energy contribution. Employing the momentum conservation law.change Rate of work done  of E in material  on the volume by   Net heat flux        =  + across the total   volume moving with  surface plus body  material volume the flow  forces    rr rr The frictional work term. dropped out and the heat flux V r r r r q t = q cond . i. Eq. A).3.e. has to be 2 separately expressed.of . i. D Dt ( ) r (r ) r (E kin + Eint ) = ∫ vr ⋅ τr dS + ∫ ρ vr ⋅ f b dV + ∫ qr t ⋅ dS (1. Etotal=Ekinetic+Einternal:= v + mu~ .31) S V S Time .30) Dt closed system m r2 Typically. the RTT reads [Recall: First Law of Thermodynamics]: DEt ∂ r r ∂t C∫. i.e.18) is scalarly multiplied through by v (see App. r (1.3.S. r r Dv D  ρ r 2   rr  r r r ρv ⋅ =ˆ  v  =  ∇ ⋅ T  ⋅ v + ρf b ⋅ v Dt Dt  2    Integration over the material (or control) volume V yields rr r DEkin Dt   r r ( = ∫  ∇ ⋅ T  ⋅ vdV + ∫ ρ fb ⋅ v dV ) V V rr r rr r rr rr The integrand  ∇ ⋅ T  ⋅ v ≡ ∇ ⋅  T ⋅ v  − T : ε . + q rad .3.3. so that with the Gaussian     divergence theorem rr r r rr rr DEkin Dt   r r ( = ∫  T ⋅ v  ⋅dS − ∫ T : ε dV + ∫ ρ v ⋅ fb dV   ) S V V where the integral rr r r rr r r  T    r rr ( ) ∫ ⋅ v  ⋅dS ≡ ∫  T ⋅ v  ⋅n dS = ∫ v ⋅ τ dS S  S  S Hence.e... ∫C .. In order to complete the energy conservation law.3 Energy Conservation Taking N ≡ E total and hence η ≡ e t .rate . the L.42 Biofluid Dynamics 1.S .H.V . ⋅ dA ≡ Q& − W& = ρ et dV + ρet v (1.

3. ( ) r r + ∫ q t ⋅ dA + ∫ ρqˆint dV C .34) ∂t r with dh = c p dT .1)] as a point of departure.3. e.35b) Dt has the same form as the species mass transfer equation Dc ------.32) Again. (1.Chapter I: Elements of Continuum Mechanics 43 distributed internal heat source term.3..V .35a) ∂t ρc p where α = k ⁄ ( ρc p ) is the thermal diffusivity.1 Heat and Mass Transfer Equations A simpler derivation of the energy equation resulting in a directly applicable form employs the GTE [Eq.3. C .3. r Setting Ψ ≡ h = u˜ + p ⁄ ρ <enthalpy per unit mass>. (1. may have to be V added.36) Dt where D is the binary diffusion coefficient and S c possible species sinks or sources (Bird et al. Eq.V .3.3.5).g.= α∇ 2 T + S T (1. .3.32) can be rewritten in differential form as rr r ρ Det  ∂e r  ( r r )   r ≡ ρ  t + (v ⋅ ∇ )et  = ρ f b ⋅ v + ∇ ⋅  T ⋅ v  + ∇ ⋅ q t (1.S . q = − k∇T after Fourier and µΦ = τ ij ∂v i ⁄ ∂x j . Eq. C ..= D ∇ 2 c + S c (1. using the divergence theorem (1.31) reads for a stationary control volume: DEt Dt = ∂ r r r r ( ) r rr ∫ ρet dV + ∫ ρet v ⋅ dA ≡ ∫ ρ v ⋅ f b dV + ∫ v ⋅ τ dA ∂t C .V . or h = c p T when c p =constant.S . C . Finally.3. 1.3. where T is now 2 2 the temperature. 2002).3.S . This scalar equation DT -------. (1.33) Dt  ∂t  1 r2 1 r2 r where et = v + u~ ≅ v + cv T and qt = − k∇T + ….3. ∫ ρqˆ int dV .(1. we obtain for thermal flow with constant fluid properties the heat transfer equation in the form: ∂T r µ + (v ⋅ ∇ )T = α∇ 2T + Φ (1. and φ ≡ νΦ <energy dissipation due to viscous stress> yields ∂ r r ( ρh) + ∇ ⋅ (ρvh ) = −∇ ⋅ q + µΦ (1. J ≡ q <diffusive heat flux>. C .

.1. an analytic solution of the heat transfer equation and hence an expression for the Nusselt number can be obtained. i.35) and (1.44 Biofluid Dynamics 1.36).3.7: Convection Heat Transfer in a Tube Considering hydrodynamically and thermally fully developed flow.7-1c) ∂r r =0 In order for the flow to be “thermally fully developed” there has to be a generalized temperature profile Θ(r ) for which . Sketch: Assumptions: qw=¢ • Poiseuille flow • axial convection balanced by radial conduction (or radial Tw(z) thermal diffusion) T(r.7-1b) ∂r r = r0 and ∂T =0 symmetry (E.e. (1. Eq. ∂u ∂z = 0 and Peclet number Pel = Rel Pr = u l / α > 100 . two sample problems are provided employing analytical solutions. Additional examples may be found in Bejan (1995) and Bird et al.3.2 Basic Heat and Mass Transfer Applications In order to highlight the use of Equations (1.3. Example 1.3.z) • constant fluid properties r • constant wall heat flux z Tm(z) Solution: Based on the stated restrictions.35b) reduces in cylindrical coordinates to ∂T α ∂  ∂T  • u (r ) = r  (E1. (2002).3. provided that q wall =¢ is the thermal boundary condition.7-1a) ∂z r ∂z  ∂r  subject to ∂T k = qw [or T (r = r0 ) = Tw (z )] (E1.

z )dA .. so that ∂T ∂Tm = =¢ ∂z ∂z Now. in the tube yields r0  dT  • qw = ρu max C p  m . • NuD=4.e. with q w = h(Tw − Tm ) =¢ and Nu D = where α = ρC hD k p and D=2r0..7-2)   r0   dz   r ∂r  ∂r    Direct integration yields u max  dTm  3 2 r4 r 2  • T (r .7-4) 4  dz  k Now. u max = 2u (E1. the governing PDE can be rewritten as   r  2  dT  α ∂  ∂T  • u  max 1 −    m  = r  (E1.364 .7-3) α  dz  16 16r02 4  The “mixing-cup” temperature Tm can now be computed from its definition and then (Tm – Tw) can be formed. 11 u max  dTm  2 Tm − T w =  r0 96 α  dz  An energy balance for a cylindrical control volume. i. Newton’s law of cooling). where Tm = T w − Tm Au A from the condition f = f ∂T ∂Tw ∂T ∂T = −Θ w +Θ m ∂z ∂z ∂z ∂z i. dTw / dz = dTm / dz . we obtain 1 With Θ = . z ) = Tw (z ) −   r0 + − (E1. πr02 ∆z .Chapter I: Elements of Continuum Mechanics 45 ∂Θ =0 ∂z Tw − T ∫ u (r )T (r . with all gradients being constant for q wall =¢ and Tw − Tm = ¢ (cf.e.

Sketch: Assume: Concept/Approach: x • Transient • Transient 1-D 1-D binary diffusion Equation h c1 diffusional • Separation of or c 0 c mass transfer Variables method initial=0 • Constant • Neglect higher-order diffusivity term Membrane or "Cell" Solution: r With the implicit assumptions of v ≡ 0 and c=c(x. Eq.8-2) leads to 2 1 . (1.36) can be reduced to 2 ∂c ----. c ( 0. among others) c ( x. Ozisik. with initial solute concentration c ( x.x=0) and then (b) apply the mathematical solution to a cell of characteristic extension h. 0 ) = 0 .t) as well as the mass flux j(t. surrounded by a reservoir with solute concentration c 0 . Find: (a) c(x.e.= – λ 2 DT ⇒ T ( t ) = Ae –λ t 2 dT dt . 0 ) = 0 . where c 1 > c 2 at all times.= D ∂-------c- • (E1. m0 = m ( t → ∞ ) .t) only. 1993.8-1) ∂t ∂x 2 Subject to c ( x.-------- d X ------.dT-----.. separated at t=0 by a thin flat membrane.3. t ) = X ( x ) ⋅ T ( t ) (E1.46 Biofluid Dynamics Example 1. 0 ≤ x ≤ h . . t ) = c 1 and c ( h. Calculate the ratio of diffusing solute mass m(t) to the solute mass at equilibrium.:= – λ 2 = cons tan t DT dt X dx 2 so that -----.= --1.8: Diffusive Mass Transfer across a 1-D Membrane or Cell Consider two solutions of constant concentrations. t ) = c 2 Postulating the product solution (cf. i.

t ) = c 1 +  ---------------- x + --- 2 1 ∑ --n. t ) = c 1 +  ---------------- x + ∑ B n sin ( λ n x ) e  h  n=1 where nπ λ n = -----. n=1. c2 – c1 c ( x. using the Fourier series expansion. t ) = c 1 +  ---------------- x + ( B sin λx + C 1 cos λx )e – λ Dt 2  h  Invoking the B..[ ( – 1 ) c 2 – c 1 ] nπ (a1) Finally.e.Chapter I: Elements of Continuum Mechanics 47 and d 2 X.8-4) ∂x x=0 i.= – λ X ⇒ X ( x ) =   B sin λx + C 1 cos λx for λ ≠ 0 2 dx As a result. t ) ≈ c 1 +  ---------------- x – --.3.. yields 2 n A n = -----. 2  A1 x + A2 for λ = 0 -------.2. with the boundary conditions for the λ = 0 case.C. h The initial condition..8-3b) M (a2) The mass flux j --------..( c 1 + c 2 ) sin  ------ exp ( – π Dt ⁄ h ) 2 2 2  h  π  h (E1. ∞ c2 – c1 • c ( x..[ ( –1 ) n c2 – c1 ] (E1. approximately . (E1..8-3a)  h  π n=1 sin  --------- exp ( – n π Dt ⁄ h ) nπx 2 2 2  h  Retaining only the terms for n=1 because the others being proportional 2 to n vanish rapidly with time. we have c2 – c1 πx • c ( x.diffusing into the membrane at x=0 is 2 L T • j 0 = D ∂c ----..s for λ ≠ 0 yields ∞ c2 – c1 – λ n Dt 2 c ( x.

… (E1. ∞ ) dx 0 ∞ 8 1- ∑ 2 2 2 = 1 – ----2.exp ( – n 2 π 2 Dt ⁄ h 2 ) n  h  n = 1.8-5) h h (b1) The solute concentration in a 1-D cell exposed to a solution reservoir for which c1 = c2 = c0 we have from Eq.8-6) (b2) The solute mass ratio for the cell can be expressed as h m ( t ) ∫0 Cc ( x.8-7) m0 h ∫ Cc ( x. (E1.= --------------------------------. so that m ( t ). ≈ 1 – ----2. 3. except for very short intervals (n>1)-terms can be neglected.3.36) can be reduced to 2 ∂c ----. ∂ c = D -------2.exp ( – π Dt ⁄ h ) (E1.+ u ∂c ----.35b) where α =ˆ D . .sin --------. (E1. the same solution method (E1. 8 2 2 • ---------.8-2) can be applied when solving simplified forms of Eq. t ) = C 0 1 – --- π ∑ --. (1. 5. Eq.8-8b) m0 π Notes: • For 1-D convection and diffusion.3. t ) dx • ----------. (E1.8-3a) ∞ 4 1  nπx • C c ( x.48 Biofluid Dynamics D 2D 2 2 • j 0 ≈ ---. (1.= D ∂-------c- ∂t ∂x ∂x 2 which can be transformed with ξ = x – ut and η = t to 2 ∂c -----. 3.8-8a) π n n = 1. ---- 2 exp ( – n π Dt ⁄ h ) (E1. … Again.( c 2 – c 1 ) – ------.( c 1 + c 2 ) exp ( – π Dt ⁄ h ) (E1.8-1) ∂η ∂ξ • Now. 5.

4). 1. here. at least approximately. i. etc. the natural transition begins with the amplification of Tollmien-Schlichting (T-S) waves. these flow perturbations (e. at least two problems have to be solved: (i) determination of the onset of transition and (ii) modeling of the influence of the transition region on the specific flow field. etc.3.. Linear stability theory allows the calculation of characteristics of these T-S waves as eigen-solutions of the linearized Navier-Stokes equations (cf. body surface roughness. energy losses.) are not damped out. numerous "irregularities and disturbing sources" exist a priori in free shear flows and on surfaces of wall-bounded flows. drag reduction with laminarized flows. flow instabilities in form of vortices. air flow in the larynx and trachea. and airflow in the upper respiratory system are frequently turbulent. Although turbulence is dissipative. Changes from laminar-to-turbulent and/or turbulent- to-laminar flows are inadvertently of importance in numerous applications including particle-hemodynamics in locally occluded blood vessels. the basic knowledge and mathematical skills gained from laminar flow studies have to be significantly extended in order to evaluate. After a brief discussion of laminar flow instabilities and turbulence characteristics.. γ . i. 1. In fact.e. In general. blood flow in highly stenosed arteries.Chapter I: Elements of Continuum Mechanics 49 1. For all practical purposes. turbulence is most desirable for high throughput and especially for enhanced mixing or dispersion. Examples then deal with turbulent flow applications. in two-dimensional flows. wall non-uniformities or vibration.e. requiring higher pressure gradients to overcome extra drag. A crude evaluation of the transition region may be obtained with an intermittency function. transition from laminar to turbulent flow and to compute quantitatively turbulence effects in a variety of practical applications.4 Turbulent Flow Equations The vast majority of natural and industrial flows are turbulent. where γ > 1 near the end of the transition region. Fig.3. and acoustic or particle environments all can contribute to the onset of transition and the development of (locally) turbulent flow fields. . Thus.. the linear and nonlinear development of initially small amplitude T-S waves into large amplitude disturbances that break down into a chaotic motion (cf. Free stream fluctuations. γ =0 in laminar flow and γ ≥ 1 in turbulent flow. applied to the turbulent shear stress τ total = τ laminar + γτ turb . approaches to turbulent flow modeling are summarized. Taylor. At sufficiently large Reynolds numbers. and transition control of fighter jets.. wall roughness effects. Warsi. free- stream turbulence. they amplify and start to interact with the entire flow field. 1999).3. just the opposite.4. Specifically.g.or Goertler-type vortices.1 Aspects of Turbulence Transitional flows. urine flow.

associated with the large-scale motion.g.4 Flat plate laminar/transitional/turbulent boundary-layer characteristics (after White.. where uτ ≡ (τ w ρ ) 2 is the friction 1 velocity. hairpin structures. they are believed to be responsible for the apparent stresses or Reynolds stresses and for most of the energy-containing motion. The traditional approach for modeling turbulence is the use of the Reynolds-averaged Navier-Stokes equations. This closure problem.. The coherent structures contain a significant fraction of the turbulence kinetic energy and they are associated with the transport of scalar quantities in the flow field. inhomogeneous. the detection of various types of spatially coherent structures in important engineering flows has spawned a new class of turbulence research. the time-averaging or time-smoothing process of these equations produces dominant (apparent) stresses that make turbulence closure models necessary. On a smaller scale. and nonstationary. A (Von Karman) vortex street behind a cylinder at high Reynolds numbers may serve as an example. e. nonlinear. there are also detached incoherent motions. However. Thus. ν 3 ε 4 . Coherent structures. The development of a general theory for turbulent flow. and it is largely dependent upon its origin and the given flow configuration. Turbulence modeling. it is chaotic.50 Biofluid Dynamics δ~x 6/7 Stable laminar δ~x 1/2 u~y 1/7 u∞ B-L flow *Tollmien-Schlichting waves y u u~y *3-D vortex breakdown 2 τturb>>τlam v *Turbulent spot increase viscous sublayer Recrit Returb x Fig. are basically recognizable patterns with characteristic orientations that recur throughout the flow field. Because of the nonlinear terms. there exists a hierarchy of eddy scales with the smaller one proportional to ν uτ . which are responsible for the energy dissipation. as discussed below. i. thus. it is necessary to model all structures requiring flow system-specific rather than universal relationships. is hampered by the fact that turbulence is very complex.3. three-dimensional. although treated as a continuum phenomenon. the necessary specification of some auxiliary equations in . In wall turbulence for example. Turbulence exhibits random velocity fluctuations. Turbulent flows. ( ) 1 eddies of the Kolmogorov length scale. Since averaging occurs over all the turbulence dynamics simultaneously. and horseshoe vortices of random position and variable shape and size. In contrast. coherent substructure phenomena exist such as longitudinal vortices. 1991). expressed in terms of the eddy dissipation rate ε.e. 1.

cf = cf . For example. v ≡ v (a) Continuity: div v + div v' = 0 With div v = div v . of course. Basic averaging axioms include f + g = f + g .e.3. the classical point of departure is Reynolds’ decomposition of the instantaneous.e.37) into a time-averaged v . i. for the reason of simplicity. − ρ v' v' . it contains nine. 1. where f and g are random functions and c is a constant. a vector. the Reynolds-averaged Navier-Stokes equation appears: • ∂v ∂t  p ( ) + ∇ ⋅ (vv ) = −∇  + ν∇ 2 v − ∇ ⋅ v ′v ′ (1. Substituting the decomposition into the basic transport equations (cf.. is usually solved with eddy- viscosity models or. we obtain div v = 0 (1. now more frequently.3) yields after averaging. and assuming that f = f . the Navier-Stokes equations (Sect. time-averaging.3.2) can be transformed as follows. (b) Momentum: ∂v ∂v ′ + + ∇ ⋅ (vv ) + ∇ ⋅ (vv ′) + ∇ ⋅ (v ′ v ) + ∇ ⋅ (v ′ v ′) ∂t ∂t  p  p′  = −∇  − ∇  + ν∇ 2 v + ν∇ 2 v ′ ρ ρ After each term is averaged or filtered (cf. i.3. 1 t2 • f = ∫ f dt (1. Note.. for example. • The dyad. 1. due to symmetry.39b) ρ Note the following: • Subtracting the last two equations yields an equation for the flow perturbations. f g = f g . actually. Warsi 1999). i. For example. v . where v could be a tensor. or a scalar. with Reynolds-stress transport models.. is called the Reynolds stress tensor. dependent variable.Chapter I: Elements of Continuum Mechanics 51 order to match the number of new unknowns.3. six different.3. so-called apparent or turbulent stress components.38) ∆t t1 the turbulent transport equations. and a randomly fluctuating component v ′ .39a) which.e. implies that div v ′ = div v ′ = 0 because v ′ ≡ 0 . in rectangular coordinates . • v = v + v′ (1. the arrow over r velocity vectors is now omitted. Sect. In any case.

a solution to the closure problem.4). 1.3. (1. an important measure of turbulent flows.3. the turbulent incompressible flow equations read ∂ vi • =0 (1. i.3. where the local Reynolds number approaches zero and a "viscous sublayer" is formed (cf.16)] . we know that total τ ijturb >> τ ijlam everywhere in turbulent flow fields except in the near-wall region.3.41) ∂xi • ∂ vi ∂t + vj ∂ vi ∂ xj =− 1∂p ρ ∂ xi +ν ∂ 2 vi − ∂ ∂ x j∂ x j ∂ x j v i' v 'j ( ) (1.40) is the turbulence kinetic energy.e.3. Eq. where in an extension to Stokes' hypothesis [cf.5(v i ' ) 2 (1. • In Cartesian coordinates. Fig.3.3. revolves around the turbulent stresses [cf.e.. the eddy viscosity model (EVM).43) ∂x ∂y and ∂u ∂u ∂u 1 ∂ p ∂  ∂u  • +u +v =− + ν − u ' v'  (1. Eq.3.3.52 Biofluid Dynamics ρ u' 2 ρ u' v' ρ u' w' ρ v' v' = ρ u' v' ρ v' 2 ρ v' w' ρ u' w' ρ v' w' ρ w' 2 • The sum of the diagonal in the form: 1 2 2 ( ) u ' + v'2 + w'2 = k := 0. i. (1.42)] τ ijturb = − ρ v' i v' j (1. using the convention of summation over repeated indices..45a) One prevailing idea is the Boussinesq concept. • For two-dimensional incompressible turbulent boundary-layer flows ∂u ∂v • + =0 (1.44) ∂t ∂x ∂y ρ ∂x ∂y ∂y  Turbulence modeling.42) 142 rr 4 3 14rr243 ∇⋅τ laminar ∇⋅τ turbulent • While τ ij = τ ijlam + τ ijturb throughout a flow field.

turbulence kinetic energy. for example. is a rather complex function. It is called a "zero-equation" model because no additional PDE is required for the solution of a particular turbulent flow.3. . Other recent advances in turbulence modeling have not matured enough and/or are too costly to be considered in complex turbulent systems analyses.46) 2 and the dissipation function ∂ v'i ∂ v' j • ε ij = 2ν t (1. controlled by sets of strange attractors. the turbulent eddy viscosity. is being investigated. turbulence time scale. Examples include the log-profiles and power-law profiles. It is also evident that there is not one turbulence model that can compute a whole range of flows to acceptable engineering accuracy. turbulence energy dissipation. Alternatively.45b) ρ ∂x j Here. to model turbulence effects. Depending upon the level of turbulence complexity. if two extra PDEs for the turbulence kinetic energy k = (v' i )2 1 • (1. which is typically associated with the type of flow geometry. traditional closure concepts are and will be used for today's and tomorrow's engineering computations. wall roughness. and each particular model requires "fine tuning" with empirical data that depend on the flow type and system geometry. ν turb . system geometry. Since the exact simulation of the turbulence dynamics for complex flows is still unattainable. and so on. For example. where the spatial and temporal resolutions have to be fine enough to represent the smallest and fastest turbulent flow phenomena.Chapter I: Elements of Continuum Mechanics 53 τ ijturb ∂v i • = ν turb (1. ν t = C k 2 ε the approach is called a "two-equation" model. The equations for k and ε. or direct turbulence simulation on parallel supercomputers. forming the k- ε turbulence model as well as the high-Reynolds number RNG k-ε and “low”-Reynolds-number k-ω models are discussed in the next subsection.3. ν t = ν t (eddy size or length scale.).3. Prandtl's Mixing Length Hypothesis (MLH) is an algebraic equation for the eddy viscosity of the form ν t [l ( y )] . velocity field gradients. we may need several ensemble- averaged PDEs in addition to the mean flow equations. In general. the cumbersome solution of any differential equation can be avoided when streamwise velocity profiles for specific turbulent flows are directly postulated. etc. fluid viscosity. Examples include the theory of dynamic systems where chaotic yet deterministic behavior of solution trajectories. On the other hand.47) ∂ xk ∂ xk ( ) are solved to form.

3. • heat diffusion time scale t h = L2 α ( ) where α = k ρc p . body length. fluid properties (i.54 Biofluid Dynamics 1. wall friction velocity. mean vorticity. Time scales: L • convection time scale t c = where L is a typical system U dimension and U is a mean flow velocity. 1 ν 3  4 • Kolmogorov length scale l k =   or with Re L = UL ν . • mixing length l m = κy or l m = κy[1 − exp(− y A)] where κ = 0. and hence it is . mean temperature. ν 1 ν  2 • Kolmogorov time scale tk =   where the eddy ε  2 U3  u'  dissipation rate ε ∝ or ε ∝ ν   with u' and l' are eddy L  l'  velocity and size scales. ν ..2 Turbulence Scales Before a specific turbulence model is discussed. it may be of interest to consider first characteristic time and length scales of turbulent flows. Numerous volumes have been written on this subject matter.e. or mean velocity.  ε    lk − 34 ~ Re L L The brief discussion of the complexities of turbulence should imply that turbulence modeling is a very challenging. shear layer thickness or duct radius.41 and A being the Van Driest damping length.4. never-ending task. here. the viscous diffusion time is about equal to the eddy convection time l' u' . ρ . Such scales are usually constructed from representative flow system parameters. Length scales: • viscous wall length lτ = ν uτ = O(l' ) where uτ ≡ τ w ρ is the friction velocity. etc. L2 • viscous diffusion time scale t v = . k. such as approach velocity. and cp).

3 Summary of Turbulence Modeling Despite some frustrating failures (in the past)." deterministic turbulence modeling approaches could be grouped as follows: 1. 1. Numerous empirical coefficients have to be tuned to match system-specific turbulent flow patterns (cf. turbulence kinetic energy k. pseudovorticity ω.4.3. more flexibility.48) ρ 3   ∂x j ∂x i   It requires the calculation of one or more turbulence quantities (e.). dissipation function ε.ω .) ρ Zero-equation models. 1999).Chapter I: Elements of Continuum Mechanics 55 only appropriate to provide here a brief summary of some viable engineering approaches. by Boussinesq (1877) and Kolmogorov (1942). etc.g. Ferziger & Peric. i. 4. which implies reduced needs for computational resources and measured "constants. Reynolds Stress Modeling (RSM) where transport equations (PDEs) for each of the important components of the turbulence stress 1 tensor. near-wall eddies. (cf. require the solution of an algebraic equation. plus the turbulence energy.τ . which requires time and length scales small enough to resolve turbulent fluctuations and tiny. etc. Pope.3.ε .e. Large Eddy Simulation (LES) with subgrid scale (SGS) modeling where the coherent. k = u i′ 2 . 2000). based on the postulates. Direct Numerical Simulation (DNS) of turbulence. k . computational fluid dynamics (CFD) modeling of turbulence is advantageous because of lower turnaround time. length scale l.. ρ u' i u' j = τ ij . Eddy Viscosity Modeling (EVM). lower cost. In other words. time scale τ. 2. 3. or constitutive equations. which are then combined and directly related to the "turbulent eddy viscosity" µt νt = = fct(l . DNS does not require any turbulence model. and occasionally higher accuracy when compared to physical modeling. such as Prandtl's Mixing Length Hypothesis (MLH) and the Van Driest extension for wall damping. In a decreasing order of complexity. large-scale structures are directly computed with the filtered Navier-Stokes equations while the small-scale eddies are modeled based on the Smagorinsky eddy-viscosity concept. for example. turb 2 have to be solved. an eddy length . τ ijt 2   ∂u ∂u j u' i u' j = = kδ ij −ν t  i +  (1..

for example. This understanding may lead to new. k and where ν t ~ k 2 ε as mentioned earlier (cf.3.49) ρg d 2g  e where f = f  Re d .  from Moody Chart or formulas by  d Colebrook. Two-equation models require the solution of two transport equations (PDEs).50) π 2 gd 5 8τ w with f = 4c f = . and pump requirements are directly related to the friction loss hf. Wilcox. for example. turbulence models. 1 Notes to 1 and 2 above: DNS is useful for studying the detailed physics of turbulent "building block" flows. One-equation models require the solution of 2 one transport equation (PDE). Von Karman. etc. where 8Q 2 Lf hf = (1.3. ∆p L v2 ∆z + = hf = f (1. y ) require knowledge of the friction velocity uτ = (τ w ρ ) 2 (Schlichting & Gersten. which relates pressure drops to losses in terms of the friction factor f. Pipe sizing. Blasius. turbulent pipe flow problems are solved with the extended Bernoulli equation. 2000). where ν t ~ l m2 . k = 1 2 ui' where ν t ~ k . 1998). For example. Current DNS examples include wall- bounded flows such as channel and boundary-layer flows as well as free .L flows   ro  Most turbulent velocity profiles u (x .L flows 7  r  δ u ~  o  (1.3. flow rates. semiempirical turbulent velocity profiles could be employed  1 1  y  or  y  7 power laws for pipe or B . or at least improved. on a more differential basis. power P = FD ⋅ v with FD = τ w Asurf . ρv 2 Alternatively. Empirical Correlations: The use of empiricism for fully developed pipe flows and flat plate boundary-layer (B-L) flows allows bypassing the equations of motion and hence eliminates the need for basic turbulence modeling.56 Biofluid Dynamics scale. 5.51a-d)  y ln  or ln y log laws for pipe or B .

i. b) . the Navier-Stokes equations have to "cover" all length scales of the turbulence field. umax or u). DNS is presently too costly for advanced engineering flows.. etc. mixing layers and plane jets. Note to 3: RSM accounts for the history and nonlocal effects of the mean velocity gradients.. tref. all flow examples are presently restricted to relatively low Reynolds numbers and simple geometries.. 3 lk − UL ~ Re L 4 with Re L = L ν where L is a reference length (e. e. j + νu i . a good indicator of what is generably feasible are the types of turbulence models available in present- day CFD software. wall curvature.g. ∆x << l k .3. LES. Thus. • Kolmogorov time scale 1 ν  2 U3 tk =   where again ε~ ε  L 1 tk − L ∴ ~ Re L 2 with t ref = t ref U Note: Time step t << tk. In any case. is still too cost-intensive and not yet suitable for complex engineering problem solutions. although presently capable of depicting more complex flows than DNS. jj ρ (1. Employing again 1 t2 u i = u i + u' i where ui = ∆t t1 ui dt ∫ results after time-averaging in the mean flow equations: u i .Chapter I: Elements of Continuum Mechanics 57 shear-flows.. However. In DNS. In conclusion.i = 0 and Du i Dt 1 ( ) = − p . countergradient transport. B-L thickness or largest eddy size) and U is a reference velocity (e. Incorporation of length and time scales in turbulent flow calculations is as follows: • Kolmogorov length or dissipation scale 1 ν 3  4 U3 lk =   with ε ~  ε  L   where ε is the energy dissipation function. nonparallel flows. including relaxational effects.g.52a.g.i − u' i u' j .e.

In conclusion. of the RST 2 equation is approximated as (Rodi. j +u' j .. .  Dt  where simultaneously solutions to auxiliary PDEs for k and ε have to be supplied.k = ˆ viscous dissipation (to be modeled) (ii) Algebraic closures: Assuming the transport of u' i u' j to be proportional to the transport of turbulence energy.58 Biofluid Dynamics Obviously. k .54) resulting in algebraic stress equations for  Dk  u' i u' j = fct. etc. the apparent stresses − ρ u 'i u 'j have to be modeled in order to gain closure.3. The Reynolds Stress Transport (RST) equations are of the differential or algebraic form.53) ( ) Dij = ν u' i u' j . ε .k − u' j u' k u i . RSM is now being incorporated into fluid dynamics software packages.e. .i ) = ˆ pressure-strain correlation (to be modeled) ε ij = 2ν u' i . although very computer. the L.H. on the viscous laminar level) Pij = −u' i u' k u j . (i) Differential closures: D Dt ( ) u' i u' j = Dij + Pij + Tij + Π ij − ε ij (1.k u' j .and data-intensive.kk =ˆ diffusive transport (i.3. However.k turbulent diffusion transport (to be modeled) Π ij = 1 ρ ( p' u' i . such as rapid near-wall variations especially in the ∏ij terms. k = 1 u' i2 .k = ˆ stress production  1 (  Tij = − u' i u' j u' k + u' i p' δ ik + u' j p' δ ik  = ρ ˆ )   . wall effects.S. 1980) D Dt ( u' i u' j ≈ k ) u' i u' j Dk Dt (1.

have kept EVM as the first choice for a large variety of engineering problems.3. at least two independent time scales accounting for the different response mechanisms of the large-scale and the small-scale motions are needed in some applications.09 or  t νt =  ( ∂y ) ν inner = fct l 2 . δ1 . and other factors.e. ∂u . typically based on zero-equation models or two-equation models. flow separation. for wall-bounded shear layers. However. model inclusions of correction factors and extension terms. l = κy (Prandtl & Von Karman) ∂y or ∂u νt = l 2 (1.. γ i for 0 ≤ y ≤ y c ( ) (Cebeci & Smith) ν touter = fct U o . Note to 4: EVM. Furthermore. RSM relies typically on a single turbulence time scale to characterize rate processes in turbulence. γ o for y c ≤ y ≤ δ . Virtually all fluid dynamics software packages entertain MLH (i. based on Prandtl's mixing length hypothesis (MLH): ∂u νt = l 2 .435 and λ = 0..Chapter I: Elements of Continuum Mechanics 59 may still cause substantial errors. Thus. proper near-wall shear-layer modeling is crucial for successful turbulent flow simulations. fluid compressibility. representing effects due to the presence of walls. relatively low Reynolds numbers. flow rotation.e.55) ∂y where κy for 0≤ y ≤ λ δ κ l= λ λδ ≤ ≤1 y (Cebeci & Bradshaw) for κ δ κ = 0. zero-equation) and k-ε as well as RNG k-ε (i. work reasonably well for nonseparating. However. streamline curvature. Again. The underlying concept is Boussinesq's analogy with molecular transport of momentum. near-parallel shear flows. in simplified form ∂u τ t = ρνt where ρν t ≡ µ t ∂y Now. two-equation) closures.

3.57) Dt k ∂xi  σ ε ∂xi  with σ k ≈ 1. energy dissipation rate ε.. at least two turbulence quantities have to be specified.09.e. For example. (iii) the RNG k-ε model. 1986). A being the Van Driest damping length.92 . Pij =ˆ k-production terms.56) Dt ∂x j  σ ∂x   k j and Dε ε 2 ∂  ν t ∂ε  = Gij − Cε +   + ν∇ 2 ε (1. i.. (i) the k-L model: νt = c k L (Kolmogorov-Prandtl) (ii) the k-ε model: 3 k2 k2 ν t = Cµ . Yakhot & Orzag. inner For a complete eddy-viscosity turbulence model. or dissipation time scale τ. . σ ε ≈ 1. and Gij = ˆ -generation terms. ε∝ (Jones & Launder) ε L where Dk ∂  ν t ∂k  = Pij − ε +   + ν∇ 2 k (1. again ν t = Cµ k 2 / ε (1. η ∞ = 4. β = 0. typically the turbulence kinetic energy k plus a length scale L.58) and the transport equations for k and ε are the same but the coefficients differ (cf.0 . capturing near-wall effects such as flow separation better than the basic k-ε turbulence model.60 Biofluid Dynamics   − y  where l = κy 1 − exp  .3. For example.3. Cε is not a constant anymore but η (1 − η / η ∞ ) C ε1 = 1. C µ = 0. and δ is the B-L thickness.e. i.015. Uo is the outer flow. and S = 2 S ij S ij )1 / 2 with S ij being the mean rate-of-strain tensor. τ t = τ touter at y = y c . γi   A  and γo are intermittency factors.42 − 1 + βη 3 where ( η = S ⋅ k / ε.38.

and near-wake effects have all been added to the basic log-law profile (cf. It has to be noted that the power law profile has a nonzero gradient at the centerline. y + = uτ (1.59) where [ C µ ≈ 0. Terms representing laminar sublayer. Spalding.3.61) can be rewritten in terms of the friction velocity uτ = τ w ρ and the wall coordinate y = ro − r . is the power law of the form: 1 u  r n = 1 −  (1. f µ = exp − 3. An alternative turbulent velocity profile. buffer zone.4 / (1 + Ret / 50 )2 ] with Ret = ρk / (µω ) . with n = 7. e ) where e is the surface roughness. which in the version of Wilcox (1998) is a low- Reynolds-number model covering laminar. Equation (1. 1961.62) uτ  ν  which is known as the 1/7-law. and turbulent flow regions: ν t = Cµ f µ k / ω (1..3.3. i. and u av = Q πro2 (cf.61) u max  ro  which can be constructed where n is a weak function of the Reynolds number.24. among others).40 and B = 5.e. neither the log-law nor the δ power law fulfill all boundary conditions at y = 0 and y =  .3. typically applied to pipe flow. and the two PDEs for k and w as well as numerous coefficients/functions for sample applications are given in Wilcox (1998).3. Note to 5: The turbulent boundary-layer velocity profile u 1 y u+ ≡ = ln y + + B. ∂u ( ∂r ) r =0 ≠ 0 .3.1).09. Re = u av D ν with D = 2ro . as 1 u  yu  7 = 8.Chapter I: Elements of Continuum Mechanics 61 (iv) the k-ω model. transitional. Table 1.60) uτ κ ν can be derived based on Prandtl's near-wall shear stress hypotheses. The friction velocity uτ ≡ τ w ρ is either indirectly measured or obtained from appropriate skin friction or friction factor correlations f (τ w ) = fct (Re.74 τ  (1. In summary. ro . The empirical constants are κ = 0.

Schlichting & Gersten. (c) Calculate the ratio τ turb / τ lam at the midpoint rm=0. Q=4×10-2 m3/s) when ∆p/l=2.3.0 8. pressure drops. (cf. Fig.3.. (a) Find the thickness δ s of the viscous sublayer (cf.1 m.806 0.791 0.8 10.865 Note: After Schlichting & Gersten (2000).3. wall shear stresses.2×106 n 6.41 & 1.0 6. Eqs. Example 1.849 0.817 0.4). tailored turbulent velocity profiles are often employed to calculate flow rates.1×105 1.9: Turbulent Pipe Flow Rather than solving numerically the Reynolds-averaged Navier- Stokes equations with appropriate turbulence model [cf. 1. i.1 Exponent n and Velocity Ratio u av u max as a Function of Reynolds Number.4.1]. 1.0 uav umax 0. Re # 4×103 2. (b) Determine the centerline velocity umax and the ratio uaverage/ umax. . 1997. Sketch: Assumptions: • steady fully-developed turbulent τw flow • • constant fluid properties D r u (r ) rm Approach: x umax • Use empirical velocity profiles l Solution: (a) Thickness δs: The viscous sublayer ends when the inner variable y + ≈ 5 . (1.1×106 2×106 3.6 7.3.59 kPa/m. among others).0 10. Kleinstreuer. Problem: Consider water at 20oC flowing through a horizontal pipe (D=0.e.865 0.42) and Sect.3. 2000.025 m.3×104 1. etc.62 Biofluid Dynamics Table 1.

9-6a. • δ s = 0.02 mm .9-4) (n + 1)(2n + 1) Thus. l r ( ) Note that τ t (r = 0 ) = 0 and that τ t r = D 2 = τ wall . Hence.07×105.Chapter I: Elements of Continuum Mechanics 63 δ s uτ y+ = =5 (E1.186u av Note: u max la min ar = 2u av Here.9-2) so that τ w = 64. from a basic force balance ∆p = 4lτ w / D (E1. we obtain ∆p τ = 2 rx := ¢ which implies τ rx ≡ τ total =Cr.9-1) y =δ s ν where uτ ≡ τ w / ρ is the friction velocity.9-3) u max  R  where (cf.3.9-5) From a force balance on a control volume. uav 2n 2 = := 0.8 N/m 2 and u τ = 0.4 because ReD =5.04 m/s (c) Shear stress ratio τ turb / τ lam r = rm : Recall that τ total = τ lam + τ turb (E1.8432 umax (2n + 1)(n + 1) or • u max = 1. Table 1. knowing u av = Q / πR 2 • u max = 6. 1 (b) Velocities umax and uaverage: Employing the -law n 1/ n u  r = 1 −  (E1.1) n=n(ReD):=8. n2 Q = ∫ u d A := πR 2 u av = 2πR 2 u max (E1.b) D l D . Now. πr 2 l .255 m/s Hence. τ ∆p 4τ w τ total = 2 w r and = (E1.

1.3. e.4 N/m 2 D (1− n ) / n du µu max  r • τ lam = µ := 1 −  dr nR  R r = rm → τ lam (r = rm ) := 0. as highlighted below and demonstrated in Chapter V. a reduced equation of motion or a special form of the Reynolds Transport Theorem..0266 N/m 2 • τ turb = τ total − τ lam . boundary conditions. the decision for deterministic problems is between a differential or integral approach. τ lam r = rm 1. Step 2 probes an understanding of the physics of a given flow system and the problem requirements.3. The sequence of solution steps for industrial or research problems is quite similar as depicted in Fig. Sect. the type of models or mathematical descriptions are labeled either “distributed” or “lumped. and submodels for closure. Step 3 in conjunction with dimensional analysis and scaling (cf. τ (r ) is linear in fully developed pipe flow regardless of flow regime and type of fluid.” When selecting the differential approach.g. Alternatively. what’s given and required is usually well spelled out and hence the initial focus should be on a detailed system sketch.5.5 outlines four basic steps with comments for academic problem analyses. Now. clear problem statements and sufficient data are typically given (Step 1). In an academic setting. which govern the .3. so that τ turb • = 1220. Figure 1.5. 2τ w • τ total (r = rm ) = rm := 32. which leads to the most suitable modeling approach. concept/ approach. although much more difficult for all four steps involved.64 Biofluid Dynamics Indeed. and assumptions/consequences rather than starting right away a solution procedure. In an academic environment.5 Solution Techniques The type and completeness of information given for a biofluid dynamics problem in conjunction with the justifiable physical assumptions and their mathematical consequences greatly determine the system- describing equations. Experimental analysis not being an option and selecting the Eulerian framework.3.3) may help to identify key dimensionless groups. 1.

e.. conditions... integral approach. and submodels (e. and accuracy of the anticipated solution • List the mathematical consequences of the assumptions. a suitable coordinate system. etc. . i. equation coupling. nonlinearities.e. completeness..e. Note: Speculate about the physics of fluid flow: (a) determine which variables and parameters play an important role in the flow problem. operating and boundary conditions. and/or (b) recouping initial or boundary conditions • Vary system parameters.2) • Solve the final equation and verify its correctness by: (a) inserting numbers and verifying the result. conditions. Assumptions & Implementations • List assumptions based on physical insight Note: Any decision at this point determines the correctness. and type of coefficients • Select appropriate solution method (cf. 1. i.5 Outline of four basic steps with comments for academic problem analyses. and what’s given & required • Sketch anticipated profiles. in case of a differential solution approach Solution • Formulate a complete set of equations. Table 1. turbulence or non-Newtonian fluids) employing symbolic math. etc. • Determine concept(s). NO NUMBERS at this stage • Check fluid properties.g. or postulates.& space-dependencies. time.3. and (b) how the flow system may respond to changes in operating or boundary conditions.g.Chapter I: Elements of Continuum Mechanics 65 Problem Statement • Check system information and data sets provided • Study requirement(s) and/or expectations Note: Read the problem statement carefully ! System Sketch & Concepts • Incorporate in the schematics. differential vs. in order to gain more physical insight and understanding • Plot the results and comment! Fig.3.

FLUENT. CFD. BVP. CFD-TWOPHASE. or elliptic  I. can also be solved with MATLAB. • Type of Equations of Linear or nonlinear algebraic equations as wellas integral equations.. ordinary differential equation. parabolic PDEs can often be transformed to ODEs via separation of variables. and/or similarity theory. finite difference method. inhomogeneous. boundary condition.. parabolic.g.i.g. initial condition. http://www.g. parabolized.66 Biofluid Dynamics Table 1. as well as Femlab and FlexPDE. 2001). SOR. successive overrelaxation.. integral.netlib. variable coefficients. integral method. http://www.netlib. finite volume method. ) ODE's (BVP) (shooting methods) (e. SOR) Nonlinear algebraic equation (Newton-Raphson) ODE's (IVP) (Runge-Kutta) (e. time-dependent. nonlinear systems of ODEs. FEM.. an asymptotic analysis using perturbation may be sufficient. 1995a. Notes: IVP. CVM) Commercial PDE-solvers: CFX. Algebraic and ordinary differential equations. PDE.  parabolic/hyperbolic. ODE.g.3. for ODEs: Polyanin & Zaitsev. IC. Runge- Kutta) PDEs (FDM. BC. finite element method. in time at t=0 or in space at x=0 • Auxiliary ∂u Conditions Neumann : ∂n Homogeneous or inhomogeneous BC Dirichlet : u Cauchy : mixed • Solution Approaches Step 1: Classification of governing equations: type of equation [algebraic. and for linear PDEs: Polyanin. second-order. as well as linear PDEs. partial differential equation. boundary value problem.2 Types of Modeling Equations and Solution Methods. CVM. mixed (integro-differential)] Characteristics of. Step 2: Employ transformations and/or approximations. differential equations: nonlinear. FVM.. e.e. direct Stiff ODEs (Gear) Nonlinear ODEs (special transformations. among other generally available software packages. Step 3: Look up textbook solution or solution procedure for a given type and given characteristics of your equation (e. PDEs. control volume method.C. computational fluid dynamics. e. . Step 4: Use appropriate algorithms or computer software for solving Linear algebraic systems (Gauss-Seidel. say. multidimensional (for PDEs). ODE.. initial value problem. FVM.g. Interest:   IVPs or BVPs. FEM. PDE.

2..3. the separation-of- variables methods with special function representation.. Finlayson (1978). e.5.e. the resulting differential equation(s) can then be classified and a solution approach identified. Still.. ∂ 2u • Neglect second-order terms in streamwise direction: ≈ 0 . one could justify the following on a case-by-case basis (cf.3. Also. system geometry.2): r r r r • Delete the nonlinear terms: (v ⋅ ∇ )v ≡ a conv ~ Finertia ≈ 0 .) it might be justifiable to drastically simplify the Navier-Stokes equations. Depending on the characteristics of the flow system (i. and associated numerical software. Specifically. the modeling equations and related auxiliary conditions have to be understood and fully classified before any solution approach can be considered (see Table 1.2). flow patterns. nonlinear. and the integral methods are very useful to gain physical insight and simple problem solutions. . Accessible texts for review include Spiegel (1971). the methods of weighted residuals (MWR) form the foundation of several numerical methods. and Hoffman (2001). desktop computers.. ∂x 2 in boundary-layer flows. 1. etc. The section concludes with the methodology of how to set up the Reynolds Transport Theorem for fluid mass and linear momentum transfer. boundary conditions. Sect. With Table 1. for creeping flows. etc.3. which is a special case of the MWR.1 Solution Methods for Differential Equations A number of sophisticated but time-consuming solution methods employing complex transformations have lost their appeal with the advent of affordable and powerful calculators. which greatly simplify the solution method. and combined (or similarity) variables. inhomogeneous PDEs. 1.. engineering workstations.5. Specific solution methods for the reduced Navier-Stokes equations are outlined thereafter. are a set of transient. academic problems or simplified industrial systems can be described with reduced forms of the Navier-Stokes equations.g.3.3. e.2 Solution Procedures for the Navier-Stokes Equations The Navier-Stokes equations. 1.g.Chapter I: Elements of Continuum Mechanics 67 flow field. etc. fully developed unidirectional flows.2. elliptic (in steady state). second-order. the similarity theory. describing conservation of mass and momentum of fluid flow with constant fluid properties. Frequently. among others. In reviewing the similarity theory and integral method. Now. the balanced approach of “physical insight” and “mathematical skill” can be pursued. The next subsection summarizes how the N-S equations are properly reduced based on physical insight. First. Greenberg (1998). the type of mathematical modeling equations and typical solution steps for differential equations are summarized in Table 1.

64) r where v = ∇φ and φ is the velocity potential. Poiseuille. ∂t • Check the problem's dimensionality as 1-D or 2-D flows.2) Laplace's Equation: ∇ 2φ = 0 (1.63) Dt ρ or (see Fig.g.3. Parallel Flows: Couette.3.68 Biofluid Dynamics • Assume a constant pressure gradient: − ∇p ⇒ ∆p/l . but if u = u(y) only as in Couette and Poiseuille flows: d 2u 1 dp = + g y :=constant (1. A. i. 1. (1. e.3. developing flow effects. • Neglect gravity effects.3. obtain pressure field from Bernoulli's equation.z. Solution Notes: • Numerical solution of Eq. end effects. Stokes' equation has to be solved: ∂u 1 dp  ∂ 2 u ∂ 2u  =− + ν 2 + 2  + g y (1.3.64). and/or evaluate the number of directions in which the velocity can vary.. ∇ ⋅ τr ~ ∇p and ∇ ⋅ v = 0 For such parallel flows v = w = 0 and the general postulate is then u =u(y.3.65) ∂t ρ dx  ∂y ∂z  Typically u = u(y.t) when transients are important.66b) . Summary of Some Basic Exact Solutions r “Frictionless” Flow: ν ≡ 0 or ∇ 2 v ≈ 0  as discussed in Section   1.3..66a) dy 2 µ dx or u ′′ = ¢ (1. ∂ • Assume steady state: = 0 if there are no time variations. r Dv 1 r Euler's Equation: = − ∇p + g (1.t). • Solve Laplace's equation analytically or numerically.63) to obtain velocity and pressure fields.e. etc. (1. in fully developed pipe or slit flows.2. suddenly accelerated plate (Stokes I or Rayleigh) flows as well as pipe flow start-up and oscillating plate (Stokes II) flow imply that r r r (v ⋅ ∇ )vr = 0 .3. • For Eq.

yy ) (1. x +ν u . y = − 1 ρ ( p . A).y) and dp/dx = fct(x) as is demonstrated with Example 1. films..2)..y) or u = u(x. employ transformations. sheets.t) y u=u(y) u=u(r) x U(t) (Couette) (Poiseuille) (Stokes II) Solution Notes: • Equation (1. Summary of Some Approximate Solutions Very Low Re#-Flows: Very slow motion around bodies and in conduits of "nonparallel" geometry (Creeping Flows).3.2. however. • For laminar flow in slightly varying conduits..e. x + vv. xx +u .e. yy ρ ) (1. quasiparallel flow may be assumed.g. y = − p .Chapter I: Elements of Continuum Mechanics 69 Application Schematics: Uo y r z u=u(y.68b) Solution Notes: Seek further reduction of the y-momentum equation.3. fibers. separation of variables or similarity method and then solve the resulting ODEs analytically or numerically (see Example 1.5 in Sect.3. . i.3. coats) • Hele-Shaw flow Neglecting or simplifying the inertia term are options.t). x + vu .6 in Sect.3.1. When u = u(x. Examples include • Flow past a sphere (Stokes.7). i. Steady Flow with Convective Acceleration: Flows around 2-D bodies flow near rotating disk. Oseen) • Lubrication theory (Reynolds equation) • 1-D stretching flow (e. xx + v . e. 1..3..1.g. nonparallel slits. y +ν v . ∇ ⋅τr (1. tapered pipes. use clever postulates as in the Von Karman solution for the rotating disk (see Sect. Re → 0. solve the equations numerically.66) can be integrated directly (see App. B. flow in conduits allow the assumption that r (vr ⋅ ∇ )vr ~ ∇p . via the "no-slip" boundary condition.67) As a result.g. 1 ( uu .68a) and uv .. etc. e. the result is that u = u(x.

3. • For simple lubrication problems with mildly varying gap size.7.7). b) where L is the biharmonic differential operator and y is the stream function.b) ∂z ρ dz dz r( z ) Solution Notes: • Low Reynolds number flows around submerged bodies are discussed as part of the problem assignments in Sect. and gravitational effects (see Sect.3. 1.3. For general lubrication problems (Reynolds): r (vr ⋅ ∇ )vr = − 1 ∇p + ν∇ 2 vr and ∇ ⋅ v = 0 (1. Thus. surface ten- sion. (1.70 Biofluid Dynamics 0 (Stokes ) r r  r (v ⋅ ∇ )v ≈  ∂v U ∂x (Oseen ) which leads to (Stokes): r ∇p = µ∇ 2 v → ∇ 2 p = 0 or νL ψ = 0 4 ( ∂x ∂ y ∂u ∂u ∂u 1 dp ∂ 2u +u +v =− +ν 2 (1.70) may be negligible.71a. The viscosity µ might be small but >> 1 near the wall ∂y as shown in Fig. 1904): δ v 1 ~ ~ << 1 l u Rel where δ (x ) is the boundary-layer thickness and l is the horizontal ∂u plate length. within a thin 2-D shear layer along a flat plate Prandtl's boundary-layer equations hold: ∂u ∂ v + =0 (1.. where p ≈ τ rr − (1. 1. r r h(x) << L. viscous.70) ρ For steady 1-D stretching flows: ∂w 1 dp dτ zz σ ρw =− + + ρg z .3.69a. the inertia term (v ⋅ ∇ )v in Eq.g.73a) ∂t ∂x ∂y ρ dx ∂y . 1. High Re# Flows: Boundary-layer flows near solid walls obey the criterion (Prandtl.3. • Stretching flows are near-parallel flows with 1-D inertia.6. e.

Chapter I: Elements of Continuum Mechanics 71 dp =0 (1.g. the new independent coordinate combines.73c.73b) dy where 1 dp ∂U o ∂U o − = + Uo and p(y) = const. has to be found such that the explicit dependence on the former (two) independent variables disappears. CVM. solve resulting ODEs numerically (e..3. and that the typical problem can be represented by parabolic PDEs (cf. 1. etc. (a) The conditions under which similar solutions exist. specifically. a similarity variable. η.3. Runge-Kutta method with two "initial" conditions).3. Solution Notes: • Use the Falkner-Skan transformation for steady self-similar boundary-layer flows.6 Laminar flat-plate boundary-layer flow.) for general cases.3. sinks or sources. (1. • Use the integral method for steady nonsimilar boundary-layer flows using suitable velocity profiles.3 Similarity Theory Similarity solutions are of interest since they indicate that it might be possible to transform two-dimensional partial differential equations into one ordinary differential equation. Fig. curvature effects. 1.3.7). (b) The selection of appropriate transformations within a particular coordinate system.5. and other singularities or nonuniformities). • Employ numerical parabolic equation solvers (FDM. these include that the given system is free of any nonuniformities (shock waves. 1. η.d) ρ dx ∂t ∂x y Uo Vo Uo u∞ p outer ( x ) δ (x ) p∞ ρ∞ pinner ( x ) u(y) B-L Separation v u x Fig. x and y as η = y g ( x ) into one variable where g(x) could be d(x) the boundary-layer thickness. . for example. Thus. Similarity theory deals with several transformation aspects. flow separation. FEM.

1. 1979. etc. converts the system equation into an ODE for f(h). The first method starts with the integration of a given set of partial differential equations that describe a given flow system.3.). requirements of fulfilling the boundary conditions.y) and f() can be determined (cf. z ) and the similarity function f (η ) are postulated on physical grounds and then nondimensionalized. η and f are postulated with unknown exponents or the governing PDE is linearly transformed. Now. again. y ) or η (x . Note: Numerous solved example problems may be found in Kleinstreuer (1997) and other texts.7 Similar and nonsimilar velocity and temperature profiles. U U δ(x) y u(y) u(y) reverse flow x x 1 x 2 T T y δth(x) T(y) T(y) x Tw Tw qw(x) Fig. or the group theoretic method.t ) or η (r .4 Integral Methods Two solution techniques dealing with integral equations are briefly discussed. known as . with unknown exponents.5. Hansen. In more mathematically formal methods. inserted into the two- dimensional PDE. the functional forms of the similarity variable η = η (x . Buckingham's pi theorem.3. and/or invariance of the original and the transformed PDEs generate equations with which the exponents and hence (x. the free parameter technique.72 Biofluid Dynamics Associated with finding this transformation is the determination of an appropriate similarity function f() that. with the intuitive/ physical approach. 1964. Na. (c) The methodologies for implementing the tasks just outlined differ in what aspects they emphasize.. For example.g. nondimensionality. e. 1.

g.3. In contrast to separation of variables and similarity theory. a two-dimensional partial differential equation is integrated in one direction. etc. φ is a dependent variable. In general. fluxes.3. remaining in the governing differential equation are approximated on physical grounds or by empirical relationships. As can be expected. jets. the entrainment . The type of weighing function W determines the special case of the MWR.3. Eq.74) ∂t where L (•) is a (nonlinear) operator. collocation method. a transport equation in normal form can be written as [cf.e. The Von Karman method is best applicable to laminar/turbulent similar or nonsimilar boundary-layer type flows for which appropriate velocity. thin and thick wall shear layers as well as plumes.. Specifically. Galerkin finite element method. Implementation of the integral method rests on two general characteristics of boundary-layer type problems: (i) the boundary conditions for a particular system simplify the integration process significantly so that a simpler differential equation is obtained. for example. concentration. i.b) where R is the residual. Solutions of such problems yield global or integral system parameters.. a "profile" or functional φ that satisfies the boundary conditions. typically normal to the main flow. (1. i... The Von Karman integral method is the most famous member of the family of integral relations. i. Now. (1. and (ii) all extra unknown functions. control volume method. etc.Chapter I: Elements of Continuum Mechanics 73 the integral method.75a.17)). such as flow rates.e. and temperature profiles are known. forces. which assures the conservation laws for a control volume. The second approach starts with balance equations in integral form. or parameters. e. integral method. and thereby transformed into an ordinary differential equation. closure is gained using. 1978).3.75c) Ω we force the weighted residual over the computational domain to be zero and thereby determine the unknown coefficients or parameters in the ~ assumed φ -function.2)] ∂φ r L(φ ) ≡ + ∇ ⋅ (vφ ) − ν∇ 2φ − S = 0 (1. In requiring that ∫ WRdΩ = 0 (1. but contains a number of unknown coefficients or parameters. Eq. L φ ≡ R (1. boundary-layer thicknesses. (cf. Thus. Von Karman integral method. the Reynolds Transport Theorem (cf.e. and S represents sink/source terms. the unknown φ-function is replaced by ~ an approximate expression.e. the integral method is an approximation method. Finlayson.3. drag coefficients. which is then solved analytically or numerically. and wakes. i. ~ () ~ () L φ ≠ 0 . shape factors. which in turn is a special case of the method of weighted residuals (MWR)..

y) and integrating the term ∫ v ∂y dy 0 by parts. and wakes or by expressing velocity and temperature profiles with power expansions for high Reynolds number flows past submerged bodies.74 Biofluid Dynamics concept for plumes. i.e..10: Integral Method applied to the Blasius Problem ( U 0 = u ∞ = ¢. i. yields [u (U − u )] dy = τ w δ ∂ • ∫ 0 ∂x ρ . jets. Example 1. ∫ udv = uv − ∫ vdu . ∂p / ∂x = 0 ) Sketch: Assumptions: • Steady laminar 2-D flow with y u∞ constant fluid u∞ properties δ(x) p∞ ρ∞ v u(y) u x Recall For the Blasius problem. the boundary-layer equations reduce to ∂u ∂v ∂u ∂u ∂u 1 dτ yx + =0 and +u +v = ∂x ∂y ∂t ∂x ∂y ρ dx • Solving for v in the continuity equation yields ∂u ∂u y v = − ∫ dy + f ( x ) := − ∫ dy ∂y 0 ∂x • Integration across the x-momentum equation yields δ (t ) 1 dτ yx δ  ∂u ∂u  τ ∫y =0  u ∂x + v ∂y dy = ρ ∫0 dx dy := − ρw δ ∂u • Inserting v(x.e..

With the previously derived momentum integral relation. 0. a suitable u(x.Chapter I: Elements of Continuum Mechanics 75 where U = u ∞ .133 ρ U dx 0 dx We also know that ∂u τ w (x ) = µ 2 := µU ∂y y =0 δ Combining both results leads to an ODE for δ ( x ) : µdx = δdδ subject to δ (x = 0 ) = 0 .356 ρU 2 ρUx Graph: τw τ w (x ) x .0665 ρU Integration yields µx • δ (x ) = 5. which matches the boundary conditions at y=0 and y = δ (x ) .y) profile has to be postulated. Now. d δ 2 dδ τ w (x ) = ρ ∫ u (U − u ) dy := 0.48 ρU and hence µ • τ w (x ) = 0. Solution: For laminar flow over a flat plate. 2 u y  y =2 −  u∞ δ δ  is a suitable profile where δ (x ) is the key unknown.

and (iii) similitude. Example 1. DA covers: (i) nondimensionalization of field equations resulting in dimensionless groups governing a given flow system. time. and design parameters.” i.11: Mean Droplet Size in an Agitated Oil-Water Dispersion (Dimensional Analysis) Consider a cylindrical water tank (D. bubbles or droplets.e. air or oil. in conjunction with appropriate scaling and order-of-magnitude estimation. ω I ). angular velocity ω I ..5. scale-up. Dimensional analysis provides significant insight. break up and disperse throughout the tank (interfacial surface tension σ 12 ). scaling.5 Dimensional Analysis and Scaling Dimensional analysis (DA). relates to all the important system parameters listed. heat. is steadily increased.. operating parameters.g. experimentation. is a powerful tool in the engineering sciences (Astarita. provide “very often 90% of what one can ever hope to know about a problem.e. A finite volume fraction α of an immiscible fluid.e. or the “by-inspection method.76 Biofluid Dynamics 1.. i. fluid phase 2 ( µ 2 . among others). i. i. Middleman. etc. −1 • the surface tension between the fluids decreases.3.. fluid properties. without actually solving the problem. H) with a rotating baffled impeller ( d I . is initially injected and the impeller speed. and CFD (computational fluid dynamics) are the key tools for theoretical engineering analysis and design. σ 12 . based on logical reasoning we postulate in steps that the mean particle size decreases as • the impeller speed increases nonlinearly [v I = (d I ⋅ ω I )n ] . Thus.” He then gives a few examples in fluid mechanics. 1997. at least how the mean particle diameter.g. dp. It would be very difficult to calculate the particle size distributions and particle concentrations as a function of space. DA. if used jointly. 1999. Obviously. which is important in physical modeling. Clearly. ρ 2 ) mixes and forms air bubbles or oil droplets in the carrier fluid phase 1 ( µ1 ...e. e. larger particles. (ii) the Buckingham Pi Theorem. ρ1 ). and estimation. Astarita (1997) even asserts that dimensional analysis. 2003). and . techniques that generate system-specific dimensionless groups similar to (i).. and impeller speed. thereby relying on results from (ii). Undergraduate fluid mechanics texts devote entire chapters to dimensional analysis and similitude (e. White. and mass transfer to underscore the point.

For example. dp v2d ρ • ~ We = I I dI σ which is the Weber number. The result is dp  ρ µ  ω 2d 3ρ ~  1 2  I I 1 d I  ρ 2 µ1  σ 12 If the fluid properties do not vary. µ1 µ 2 . experimental results for light oil droplet dispersion as depicted in the graph were given by Chen & Middleman (1967). Now. If large enough. the size of the tank (D.Chapter I: Elements of Continuum Mechanics 77 • the fluid property ratios increase. by inspection. we note that or M / T 2 and hence a density [ρ ] = M / L3 and one [σ ] = F / L more length scale [d I ] = L are required to nondimensionalize the RHS. curve-fitting of the data generates a nonlinear equation for d p ⁄ d I . ρ1 ρ 2 . H) is immaterial. we have. with n=2  ρ µ  ω 2d 2 d p ~  1 2  I I  ρ 2 µ1  σ 12 Trying to obtain dimensionless expressions. Thus. Graph: 102 dp/dI 101 100 101 102 103 104 We # .

In general.12: Derivation of the Reynolds Number (Scale Analysis) Scale analysis. 1997). (1. and contaminated body fluids. Of this diverse research field. we scale the force ratio r r (v ⋅ ∇r)v := u l −1 u ul = ν∇ v2 νl u ν −2 i. two-phase flow is best described as the interactive motion of two different kinds of matter or media. solid) and/or its multiple chemical components.3.. Such circulating particle suspensions fall into the research category of multiphase flow (Kleinstreuer..3). 1998. 1998. gas.g. Sect. 1. blood as well as inhaled air with particulate matter. 1. and to form suitable profiles of dependent variables (cf. we know r r r r r FI ~ (v ⋅ ∇ )v and Fv ~ ν∇ 2 v Choosing the scale parameters u (average or reference velocity scale) and l (system length scale). Fan & Zhu. e.. ul • Rel = ν 1. we focus on two-phase liquid- solid and gas-solid flows with quasi-spherical particles of diameters dp>1 µm (fine) and dp<1 µm (ultrafine).e. Sect. Eq. starting with the definition inertial forces FI Re ≡ = viscous forces Fv From the Navier-Stokes equation in vector form [cf.4. liquid. which evolved from dimensional analysis (DA) and relative-order-of-magnitude analysis (ROMA).4 TWO-PHASE FLOWS Most biofluids are actually fluid-particle mixtures. Kleinstreuer. is a useful modeling tool to develop dimensionless groups characterizing the process dynamics at hand (cf. 2003.78 Biofluid Dynamics Example 1.2).3.1.g. as illustrated in Table 1.19)]. to determine similarity variables for PDE-to-ODE transformations (cf. or Soo. Crowe et al.6. The difference between the matter (or media) can be its thermodynamic state. called the phase (e. . 1990).4. The example given here is the derivation of the Reynolds number.

droplets. turbulent particle trajectories).. multiphase flow becomes “flow of multicomponent fluids. because of the mathematical (and computational) complexities involved.g. with their particular definitions and modeling approaches. many systems have been simulated with single-phase flow models or as (transient) one-dimensional two-phase flows. combustion engineering.e. i. simulation. e. multicomponent flows) After Crowe et al. 1999) label “phases” as “components. etc. (three- multiple-component flows) phase. Phase Component Single Multiple Flow of water.4.e. deterministic transfer equations (cf. 1.. and design still relies on system-specific data correlations. or bubbles in a continuum. (single-phase. historically the development of two-phase flow analysis has not followed any rigorous approach. two-phase flow modeling. * Note: Some authors (e.1 Examples of Flow Media*: Phase of Matter vs.4.e.g. It has evolved from separate contributions in several application areas.. a gas or liquid stream.1). oxygen. Table 1. Examples include the study of single or monodisperse solids..e. all somehow embedded in continuum mechanics laws and/or statistical mechanics concepts.Chapter I: Elements of Continuum Mechanics 79 Table 1.” i. oil.. i.g. single.4. Free surface . single-component flow) component flows) Flow of air. Boltzmann equation with suitable collision integral). Drew & Passman. such as nuclear power. semi-empirical equations. phase.1 Modeling Approaches There are two basic types of models. bubbly slurries.1 and Fig. Multiple mixture. (1998). liquid polymer Flow of air-water-particles. (single-phase. enhanced by probability density functions (cf. and aerosol science.. chemical processing. assumed probability density functions. or with random perturbation functions added (e. For both approaches. As a result. Furthermore.” 1. Component of Matter.4. etc. molecular dynamics). corpuscular/molecular models on the subcontinuum scale (kinetic theory) and continuum models on the micro/macroscale (continuum mechanics) (cf. However. the conservation laws are applied. or the use of random number generators.. Flow of water & steam (two- Single etc. i.

.and submicron-size liquid carrier fluids) particle flows with effective • Free surface or two- mixture properties layer/two-fluid flows • Dispersed phase with particle • Particle or particle cloud drift velocity relative to carrier trajectories fluid flow • Non-uniform.80 Biofluid Dynamics (a) Hierarchy of multiphase flow models Multiphase Flow Models Two-phase Flow Models • Molecular Dynamics Models • Continuum Models (a) Separated Flow Models (interacting media) Two-Fluid Models Particle Trajectory Models (b) Flow Mixture Models (pseudo two-phase) Drift Flux Models Homogeneous Equilibrium Flow Models (b) Two-phase flow model applications Two-phase Flow Models Flow Mixture Models Separated Flow (well-mixed or well-dispersed Models media. pseudo two-phase flows) (e.e. . non- • Non-Newtonian fluid flows equilibrium particle • Quasi-homogeneous suspension flows equilibrium flows with averaged • Two interacting-fluid flows properties Fig..1 Two-phase flow models: (a) hierarchy and (b) sample applications. i. side-by-side interacting • Dense. 1. uniformly dispersed fluids. or "particles" in gas or micron.g.4.

1b. i.4. Densities. • The number of particles per unit volume δN n = lim (1.4. where δN is the number of particles in the mixture volume δV and δV ′ is the limiting mixture volume in line with the continuum assumption. The volume or void fraction of the continuous (i. the parameter definitions are basically the same (cf.4. and concentrations.1.4. the dispersed phase is commonly quantified in terms of particle densities.. fractions. Some of these parameters appear in the two-phase flow modeling equations while critical values of others are used to determine an appropriate modeling approach in the first place. especially dispersed particle flows falling into one of the two modeling categories.1.1 Definitions A number of dispersed flow terms are briefly introduced in this section and further elaborated on in subsequent chapters. Clearly. 1.4.e. Specifically.2c) • The bulk (or effective or apparent) density of the mixture is . for the dispersed phase contains a sufficient number of particles so that a stationary average is warranted..e. i.Chapter I: Elements of Continuum Mechanics 81 flow and separated multifluid flow problems have been treated as single- phase with coupled boundary conditions at the phase interface(s). 1. in order to analyze multiphase (or just two-phase) flow systems. the limiting volume. or bubbles in the mixture δV d α d = lim (1. In order to characterize and quantify two-phase flows. They are further highlighted in Fig.2a) δV →δV δV ′ is the volume fraction of the dispersed phase.e.1. i. • The space occupied by the solid particles. loadings. and/or concentrations. approximations to the actual flow phenomena.1) δV →δV ′ δV is called the number density. 1. Their complexity increases sharply with the hierarchical level as depicted in Fig. liquid) phase is δV c α c = lim (1. δV ′ .4. fractions.. droplets. have to be employed. Appendix C). mathematical models.2. either mixture flows or separated flows.4.e. While the nomenclature may differ from source to source.1a.. The definitions of characteristic system parameters are based on the continuum assumption discussed in Sect.2b) δV →δV ′ δV where αd +αc = 1 (1.

4.7d) so that the slip ratio is vc v s= = 1+ r (1.s = v m (1. • The mass concentration of the dispersed phase is c = ρd / ρc (1. often approximated as κ ≈ c = ρ d / ρ c .4. is v drift = v d − v m (1. while the phase or actual mean velocities are j j v c = c and v d = d (1.7f) • The time it takes for a particle to respond to a (local) change in fluid velocity or fluid temperature can be estimated from the .e.5b) and the volume flux is Q + Qd j m = j c + j d = G m / ρ m := c = v c . i.4.4. i = c.4. the relative (or slip) velocity can be defined as vr = vc − vd (1..5a) Now.4b) m& c is the total loading.3a. d (1. ρ i is the actual density of material i. while the ratio of overall mass flow rates m& κ= d (1.4. the mass flux of the mixture is G m = m& m / A = Gc + Gd (1.4.6 a-c) A where vm is the superficial velocity of the mixture.7a.7e) vd vd whereas the drift velocity.82 Biofluid Dynamics ρ m = α c ρ c + α d ρ d := ρ d + ρ c and ρ d = nm p (1.s + v d . s = α i v i .4a) • The mass flux ratio (ρ d v ) / (ρ c u ) is the local loading.4. • The mixture mass flow rate can be rewritten and applied to two- phase pipe flow as m& m = m& c + m& d := (ρQ )c + (ρQ )d (1.c.b) where m p is the mass of a single representative particle.7c) Finally. and ρ i is the bulk density of phase i=d.4.4. dispersed phase velocity relative to the mixture velocity.4.b) αc αd This implies that the phase “i” superficial velocities are v i .

. (u − v )d p Re p = < 1. Here. no kinetic pressure due to particle velocity fluctuations or collisions. which greatly complicate two-phase flow modeling and analysis.4.. respectively. For example. Thus. i. we have ν Re p 18µ dT 12k c dv = (u − v ) and d = (Tc − Td ) (1. Assuming low relative particle Reynolds numbers.b) dt ρ d d p 2 dt ρc d d p2 which allows the velocity (or momentum) response time to be defined as ρ d d p2 τv = (1. 1. i. While in continuum fluid mechanics the key thermodynamic property is the pressure that powers the fluid flow.e.. πd p2 ρ c (u − v ) u − v dv 1 m = CD (1.e. (1.4. and hence information transfers along particle trajectories. the carrier fluid determines particle motion.8b) dt where u is the fluid velocity and v is the particle velocity.e. in a dilute particle phase there is no particle pressure. one-way coupled or “uncoupled” phases) when there are no particle-particle collisions. Obviously. i.10b) 12k c • Another characteristic parameter of interest in dense particle suspension flows is the time between collisions. which is a function of the particle loading [Eq. or mass transfer.2 Phase Coupling There always exists some degree of interactions between the dispersed and continuous phases due to momentum exchange. two-way coupling is occasionally negligible. −1 τ col = f col (1.4.10c) where f col is the collision frequency.4b)]. C D ≈ 24 and Nu ≈ 2.Chapter I: Elements of Continuum Mechanics 83 solutions of simplified forms of the equations of particle motion and particle heat transfer.8a) dt 2 4 and dTd mc d = πk c d p Nu (Tc − Td ) (1.4.e.0 .10a) 18µ and the thermal response time ρ d c d d 2p τ th = (1. criteria that determine the degree of phase coupling are very useful. heat.9a. Fortunately.4. particle suspensions are considered to be dilute (i. which ..

/ m& syst . the mixture is dilute as long as κ ≤ O(1) .e. c.84 Biofluid Dynamics also implies that the particles stay so far apart that the flow field remains unaffected and hence the fluid flow forces on the individual particles stay the same. . temperatures.. turbulent dense particle suspensions exhibit strong two- way coupling effects. the flow is dilute because collisions occur infrequently. buoyancy) and hence influences the local flow field. In contrast. . Still. pressure.004 for solids and drops and a d ≈ 0. and energy coupling between the two phases could affect the carrier fluid’s velocity. α p = κ ρ g ρ p << 1 . greatly determine the probability of interparticle collision in conjunction with the local flow pattern and changes in system geometry. the number density n = m p ρ p ≈ N Vsystem . or ad may vary regionally in a flow field due to transient or spatial changes in particle distributions. These examples reflect momentum coupling. and ( ) l particle / Lsystem ≥ O 10 −2 . which may alter particle shape and motion (e. Clearly.. or loading κ = m& disp . considering gas-particle (solid or liquid) flow. momentum.001 for bubbles. i.. and droplet vaporization in a hot gas stream is an example of two-way mass coupling. Experimentally it was found that ad1/3 <<1 works as a criterion for dilute particle suspensions. 2000). In the average: (i) When τvel τcol < 1 and ad <<1. sizes. where ad ≈ 0. d p . two-way mass. n: • nd 3p << 10 −3 for dilute case where the average particle distance is l = n −1 / 3 . where ρ p / ρ g ≥ 10 3 . and possibly number densities. Another criterion for dilute suspensions when ρ p / ρ f ≈ 10 −3 can be given in terms of particle density number. for any given system.g. In summary. a measurable pressure gradient around the exterior of an individual (larger) particle induces a stress field inside a particle (Patankar et al. For example. It should be noted that the loading parameter values of κ . known as “turbulence modulation” due to the presence of particles. while hot solid particles in a cool gas stream changing the local gas properties is an example of two-way energy coupling via heat exchange. and the particle size. as well as properties and the particles’ velocities. even in very dilute particle suspensions. temperature. l >> d p and • nd 3p ~ 10 −3 for intermediate case.

e. a reliable estimate of possible phase coupling is very important. especially near walls.. combining the total particle loading κ with the Stokes number yields a momentum coupling parameter (Crowe et al. the particles have no time to respond to the fluid dynamic forces before the next collision.e. nicely describe such phenomena. even in steady flows. the volume fraction of the dispersed phase is V sphere πd 3p αd = = Vcube 6l 3 . i. For example. or because τ vel > 1 .4. i. Example 1. the flow is dense because τ col << 1 indicates that particles are close together. whereas St >> 1 means that particle inertia is dominant ignoring any fluid fluctuations. particle pathlines differ from gas streamlines and hence particle clouds. ready to collide. St << 1 implies that particles follow readily the fluid flow including eddy fluctuations. while dilute flows can be often treated as uncoupled flows. each in a cube of a carrier fluid of extent l.e..way coupling KM ≡  (1. Considering spherical particles a distance l apart.13: Estimate of Phase Coupling Because dynamically uncoupled flows are much easier to analyze. with which the ratio l/dp can be evaluated. and particle-free regions may appear (cf.. (ii) When τ vel τ col > 1 . locally denser concentrations. dense particle suspension flows contribute two- way momentum coupling between the two phases. Particle trajectory models with appropriate point forces. large particles contribute to collision events because τ vel ~ d 2p and τ col ~ d −p 2 .e.way coupling Aspects of phase coupling and proper model selection with suitable solution method are summarized in Table 1. For example.. i.11) 1 + St  > 1 L two .4. κ  << 1 L one .Chapter I: Elements of Continuum Mechanics 85 Nevertheless. Obviously. Associated criteria of particle behavior are the Stokes number St = τ p / τ f and the density ratio. 1998) as mentioned in Example 1. Zhang & Kleinstreuer. trajectories are determined by the mean convective flow plus gravity.9. 2002).2. i. in both situations. One key parameter is the “minimum” particle spacing l..

where d p >> ∆h . and mixture properties µ .4.. • Direct simulation Monte Carlo (DSMC) method for relatively high Knudsen number flows All footnotes are given on the next page. properties. j drift or φ drift . ρ .. heat flux. dimensionality. (i) Smooth interface temperature. Parameter Ranges r Solution Methods# Two-phase Flow St = τ p / τ f † Re p ~ ∆v ‡ Vp § ρp § Models α = κ ≅ V ρf Flow Mixture Models General: Single-phase flow solution with modified fluid Quasihomogeneous flows ≈0 ≈0 Variable ≈1 µ = µ (α). Separated Flow Models • Stratified two-fluid flows in smooth constant layers Specific solutions of two moving fluids interacting at a Layered two-fluid flows Main criterion : τinterface ⋅ A ≤ σ critical ⋅ L . for example. Two-interacting fluid flows Two-way coupling: (i) Gas-solid • Point-volume formulation with additional collision (ii) Gas-liquid and (random) interaction forces (i. or Reδ* < Recrit straight or wavy interface.2 Parameter-Range Criteria for Two-Phase Flow Model Selection and Possible Solution Method. ρ =ˆ ρ mixture . 0<St< O (10 ) 0<Rep< O (10 ) (i) Dilute/uncoupled Limited number only restrictions Note: “resolved-volume” representation for a very few. and stress at interface (see above). flow regime. (ii) Wavy interface One-way coupling: Particle suspension flows Finite Eulerian (fluid) Lagrangian (limited number of (Trajectory model) particle particles/points). Eulerian- (v) Liquid-liquid Eulerian approach). .Table 1. (iv) Liquid-gas restrictions • Interpenetrating phases (i) –(v) (i.5 ≈1 µ → η = η( γ& ) where γ& ~ ∇v . etc. r Non-Newtonian fluid flows ≈0 ≈0 0<α≤0. Eulerian- Rep < O (10 ) (iii) Liquid-solid St < 1 α<1 Limited Lagrangian approach). O (1) Flows with component drift << 1 << 1 Variable Specifics: Mixture solution method depends on system’s flux time dependence. (ii) Dense/coupled 0<St<1 0< Rep <1 nonspherical particles only.e. matching of velocity.e.

Rep << 1 may imply dilute particle suspensions when particle-particle collisions are negligible and presence of small particles do not affect fluid flow.. the typical St number range is 0. Fluent or CFX for solving Flow Mixture Models.2: † St ≡ ρ p d 2pU /( 18µD ) .3 and κ ≤ 50 ~ 500 for highly concentrated dispersed flows • 0. or CFD-TWOPHASE for two-phase flows) • CFD books with Fortran codes and review articles (e.4. Tannehill et al. ..01<St<0. Cebeci.g. In summary. 1998. 2004) * Critical Reynolds numbers. r r ‡ Re p ≡ v − v p d p / ν f . Anderson et al.3 ≤ α ≤ 0. In lung airways with d p > 1µ m . Re = 4δv / ν . commercial software (e. for example. and Re > 1500 (turbulent flow). 2000. Experimentally it was determined that α 1 / 3 << 1 for dilute particle suspensions.. • St < 1 implies that particles may deviate somewhat from fluid streamlines. for moving film: For Re < 20 (negligible rippling). • α ≤ 4 × 10 −4 and κ ≤ 1 / 3 for dilute gas particle flows • 4 ×10 − 4 ≤ α ≤ 3 ×10 − 2 and κ ≤ 5 ~ 50 for dense suspensions • 0. 1998. where • St → 0 implies that particles readily follow fluid motion including eddy fluctuations.001 for bubbles.g. 20 < Re < 1500 (laminar flow with pronounced rippling). similar to κ ≡ m & p / m& f or c = ρ p / ρ f may vary significantly in the flow field. decoupled fluid-particle flows. Hoffman. 1997. “crossing effect” (Notes: Brownian motion has to be considered when d p < 1µm ). for Rep >> 1 see St >> 1. 1999. Roache. Ferziger & Peric. the particle volume fraction.2. etc. while • St >> 1 implies that particle trajectories are due to drag force and gravity. where for Rep → 0 see St → 0. § α.65 and κ ≤ 1 ~ 50 for moving packed bed # Resources include • User-friendly. α = 0.. 2001.03 ≤ α ≤ 0. Loth. i.004 for solids and drops and α = 0..Chapter I: Elements of Continuum Mechanics 87 Footnotes to Table 1.e..

However.e. or those for which one phase migrates relative to. i = d . and (b) actually nonhomogeneous fluids..4. ultrafine particle suspensions.88 Biofluid Dynamics or the length scale ratio is 1/ 3 l  π   ≥ 10 for dilute particle suspensions • = :=  d p  6α d  ≤ 2 for dense particle suspensions An alternative expression for α d in terms of dispersed phase mass fraction relies on the definition of the bulk densities.4. if. i. i.. fluids such as air..e.2. for example. 1. i.2 Mixture Models The mixtures of interest are those which are either quasi-homogeneous.e. that are treated as “homogeneous” mixtures..1 Homogeneous and Non-Newtonian Fluid Flow Models There are two types of materials or media which may fall into the category of homogeneous equilibrium models (HEM): (a) truly homogeneous. although possibly multicomponent. i.  δVi  ρ i = lim  ρ i  = ρ iα i .7 which indicates that the particles are so close that the flow is considered to be dense and particle interaction as well as phase coupling effects may be important. particles can be treated individually and the suspension flow is dilute. multi-component fluids in thermodynamic equilibrium. 1. a bulk motion. a drift flux exists. . c δV →δV '  δV  Now we estimate α d from κ αd = ρd ρd = 1+κ where ρ κ= d ρc and hence 1/ 3 • π 1+ κ  l =  dp 6 κ  In any case. say. α d = 10% then l / d p ≈ 1.e. when l d p ≥ 10 .

1969. Patankar & Hu.4.e. 1932.4. paints. α ≤ 0. 1.g. Marking the two phases with indices k = 1 (carrier fluid) and k = 2 (dispersed phase). in smaller bifurcating conduits as a non- Newtonian fluid mixture.4. mists.. α ≡ Vd V . 1970. In order to use this model. stimuli response is identical everywhere. Several theoretical expressions for the mixture viscosity have been derived. and certain food stuff). among others). the HEM treats the mixture as a pseudofluid.e.. i.4. which obeys all the laws of single-phase flow. carbonated beverages) and other well-dispersed particle flows (e. However.4. 1999). quasi-homogeneous behavior can be assumed if the different substances in solution are near uniform and well mixed. i. the kinematic mass concentration. and T1 = T2 . Thus. is employed. 1.1.05 (Taylor. as with most fluids and especially mixtures.1). slurries. sprays.1.e. (1. χ ≡ m & d m& .. alternatively.. Thus. In ″homogeneous″ mixtures. the time scale for the transport between phases or components is very much shorter than the overall characteristic or system time scale. Batchelor. the mixture or effective density is (see Fig. suspensions of fluid spheres at low concentrations.2) ρ m = αρ 2 + ( 1 − α )ρ1 (1. Soo. Appendix C). 1990).13) ρm ρ2 ρ1 where for homogeneous equilibrium flow the static void fraction α = V2 / V is equal to the kinetic volume fraction β = Q2 / Q and mass fraction χ = c . can be treated as a homogeneous Newtonian fluid. suitable thermodynamic and transport properties are needed for the mixture as well as a mixture equation of state (cf.e.. i. and in capillaries as separated multicomponent fluid flow (see Fig.12) or 1 χ 1− χ = + (1. m& 2 / m& = m 2 / m (cf. The mean density can be expressed as a function of the static concentration. the material behavior depends on the geometric flow scale (cf. commonly called the particle volume or void fraction [see Eq.2) in Sect. For example.Chapter I: Elements of Continuum Mechanics 89 such as bubbly flows (e. Mixture properties and solution techniques. p1 = p 2 . i. 2001. may have a mixture viscosity .. Wallis. v1 = v 2 .4. This assumption leads to the requirement that the two phases are in thermodynamic equilibrium. blood flow in large conduits with shear rates over 200 s −1 . commonly called the quality. blood.g.1]. For example. Gad-el-Hak.

4. As stated. 1. α < 0.4.14) Equation (1.15. water droplets in air. 1.5α   µ 2 + µ 1  (1. 1.19) j j ..90 Biofluid Dynamics  µ 2 + 2 5 µ1  µ m = µ1 1 + 2.4. 1964) (1.2 Conversion of two-phase flow to a single mixture flow. The following mixture viscosity expressions have been proposed for evenly distributed gas-liquid flows. 1984) < µ m = µ1 (1 + βα ) .18) or jf jg µm = µf + µg (Dukler et al.4.4. 1960) (1. 1.4. 1957) (1. for example.4..µ 2 Fig.05. µm ρ 2 . (see Fig. To account for mixtures with larger void fractions.4.14) implies two special cases: • Suspensions of solid particles where µ 2 >> µ1 (Einstein.16) are valid only for relatively small void fractions.17) µm µg µf or µ m = χµ g + (1 − χ )µ f (Cicchitti et al.4.e.4..14. sprays. the above three equations (1.. β = 5 / 2 or β 1 (1.4. µ 1 ρm .2): 1 χ (1 − χ ) = + (McAdams et al.4. i.. Isbin et al.15a-c) > and • Suspensions of low-viscosity gas bubbles µ m = µ 1 (1 + α ) (1.16) ρ1 . 1906. Brady. 1942. several authors have proposed averaged expressions that converge to the individual phase viscosities for the limiting cases of α → 0 and α → 1. etc.

e.e.079 Re-1/4.Chapter I: Elements of Continuum Mechanics 91 Additional correlations for effective viscosities of suspensions and emulsions may be found in Zapryanov & Tabakova (1998).42 kg/s) and air ( m& g = 0. i. Sketch: Assume: Concepts: ul ug • steady turbulent • homogeneous m& p2 fully developed equilibrium model two-phase flow (HEM) • area-averaged • Dptotal= Dpfriction r (i. the volumetric flow rates.. • Newtonian fluid p1 slip ratio D u g ul = 1 Solution: Based on mass conservation. void fraction. find the total pressure drop. Example 1.14-2)  ρg ρ   l  and the effective viscosity −1  x 1− x  µ = +  (E1. L=45 cm). m& = m& g + m& l (E1.14-3)  µg µ   l  .14-1) so that with u g = u l = u = m& / (ρA) and with quality x ≡ m& g / m& . Given a friction factor of f = 0. the mixture density −1  x 1− x  ρ = +  (E1. and mean water and air velocities.14: Quasihomogeneous Mixture Flow Consider a mixture of water ( m& l = 0.. uniform) +Dpstatic g velocities from extended • thermodynamic Bernoulli equation z equilibrium.01 kg/s) flowing upwards in a vertical pipe (D=25 mm.

g..0053 Re D Note: For smooth-walled pipes. Stokes’ hypothesis of a linear relationship ( ) rr rr r r between shear stress and shear rate.25 := 0.079 f = 0.64 kg/m 3 . exhibit unusual behavior due to their component make- up or molecular structure. from the extended Bernoulli equation p1 v12 p v2 + + z1 = 2 + 2 + z 2 + h f (E1.0042 m /s (water ) 3 so that the void fraction Qg α= = 0. the effective viscosity is shear-rate dependent. the Reynolds number is Re = ρuD / µ := 49.952 Q and as stated u g = u l = u = 17. and µ = 4. with u = 17. There are two “sources” of nonlinear viscous flow behavior: .023 and at T = 20 o C . Blasius (1910) suggested f = 0. x = 0.65 m/s Basic non-Newtonian fluid models.14-4) ρg 2 g ρg 2 g where 2 Lv hf = f   (E1.388 so that 0. Now.00833 m 3 /s (air ) Qi =   =   ρ  i 0.92 Biofluid Dynamics Thus. is invalid for such non-Newtonian type fluids because they may have memory.435 × 10 −4 kg/m ⋅ s . τ = µγ& = µ ∇v + ∇v tr . although treatable as single-phase flow.14-5)  D  2g ∆p total = 3169 N / m 2 The volumetric flow rates are  m&  0.316 Re D−1 / 4 for 4000< ReD <105.65 m/s . ρ = 49. e. etc. Some mixtures.

Bird et al. 1. Books on single-phase non-Newtonian fluids include Macosko (1994).3. Tanner.3): µγ&ij for Newtonian fluids τ ij =  η (γ& or τ )γ&ij for non . Flow phenomena such as rod climbing and jet swelling or viscoelastic effects such as fluid recoil. 1985. syrups.Newtonian or "generalized Newtonian" fluids Some practical fluid models and their applications can be summarized as follows: • The "power-law" model of Ostwald and deWaele is a two- parameter model for a wide variety of shear-thinning (n < 1) or shear-thickening (n > 1) aqueous solutions. slurries. the shear stress tensor τ for Newtonian fluids was rr r r rr linearly related to the shear-rate tensor γ& = ∇v + (∇v )tr .. τ has to be rr nonlinearly related to γ& . γ& ≡ γ& ≥ 1 s .Chapter I: Elements of Continuum Mechanics 93 • genuine non-Newtonian fluids such as exotic lubricants. Fig. As in multiphase flow modeling (cf. Churchill (1988). the Peclet number Pe = u r0 / Deff is indicative. 1998). Bird et al.. blood at low shear rates.20a. and other food stuff.e. i. simple polymeric liquids. Clearly.1). 1987). relative convection by an imposed flow. as well as polymeric liquids with high molecular weights. η = η (γ& ) .b) 2 i j . The steady-state shear flows of interest here can be basically described with an analytical representation of the shear-rate dependent viscosity. and stress overshoot are discussed elsewhere (Bird et al. Macosko. for incompressible fluids (cf. We deal only with the continuum approach and focus on simple polymeric liquids with a viscosity that is shear-rate dependent.4.4. two approaches for the development of “rheological equations of state” are commonly used: the continuum mechanics theory and the molecular dynamics theory. (1987). Zapryanov & Tabakova. where Pe = O(1) in order for the flow to be able to disrupt significantly the suspension microstructure and to produce a nonlinear. Sect.2. paints. pastes. rr In Sect.. 1987. 1. which fits specific experimental data sets (cf. Now. rr rr −1 1 η = mγ& n −1 . MW>104. stress relaxation.. etc. and so on. and • uniformly dispersed particle suspensions such as droplet sprays.4. where the presence or absence of “non- Newtonian effects” in homogeneous mixtures depends primarily on the rate of particle diffusion vs. Thus. time-dependent rheology (cf. and Tanner (1985). γ & = ∑ ∑ γ& ij γ& ji (1. 1. 1994).

b) µ p + γ& for τ > τ o  • The Casson model is a nonlinear extension of the Bingham model suitable for simulating suspensions of (spherical) particles in polymeric solutions. λ is a time constant that represents the “fading memory” of certain polymers.4. the dimensionless exponent n is equal to 1.21) where η o is the zero-shear-rate viscosity.22a. 1. • The Bingham model is a two-parameter formula for pastes.3 One Newtonian and three non-Newtonian fluids. For Newtonian fluids. i. including ketchup.4. .94 Biofluid Dynamics Bingham Newtonian τ plastic fluid Shear-thinning fluid τ0 Shear-thickening fluid γ& Fig. ∞ for τ ≤ τ o  η= τo (1. which implies that m becomes µ. and slurries exhibiting a threshold or yield stress τ o resisting motion.. λ = O(10 s ) to O(100s ) . • The Carreau-Yasuda model is a five-parameter expression that extends the application of the power law model to concentrated polymer solutions and melts.4.e. and a (typically a = 2) is a dimensionless parameter describing the transition between the zero-shear-stress and the power-law regions. η −η∞ ηo −η∞ [ = 1 + (λγ& )a](n −1) a (1. η ∞ is the infinite-shear- rate viscosity (typically η ∞ → 0 ).

Kleinstreuer et al.4. that is.4.b)  τ o + µ o γ&ij for τ > τ o • A modified Casson model can be employed to simulate blood rr rheology. which is represented by the stress tensor τ with the following relations: rr rr τ = 2η (II D )D (1.2 (dyne/cm2)1/2 and C2 = 0. 2 2 and the apparent viscosity η is a function of the shear rate 2 η (II D ) = 1 C (Ht ) + C (Ht ) 2 II  (1. (1.4).4. they took η = η γ& =1 = 0. (2001) discussed µ =0. Fig.24a) [ ] rr rr 1 r r r 1r Here.24b)] with shear rate γ& = 2 II D . which is the “zero-shear rate” condition (cf.4. based on Merrill’s experimental data (cf. For human hemodynamic studies. At the other end of the curve.24c) 2     Writing out the right-hand side in component form yields II D = D11 D22 + D11 D33 + D22 D33 (1. .4.24d) − D12 D21 − D13 D31 − D23 D32 The coefficients C1 and C2 were determined for Ht = 40% as C1 = 0.Chapter I: Elements of Continuum Mechanics 95 0 for τ ≤τo τ ij =  (1. ε ≡ D = ∇v + (∇v )tr = γ& is the rate-of-strain tensor. since Casson’s model is suitable only for γ& > 1 (s-1). 1.0348 (dyne.23a. 1968).24b) 2 II D  D  1 2  where Ht is the hematocrit and IID is the second scalar invariant of rr D . that is. An even better representation of the blood rheology is achieved with the Quemada model (see Buchanan & Kleinstreuer.4. which guarantees a smooth transition from the Casson fluid to a Newtonian fluid. Merrill.033 (cm2/ s). The following graph shows the variation of blood viscosity η [Eq. 1998). ν =0.1444 (dyne ⋅ s/cm 2 ) when γ& < 1 (s-1).s/cm2).4. rr 2 rr  1 II D =  traceD  +  traceD 2  (1.18 (dyne.s/cm2)1/2.

15: Casson-Fluid Flow through an Inclined Tube Calculate the volumetric flow rate of a Casson fluid through a slanted circular tube of radius R and length l. System: Assumptions: • steady laminar fully developed flow. x=l Postulates: du v = w = 0 and u = u(r) only.4. where − dp dx + ρg sin θ ≈ ∆p l = constant. γ&ij ⇒ γ&rx = . γ& rx = 0 for dr 0 ≤ r ≤ ro . 1.M errill(1 9 6 9 ) η [dyn s / cm ] 2 -1 10 -1 0 1 2 3 10 10 10 10 10 γ• [ se c -1 ] Fig. u(r) • uniform fluid motion x=0 r=R x when τ rx < τ o (where τ 0 r=r0 u0 r is a yield stress).4 Example of low-shear-rate blood rheology. p0 • shear-rate-dependent y r viscosity.96 Biofluid Dynamics 0 10 N ew to n ian C as so n E x p erim en t . and pl g • gravitational effect is θ r=0 absorbed in ∆p / l . Example 1.

we obtain ∆p at r = ro: τ rx r = r = τ o = ro (E1. applying integration by parts. (1. or (ii) expressing the equation for Q = ¢ in terms of r and integrating to find Q directly. Appendix C) 0=− dp 1 d + (rτ rx ) (E1.15-4) o  o ro  Solution: From the Casson relation [Eq.15-2) 2l Invoking the Casson model [Eq.4. the reduced equation of motion in the x-direction reads (cf. ro = τo.4. A = τ w / R . For the second approach we recall that τ R 2l 2l r = rx R or dr = dτ rx . .23b)] we know that − γ&rx = − du = 1 dr µ o ( τ rx − τ o 2 ) for τ > τ o (E1. (1.15-5) Now we have two possibilities: (i) Inserting τ rx = Ar .23)].Chapter I: Elements of Continuum Mechanics 97 Governing Equations: The Navier-Stokes equation is not applicable.15-3a) o 2l and ∆p at r = R: τ rx r = R = τ w = R (E1.15-3b) 2l r so that τ rx = τ w as for Newtonian fluids.15-1) dx r dr Integration yields ∆p • τ rx = r where τ rx (r = 0) = 0 (E1. Thus. and integrating to find u(r). Continuity is preserved R in terms of R  ro R  Q = 2π ∫ urdr := 2π  ∫ u o rdr + ∫ u (r )rdr  = const (E1. R = τw τw τw ∆p ∆p and.

Thus. a particle ensemble and the carrier fluid. [ ] 3 π  R  τw 2 Q=   ∫ τ rx τ rx − 2 τ rx τ o + τ o dτ rx µo  τ o  τ o 2lτ o ∆p Introducing κ = and using τ w = R . among many others. Additional applications of non-Newtonian fluid flow may be found in the books cited. countercurrent flows. can sufficiently model the dynamics of the two-phase flow field. is based on the concept of analyzing the mixture as a whole. The primary assumption associated with this model is that the mixture momentum equation.2 Drift-Flux Model The drift flux model. 1.” Examples include turbulent particle suspension flows where a drift parameter can be defined as the ratio of the particle ensemble drift velocity to the fluid phase rms fluctuation velocity. as well as in the articles by Kleinstreuer & Agarwal (1987) and Buchanan & Kleinstreuer (1998).4. we obtain after ∆pR 2l integration πR 4  ∆p  16 4 1  • Q=  1 − κ + κ − κ4 8µ o  l  7 3 21  Note: For τ o = 0 . along with the constitutive equations describing the drift or relative motion between the phases. cell migration toward the vessel wall in blood flow. making it possible to model flows in which a measurable drift exists between.2. etc. .98 Biofluid Dynamics R  R R  R  du  Q = 2π ∫ urdr = π ur 2 − π ∫ r 2 du  = −π ∫ r 2  dr o  o o  ro  dr  After substitution. the drift flux model is most appropriate when the motions of the phases are strongly coupled. κ = 0 and with µ o ≡ µ the well-known Hagen- Poiseuille solution is obtained. rather than as separate phases. like the pseudo-homogeneous equilibrium model. the advantage of the drift flux model is that it accounts for the relative motion between the phases. particle settling in sedimentation processes. say. particle pathlines differ from fluid element streamlines. Hence. However. known as the “crossing trajectory effect.

4.3 Separated Flow Models It should be noted that the adjective "separated" refers to (a) the physical separation of two immiscible fluids flowing in layers. However. they may shrink due to evaporation or grow due to condensation. while the generalized two-phase flow equations describing interacting gas-solid. i. and all relevant “point” forces acting on ..e. inhaled droplets may deform and in thermal flows or high humidity environments. the dispersed phase. Table 1. or distinct but interacting fluid/solid phases such as dense. or distinct particles in a carrier fluid such as solids in a gas or liquid. i. the trajectories of many individual particles can be directly calculated. and resuspension. is uncoupled from the continuous phase and. First the modeling equation for particle trajectories are based on the Eulerian- Lagrangian approach. and • fluid-particle interactions as expressed with Newton’s second law of motion. one describing the particles dispersed in a carrier fluid. such as size and composition.4.2). nonuniform particle suspensions. and bubbles in a liquid stream.3. • particle-wall interactions such as direct impacting. gas-liquid. rolling. bouncing. for accurate results the velocity field of the carrier fluid.1 Particle Trajectory Models There are various forms of (solid) particle interactions causing up to four-way coupling phenomena. • particle-wall interactions such as direct impacting. or bubbles. or an apparent viscosity of the particle phase. 1. separated flow examples include two stratified fluids such as oil on water and steam in a partially filled water pipe. or particle aggregation/droplet coalescence. interception. 2003) require the Eulerian-Eulerian approach. interception. respectively. solid particles. and the other describing the continuous phase (the fluid). avoids numerical diffusion and allows readily incorporation of particle characteristics. Examples include: • particle-particle and particle-fluid interactions influencing individual particle trajectories and fluid flow between neighboring particles. as well as turbulence modulation in dense particle suspensions. difficulties arise when the particles are nonspherical. Thus. are interacting or randomly diffusing.Chapter I: Elements of Continuum Mechanics 99 1.e. or with dispersion terms in the enhanced mass transfer equation. quite frequently (cf. as well as (b) the mathematical need for two separate sets of equations. generate large wake effects. bouncing. the Lagrangian description of particles. or a drift flux. Furthermore. In any case. given the particle release conditions. and liquid-solid media (see Kleinstreuer. rolling. and resuspension. the particle characteristics. In fact.4. droplets in a gas. droplets. • random interparticle collisions resulting in Brownian-type motion for submicron particles.. However. particle tracking.

25a) dt where r r r r r r 1 dˆ( v p − v) ∑ Fvolume = Fbuoyancy − Fvirtual mass = ( m p − m f ) g − m f 2 dˆt (1. i.4. In Eq. for its position in the fluid flow field. (1. Hence.4.density ratios 5 ≤ ρ p / ρ f ≤ p 200.4. D Dt ≈ dˆ dˆt = d dt .26) to (1. single particle transport will involve the solution of a general equation of particle motion. (1. However.25c) Equation (1.4. i. 2001). i. The initial assumptions will be that the calculated flow field is undisturbed by the presence of particles and that particles are spherical with negligible Magnus effect. Note that in these equations r dˆ dˆ t = ∂ ∂t + v p ⋅ ∇ .4. Single spherical particles..25b) and r r r r r r ∑ F surface = F drag + F pressure + F Basset + Flift + Finteraction (1. in contrast to the substantial derivative. particle rotation is insignificant.4. Ideally. DNS is still cost prohibitive (cf.25b) and (1. a time derivative following the moving r particle. pressure. all surface and exchange forces should be accurately integrated from the fluid-particle and particle-particle interactions. Basic particle equation in laminar flow..4. Newton’s second law. The kinematics of a particle in a fluid depends on the external forces imposed on the body by the suspending medium or carrier fluid. and stress fields surrounding each particle. when Re p = u − v d p / ν << 1 . which follows a moving fluid element. the first .100 Biofluid Dynamics the particles have to be known. Joseph. for example. employing direct numerical simulation (DNS) of the velocity. A historical review of the various particle trajectory equations and their applications has been provided by Michaelides (1997) and Kim et al. (1998)..e.e.25b).31). who also proposed a new particle trajectory equation for wider ranges of particle Reynolds numbers 2 ≤ Re d ≤ 150 and particle-to-fluid. The starting point in this section is the complete dynamics equation for a single solid sphere.25a) is simply a form of Newton's second law where the volumetric and surface forces are actually “point forces” as specified in Eqs.4. The equation of motion for a representative particle in the Eulerian-Lagrangian framework may be written as: r dˆ v p r r m p ˆ = ∑ Fvolume + ∑ Fsurface (1. however.e. D Dt = ∂ ∂ t + v ⋅ ∇ .

4. Re p = v − v p d p / ν . i. The Basset force. 2 FFaxen  d p  ~  (1. requires the addition of a correction. r ρ 1 r r (r r ) 3 Fdrag = m p ℑ v − v p .646 24 / Re p for 1 < Re p ≤ 400 while for liquid spheres.28) t0 π v ( t .26f) FStokes  L  The pressure force. r r F press = − m f Dv Dt ( rr ≈ −V p ∇p + ∇ ⋅ τ ) (1.4.4. CD = ( 3. i. and for solid spheres.Chapter I: Elements of Continuum Mechanics 101 term accounts for the inclusion of the buoyancy effects indicated by the difference in densities and the second is the virtual mass due to the inertia of the particle. The friction coefficient CD is a function of the particle Reynolds number.27) captures pressure gradient effects. The surface forces accounted for in Eq.e.v& p -factor).4.t′ ) dˆt ′ arises from the acceleration of the fluid around the particle.4. d p = 2a . Clearly.4.e.26a. the Faxen force (Happel & Brenner. r r r t dt ′ dˆ( v − v p ) FBasset = 6 π a 2 µ ∫ (1.25c) are the viscous drag. or time history term.. a ratio can be formed that determines the relative importance of the Basset force . r d p3 2 r Fdrag = 3πµd p (v − v p ) + µπ ∇ v r r 144244 3 148243 (1..26c) 0. 24 / Re p for 0 < Re p ≤ 1 CD =  (1. Considering the history term as an additional form of resistance. Hence. 1973). a temporal ( ) velocity changes with time (see v& .26e) Stokes Faxen r where ∇ 2 v is evaluated at the deposition of the particle. Thus.4. nonuniform shear field.05 783κ 2 + 2142κ + 1080 Re −p0. ℑ = CD f |v−vp | (1.26d) (60 + 29κ )(4 + 3κ ) for 4 < Re p ≤ 100 with κ ≡ µ p /µ .b) 4 ρp d p where ℑ is the inverse time constant of momentum transfer. assuming uniform Newtonian fluid flow. (1.4.74 ) (1. especially near the wall.

” has to account for the influence of a boundary on wall- normal and wall-tangential particle motion (see Sect. unit variance-independent Gaussian random numbers. ∆t .31b) π d p S Cc . and the spectral intensity is µkT S 0 = 216 2 5 2 (1. when particles are large. For example. as in near-wall regions.e.102 Biofluid Dynamics FBasset 18 ρ = ar (1.31a) ∆t where Gi are the zero-mean. and/or Brownian motion effects. (cf.30a) ρ d d p (ε lk ε kl )1/ 4 where the deformation-rate tensor is  1 ∂v ∂v j  ε ij =  i +  (1.5. Lift forces are caused by particle rotation either induced by strong shear flow where velocity gradients generate particle rotation leading to top/bottom pressure differentials (cf. rotate. For example.4. or due to prerotation. the bombardment of molecules of the carrier fluid on submicron particles is encapsulated in the Brownian force.29) FStokes π ρp where a r = τ p / ∆t is the acceleration rate with τ p = ρ p d p / (18µ ) being 2 the particle relaxation time. Clearly.. Saffman lift).30b) 2  ∂x j ∂xi  Alternative forms for Flift may be found in Crowe et al. when a r <<1 and/or ρ ρ p <<1. Interaction forces.4. etc.4. a modified Stokes drag. i.2m p ν ( ρ v j − v j ε ij p ) (1. Li & Ahmadi (1995) suggested a Saffman lift force for supramicron particles in flows with high shear rates as Fi lift = 5. called the “lubrication force. particle collision. and/or are subjected to strong shear flows. the history term is negligible. In wall-bounded suspensions. flow turbulence. Clearly. It can be modeled as a Gaussian white noise random process where the amplitudes of the components at every time step. Fint er . 1992) mp πS 0 Fi Brown = G i (1. wall contact.4. are due to random processes.2). Magnus lift in conjunction with a particle torque equation). lift forces should be considered. (1998) and Fan & Zhu (1998).4. are (Li & Ahmadi.

turbulence effects are incorporated in the fluid velocity vector.38 × 10 −23 J/K is the Boltzmann constant. an individual particle trajectory or dispersion of fine particle clouds is measurably influenced by the random velocity fluctuations of the continuous phase. S is the ratio ρ p ρ .4. v = v + v' . Sect. and C c is the Cunningham slip correction factor 2λ Cc = 1 + dp [ 1..4.31c) i.e.4.. only the time-smoothed or bulk flow velocity vector is calculated (cf.0 when d p >> λ . k = 1. Rather than creating a separate turbulent fluid-particle interaction force.e. when employing Reynolds’ r r r decomposition of the instantaneous fluid velocity.e. v ′(i2) is the mean-square of the ith fluctuation velocity. 1999) 1/ 2 r 2  v' ≡ v 'i = λ  k i  (1.1d p / 2λ ) ] (1. For example. r r This implies that either v' or v has to be recorded and integrated into a particle trajectory/dispersion equation in order to simulate the stochastic process of particle motion in a turbulent flow field.Chapter I: Elements of Continuum Mechanics 103 Here.32) 3  2 where λ is a random number. In turbulent flows. i. the time-smoothed fluid velocity v := v i is augmented with Eq.. the r fluctuating components of v' can be expressed as (Gosman & Ioanuides. − 1 ≤ λ ≤ 1 . 1981. especially the normal component.3k / ε is the particle integral time. and ς i (t ) is a Gaussian vector white noise random process given by ς i (t ) = Gi / ∆t at . Specifically. and k ≡ 1 2 ui' is the r turbulence kinetic energy. which has to be accurately simulated (or measured). C c ≈ 1. 1999): 1/ 2  2  dv i v i − v i  2 v ′(i )  =− +  ς i (t ) (1. the instantaneous fluid velocity is (directly) obtained from a stochastic equation (cf. Alternatively.33) dt τI  τI    where τ I ≈ 0.. He & Ahmadi.4. Now.3. i.257 + 0. the turbulence intensity.e. i. Katz & Mortonen. which depends on the turbulence kinetic energy k and dissipation ε . T is the fluid temperature [K].4 exp(− 1.4).32) to form the instantaneous velocity v i + v ′i = v i . the molecular mean free path. µ is the fluid viscosity. 1. (1. here related to the Lagrangian fluid element integral time.

.e.25) since the instantaneous drag may differ radically from the corresponding steady drag. Another (random) interaction force is due to particle-particle collisions.25) is a very challenging problem especially if the Basset (i. then the transient effect terms (i. bubbly flows..4. unit variance-independent Gaussian random numbers. 1994). Thus. e. for example. if the unsteady motion is of type I. Fortunately for many problems Eq. the virtual mass. 1998. (1. and Basset forces are often negligible due to the relative density ratios. pressure.25) can be substantially reduced. However.104 Biofluid Dynamics every time step ∆t with Gi being zero-mean.g. Basset and virtual mass) should be retained in Eq. (1978). among others).4.e.4.34) 2 where CI is a semiempirical interaction coefficient that depends mainly on the probability of particle collision and intraparticle collision impact. from the “soft sphere” collision model (Crowe et al. Obtaining a numerical solution to Eq. In practice the type II condition usually corresponds to gas-particle systems with their corresponding high density ratios... 1998. where 0 ≤ λ ≤ 1 is a random number and the normal/tangential impact I n . This condition is commonly found in liquid-particle systems where the density ratio is low to moderate. characterized by rapid changes in Reynolds number with the displacement modulus. assuming appropriate (collisional) probability density functions.t can be obtained. Also Brownian motion may be ignored for the particles with effective diameters d p > 1 µm. I n .25). For unsteady motion of type II (characterized by slow changes in Reynolds number with the displacement modulus). (1. Simplified particle equation. Specifically. Alternatively. Numerous models have been postulated (cf.25) can be reduced to the following form: r d 2x p r r r mp = ∑ F p = ∑ Fbody + ∑ Fsurface (1. history) term must be retained. kinetic theory formalism can be applied to account for interparticle collisions in the model approach of the dispersed phase. Crowe et al. (1. As pointed out by Clift et al. thus greatly simplifying the solution process. especially in “dirty gas” flow.e. one rather simple approach given here is to construct r r an expression for Fcollision similar to Fdrag .t ) . i. Bird.. (1. particle and molecule systems have basic differences as discussed elsewhere (cf. it is generally acceptable to drop the transient effect terms such as the Basset and virtual mass terms in Eq. As a result. while the interaction force is neglected if the suspension is dilute. r ρ r Fcollision = C I As (v p − v ) v p − v r r r (1. Fan & Zhu.35a) dt 2 . Eq. C I = C I (λ . 1996). (1..

Example 1.4.35) can be reduced to ρ • mp du p dt ( = FD = AC D u − u p u − u p 2 ) (E1.687  − r = 1 + 0.16-1) ( ) Here.15  u r0.16: Accelerating Particle in Uniform Steady Flow Sketch: Assumptions: Concepts: u00=u • steady unidirectional • carrier fluid r flow w/constant v = (u .4.1 and 5.3). in terms of ur.0 ) mp properties where u=¢ • spherical particle • particle up • drag dominated dynamics particle motion du p ρ.1. (E1. In case of turbulent flows. as discussed (see also Sects. particle dispersion has to be considered via an r instantaneous system velocity v defined as the sum of the mean velocity and a randomly fluctuating velocity component.4. 3.35c) r For laminar flows the carrier fluid velocity vector is v . noting that du r / dt = − du p / dt . C D = 24 Re d 1 + 0. Eq.35b) 8 and r r = Fbuoyancy = πd p3 (ρ p − ρ )g 1 r ∑F body 6 (1.Chapter I: Elements of Continuum Mechanics 105 where r r ∑ Fsurface = Fdrag = πρd p C Dp (v − v p ) v − v p 1 2 r r r r (1.16-2) dt ρ p d p2    µ    .4.16-1) reads with A = d 2p π / 4 and m p = ρ p πd 3p / 6 .687  u r2 (E1.15 ⋅ Re 0d. where Re d = ρ urel d µ and u rel ≡ u r = u − u p . µ mp = FD dt Solution: Equation (1. du 18µ   ρd  0. Now.0.687 .

” i.. fluid flow Release position of red i. ur ur time t Example 1.e. i.15  and b =  µ  ρ p d 2p Graph: u p . v(t ).687  ρd  18µ a = 0. ω t = −71.4o of a cell (RBC) pressure.e. sinusoidal input pulse • drag dominated and “virtual particle motion mass. 1977) Sketch: Assumptions: Concepts: u(t) Centerline • transient • carrier fluid r axisymmetric v(t ) = (u (t ). angle –71.0) A. µ constant properties motion due to • spherical particle.. hardened induced “point blood cell A with phase human red blood forces.687 ⋅ exp(0. d p / D << 1 .106 Biofluid Dynamics with the particle initially at rest..” Solution: A one-way coupled point-force model can be applied to simulate the trajectory of the particle in the annular expansion due to the size of the particle in relation to the tubular diameter.4 o internal flow with • 2-D particle ρ. Integration of this nonlinear ODE after separation of variables yields (see graph) [( ) u r ≡ u − u p = a + u r−0. u r (t = 0 ) = u .16-3) where 0.e.687 (E1.17: Particle Trajectory in a Recirculation Zone (Karino & Goldsmith.687b t ) − a ] −1 / 0. drag.

the computational effort required has increased 40 times. the inclusion of only the drag term under identical release conditions reduces the number of particle loops by one and slightly modifies the location at which the particle ends its recirculating trajectory. The first trajectory.531 × 10 −6 (E1. i. If virtual mass effects are to be included. Karino and Goldsmith (1977) also studied the motion of a 55-mm aggregate in the 504-mm- diameter expansion. 1. respectively. is nearly identical to the RBC trajectory of Karino & Goldsmith (1977).17-2) ρd The governing dimensional equation.17-3a) dt Dt τ p where vi and ui are the components of particle and local fluid element velocity. and virtual mass terms over separate runs are illustrated in the Graph. Using CFX-4.4. as indicated in the trajectory plots by the number of time steps required by the quality control integration routine. such as an aggregate of blood cells. assuming only pressure and drag effects. the above equation can be written as dv i 3 Du i β = βκ + (ui − v i ) (E1.. the results of sequentially adding drag.693 (E1. can be written as [cf.17-3b) dt 2 Dt τp where 1 β= := 0. Properties of the RBC include a particle response time of 2ρ d a 2 τp ≡ = 3.1. Including both pressure and drag effects modifies the particle trajectory such that it again very nearly resembles that of the RBC. assuming fluid element pathline track.4.17-1) 9µ c and a continuous to dispersed phase density ratio ρc κ= ˆ = 0. may not be sufficiently small to have a negligible effect on the surrounding flow field. Then.17-4) (1 + κ ) / 2 Solution: Particle A (red blood cell) trajectory plots considering different point forces.Chapter I: Elements of Continuum Mechanics 107 Larger irregular particles. Interestingly.(1.25a) with Eqs.3. For example.885 (E1. pressure.e. Eq. both . (1.4.3 on an engineering workstation. Sect.27)]: dv i Du i =κ + 1 (u i − v i ) (E1. Furthermore. Resolved-volume techniques may be required to simulate such a motion.26) and (1.4.

however. e. Pathline (Nearly Identical to h. ….3.4 Number of Steps: 75. particle diffusivity DM..b.e.108 Biofluid Dynamics pressure and drag terms have an influence on particle trajectory.e.(1. 1977) Same as in Karino & Goldsmith (1977) Particle Trajectory with Drag Release Time: 0. i. Pressure. For example. can be written as [cf. the suspension is categorized into n domains according to the .36) ∂t ∂x j ∂x j  ∂x j  where Mk represent the number (or fraction) of particles in section k. i. an Eulerian-Eulerian approach should be considered. k = 1.534444 Phase Angle: -71.4 Number of Steps: 1811 Loops: 9 (as with Karino. Trajectory) Release Time: 0. the general dynamics equation. and particle sink/source term QM.4 Number of Steps: 71.3..4. requiring knowledge of the continuous phase velocity field uj.000 Loops: 8 Particle Trajectory with Drag and Pressure Release Time: 0.c.g.210 Loops: 9 Particle Trajectory with Drag.2 Species Mass Transfer When submicron particles are convected by the carrier fluid.36)]: ∂ ∂ ∂  M ∂ ( ρM k )  ( ρM k ) + ( ρM k u j ) =  Dk  + Qk M (1. and Virtual Mass Release Time: 0.000 Loops: 9 1. these terms apparently cancel one another.534444 Phase Angle: -71. their net effect is essentially represented by the motion of a pathline for this system.4 Number of Steps: 75.534444 Phase Angle: -71. n. Eq. inhaled air. 2.534444 Phase Angle: -71.4..

Although it provides mathematical convenience in describing. the pressure drop and the resulting velocity field (or flow rate) are the key unknowns.or gravity-driven fluid. particles packed or suspended in a container.. Extending Eq (1. DT is the turbulent eddy viscosity. 1978). PrT=0. the reader should keep in mind that the use of Darcy’s law in biomedical engineering is controver- sial. geo- logic material such as sand. α. as well as tissue modeled as a fiber matrix. 1. while “two-fluid” modeling.9 is the turbulent Prandtl number.37b) 3πµd p Here. tissue perfusion. Cslip is the Cunningham slip correction factor (Clift et al.= fct  Re.38×10-23 J/K.e. Clearly. and pebbles. Clearly.3 material may be found in Sects. is discussed by Kleinstreuer (2003).4. shale. ∂C ∂C ∂   +uj =  ( D p + DT ) ∂C  (1. or granular struc- ture.4 Porous Media Flow Flow equations. limestone.4. ---- ------p- -------.37a) ∂t ∂x j ∂x j   PrT ∂x j  where C is the particle concentration. 1. clay. a proper analysis of “flow through porous media” is important in chemical/petroleum engineering. etc .6.38)  l  2  d  ρ u0 p .Chapter I: Elements of Continuum Mechanics 109 given particle size distribution.4. kB=1. Nevertheless. environmental/civil engineering. 3. ξ. e- . the interaction of two separate phases. i. for monodisperse nano-particles k=1. ϕ. for example. A porous medium can be a solid structure with continuous interconnected conduits.. it can also be written for ultrafine particles (or vapor) in more familiar form as the convective- dispersion equation.. Specifically.4. Thus. and mechanical/biomedical engineering. soil. As with all internal flows. Applications of Sect. i. a functional relationship for the dimensionless “pressure gradient” can be stated as  –∆ dp .36) to turbulent flow.4. (1.e. and the particle diffusivity is k B TC slip Dp = (1. it ignores the intricacies of different living tis- sues. fixed or suspended particles in the flow path may greatly increase the resistance encountered by the pressure. and dp is the particle diameter. sponge-like structure.3 and 5.4.

where k is the medium permeability. i. ν dp ξ = ---------------. i.110 Biofluid Dynamics 0 p u d where d p is the effective particle (or cell) diameter..e. in 3-D it reads r k r v 0 = − (∇p − ρg) (1..b) ∂t A simpler version of the momentum equation for steady incompressible flow through an isotropic porous medium was suggested by Darcy in 1856.22)].18: Evaluation of the Hydraulic Conductivity K = k ⁄ µ Assuming a porous medium to be a solid structure with n parallel tubes. Example 1. and k ⁄ µ ≡ K is the hydraulic conductivity. Sketch: Concept: x Compare steady laminar fully-developed pipe flow result with Darcy’s law in D 1-D.3. employ the Hagen-Poiseuille result for steady laminar pipe flow to estimate K.(1. i. ϕ is the e system sphericity. relatively small volume containing both the solid matrix and the fluid. Q ---.= v 0 = – K dp ------ A dx Q .40) µ r where v 0 is the superficial velocity vector which is averaged over a local.. Re → 0 and Stokes flow can be assumed [see Eq. and e ⁄ d p is the relative surface roughness.4. Re = ----------.39a. α is the medium porosity (or void fraction).e.e. r r ∂v r r ∂ (αρ ) ∂t + ∇ ⋅ (αρ v) = 0 and ρ = −∇p + µ∇ 2 v + ρg (1.. With d p « 1 or the equivalent conduit or open passage hydraulic diameter D h « 1 .4..

nonlinear inertia effects appear and when Re ≥ 100 . such insurmountable challenges were answered with (one-dimensional) empirical correlations.= µ ----------0. the void spaces containing two fluids (e. Extended Darcy equation. we set dx A 4 Q H – P = Q Darcy and the hydraulic conductivity turns out to be 2 D • K = --------. all fluid passway geometries have to be known as well as inlet/outlet conditions and the conduit wall roughness.e.18-1) 32µ In order to solve Eq.6).. 2 r 0 dp 1 –  ---- .= 2π ∫ n 0 128µ  dx 2 dp Q D π Thus. where C = 0. r 2 • v ( r ) = v max  r 0 4µ  dx we obtain for the volumetric flow rate per pore/tube: D⁄2 π D dp 4 v r dr = – -----------. v max = – -----. When the Reynolds number is greater than one. with v 0 = – K -----.55 is typical (see Nield & Bejan. (1.. u 0 is the volume-averaged (i. .– -. A = n ---------.(1. ------ . ------ Q ---.e.4. ρC 2 2 u 0 + ------.39). Historically.40) which was extended by Forchheimer in 1901 and Brinkman in 1947.Chapter I: Elements of Continuum Mechanics 111 Solution: Considering Poiseuille flow (cf. for steady fully-developed laminar flow in a 2-D isotropic porous medium 2 dp ˆ d u µ- -----. Further complications may arise when the solid phase moves.7.1 to 0. 1999).g.41) dx dy k k [Stokes/ [Darcy] [Forchheimer] Brinkman] where pˆ = p + ρ gx . Specifically.. i.4. air and water as in unsaturated geologic media) or a non-Newtonian fluid. superficial) axial velocity. (E1.= ---.4.u 0 (1.. the flow is turbulent.. starting with Eq. k is the medium-dependent permeability and the empirical factor C varies from 0. Example 1.

= 0 (E1. inhomogeneous second-order ODE with constant coefficients has the solution cosh  -----.4.= 1 – --------.19-4) k ⁄ µ dp ⁄ dx h  2 k . Sketch: Assume: Concept: • Creeping flow. 2 k. h • --------. (E1.u 0 = cons tan t dp d u • (E1. i. kh dp 2 k h • (E1. e.--------------.19-1a) dx 2 k dy subject to du 0 u 0 ( y = h ⁄ 2 ) = 0 and -------. ------ 1 – ---------. y µ u0=u0(y) only Use of Brinkman Q x h k • Constant pressure equation to invoke gradient no-slip wall • Constant properties conditions Solution: Equation (1.19-2) µ  dx cosh  ---------.41) can be written without the “high-speed” Forchheimer term as : 2 -----. h 2 k  h⁄2 ∫0 ˆ = 2 The flow rate per unit depth Q u 0 dy is u 0 ( y ) = – -----. the uniform space between two cells where lymph flows through an extracellular matrix.– --µ.= µ ----------0.19: Stokes’ Flow in Interstitial Spaces Consider a porous-medium channel.112 Biofluid Dynamics Example 1..tanh  ---------.. y  k u 0 ( y ) = – --.e.c) dy y=0 This linear. K .19-3) µ dx   h  2 k and the dimensionless effective hydraulic conductivity is u0 tanh  ---------.≡ . ------ 1 – ---------------------------- k dp • (E1.19-1b.g.

43) 180 ( 1 – α ) where it has to be recalled that the porosity at a wall may double. i.d . α  -------- ∆ pˆ.= 180µ (1 – α)- -------------------------------- 2 u0 (1. (1. α.2 0 -2 x 10 10-1 100 101 102 103 h/(2k0.5) A celebrated special case of Eq. and porosity. (1.40) is the Carman-Kozeny equation describing laminar ( Re ≤ 1 ) fully-developed (Poiseuille-type) flow in a tube filled with spheres of diameter d p .4. passages.e.8 0.4 0.= C f (1.4.Chapter I: Elements of Continuum Mechanics 113 Graphs: y 1 y=h/2 0.. permeability. In turbulent flow through packed beds. creating a constant void fraction α: 2 ∆ ˆ ------p. ----------. have been related as 2 3 dp α k = ----------------------------2.44) L  1 – α ρ u 2 0 . and u 0 = Q ⁄ A tube = α u n is the superficial velocity (or volumetric flux density) while u n is the corresponding intrinsic average or mean velocity through the pores.6 K/(k/µ) u0(y) 0.42) L ( α dp ) where pˆ = p + ρ gL – p 0 . or channels.p. void space.4. k. the pressure drop is largely due to inertial effects which leads to the Burke-Plummer equation 3 –--------.4. For a packed bed of spheres. L being the (vertical) tube length.

20: Radial Flow through a Porous-Walled Tube Consider pressure-driven flow through a porous-walled tube. Ergun (1952) combined Eqs.45)  L   1 – α 2 ρ u0 dp ρ u0 Example 1.---..42) and (1.40). r k ∇ ⋅ v 0 = − ∇ 2 p and recalling that for an incompressible fluid µ r ∇ ⋅ v 0 = 0 .20.4.(1. after successful comparisons with experimental data sets.0 .1) r dr  dr  subject to p ( r = R 1 ) = p i and p ( r = R 2 ) = p o . Sketch: Assumptions: Concept: p0 • steady radial seepage • Darcy’s through R 1 ≤ r ≤ R 2 equation • constant pressures • mass V0 r R2 and system properties balance R1 pi m& L Solution: Taking the divergence of Eq.4.4. seepage into a lymph vessel. we have to solve in cylindrical coordinates: 1 d.2 ≤ C f < 4.e.e.75 (1.114 Biofluid Dynamics where the friction coefficient is in the range 1.. i. dp- r = 0 (E1. Double integration yields p ( r ) = C 1 ln r + C 2 and ultimately .75 . ----- --.4. d -------------------. (1. i. replacing the factor 180 by 150 and setting C f = 1.44).+ 1.= 150 µ ( 1 – α )-  –--------. for example. 3 ∆ pˆ-  α  -------- -----------. Find the p(r) and v 0 ( r ) profiles as well as the added mass flow rate over the tube length L.p.

7]:  ∂T  (1 − α ) ρc  = (1 − α )∇ ⋅ (k∇T ) s + h As − f (T f − Ts ) + (1 − α )q&s  ∂t  s ∀M (1. . T f . h. (E1. differs locally from the fluid temperature.20.4. so that 2π k ρ∆ p • ∆ m· = -----------------------------.Chapter I: Elements of Continuum Mechanics 115 p–p ln ( r ⁄ R 1 ) • ---------------i = ------------------------- . soft tissue constitutes a major challenge. h is the convective heat transfer coefficient. say. T s .5) µ ln ( R 2 ⁄ R 1 ) Heat and mass transfer in porous media. and Nakayama et al. Recent porous medium applications may be found in Alazmi & Vafai (2002). (2002).2) po – pi ln ( R 2 ⁄ R 1 ) r For purely radial flow v 0 = (0. Nield et al.4.L (E1.35) and HWA in Sect.3.--. (1. and A s – f for.3) µ  dr  so that with Eq.0) and Eq.4) µ ln ( R 2 ⁄ R 1 ) r A 1-D (radial) mass balance yields ∆ m· = – ρv 0 A surface where A surf = 2π R 1 L . v 0 . A s – f is the solid-fluid contact area. and q· is 3 the local heat generation rate per unit volume [ J ⁄ ( m ⋅ s ) ].2) k ∆p 1 • v 0 = – --.46b) where α is the porosity.4. two heat transfer equations can be written as [see Eq.20. (1. ∀ M is the porous medium volume. Clearly.20.46a) ∂T   α  ρc p r  + ( ρ c p v 0 ⋅ ∇T ) f = α∇ ⋅ (k∇T ) f + h As − f (Ts − T f ) + αq& f  ∂t  f ∀M (1. (2001). For a volume-averaged homogeneous medium where the solid-phase temperature.20. ------ k dp (E1.40) reduces to v 0 = – --.-------------------------. 1. (E1. determining system parameters such as α .20. (E1.

49) where σ is the reflection coefficient [ 0 ≤ σ ≤ 1 .21: 1-D Diffusion across a Neutral Biomembrane The selective exchange of material between fluids and compartments occurs through two types of biomembranes.4. noncharged) biomembrane (or filter). plasma (or cell) membranes and capillary walls. the species mass transfer equation describing transient convection.48a) r • Turbulent dispersion: j = -DT ∇c (1. c s is the average solute concentration within the membrane. r k v = − (∇p − σΠ ) is the average mixture velocity..e. σ = 0 no restriction for solute passage through pores and σ = 1.4.4. where D M =æ D p with æ<1 and D p given as Eq. 1997) could be formulated as r r js = (1 − σ ) v 0 cs − DM ∇cs (1. diffusion..4.46b and 1. j is the flux vector.0 rejection of solute passage].116 Biofluid Dynamics Similar to the scalar equations (1. Starling’s hypothesis (Michel. Example 1.4.4. (1.4.. Both are very thin (w = 7 -500 nm) . i.48c) • Nernst-Planck ion mass flux due to an electrical potential gradient across a barrier B: r zF j = -DB c∇ψ (1.4.4.e. and conversion can be written as ∂c r r α + ( v 0 ⋅ ∇)c = −∇ ⋅ j + ∑ S c (1. 2002): r • Fickian diffusion: j = -D∇c (1.37).37b).48b) • Soret mass flux due to temperature gradients: r j = -DS ∇T (1. (1.48d) RT For solute flux across a neutral (i.47) ∂t where c is the solute concentration of the homogeneous mixture flowing r [Eq.40)] in a porous medium of porosity α .4. p is the fluid static µ pressure and Π is the osmotic pressure. and D M is the membrane diffusion coefficient. accommodating potentially (Bird et al.

and ∂ c ⁄ ∂ x = 0 at x = w ⁄ 2 . Consider transient 1-D diffusion across a thin neutral membrane and develop an equation to obtain the membrane diffusion coefficient for different solute-membrane conditions. • No solute sinks time w or sources inside membranes Solutions: Based on the assumptions. static pressure difference.48a) can be reduced. nπx  n π DM  sin  --------- exp  – ------------------- c-------------- c0 = 1+ ∑ ----------------------- nπ  w   α w2  . Eq. (1.= D M -------2. 1993 or Gates & Newman.4. DM c/c 0 vs. w ) = c 0 = const : 2 ∂c ∂ c α ----.Chapter I: Elements of Continuum Mechanics 117 with tiny pores ( d ≤ 6 nm ) where solute and water transport are caused by diffusion.t) α. with Eq. etc. electro-chemical potential. Sketch: Assumptions: Approach: x Reservoir • Simple 1-D • Analytical c0 = c diffusion solution of c 0=c t c1 =  only the diffusion cw = c equation • Constant properties • Plotting of c(x.1) ∂t ∂x subject to c ( t = 0 ) = 0 for 0 < x < w . (a) Analytic solution when c ( x = 0.t (E1.2) n=1 . Separation-of-variables method yields (Özisik. t -) cos n π – 1. (1. osmosis. 2000): ∞ 2 2 ( x. (E1. c ( x = 0. w ) = c 0 .

21-4) 2  w  .t (E1.5 0.. t ) ⁄ c 0 is known.0 2 2 [π D M /(αw )]t 2 π DM Once c ( x.6 c/c0 x/w=0.21-3a.0 1.0 5.b) dt ∀ w w Integration.8 x/w=0.0 0. i. ∂c0 c −c ∂c −∀ = ADM 0 w = ∀ w ∂t x =0 142 w4 4 43 ∂t x=w 0< x < w so that with Α/∀:= w <system length scale> d (c0 − cw ) = −2 A jx = −2 DM c0 − cw (E1. subject to c 0 ( t = 0 ) = c initial =¢ and c w ( t = 0 ) = 0 .118 Biofluid Dynamics Graph: 1. yields ci 2 DM c 0 ( t ) = ---.1 0.e.2 0.1 + exp  – ----------.t from the chart and 2 αw then calculate the diffusion coefficient D M .4 x/w=0. (b) Approximate solution when c ( x = 0 ) = c0 ( t ) and c0 t c ( x = w ) = cw ( t ) : ∀ ∀ For thin membranes the solute flux cw can be linearized as A j x = D M ( c 0 – c w ) ⁄ w so that for a w 1-D mass balance.0 0. we can read -------------.4 0.3 x/w=0.0 4.0 2.0 3.2 x/w=0.

5 2.21-4) indicates a characteristic diffusion time for the membrane 2 tM = w ⁄ DM (E1. we can read -------.5 1. .Chapter I: Elements of Continuum Mechanics 119 Equation (E1. Graph: 1.8 c0/ci 0.5 3.4 0.0 0.0 (D M /w2)t M D Again.0 1.0 0.21-5b) 2∀ M where ∀ >> ∀ M and hence t M « 1 .0 2.21-5a) whereas the characteristic times for solute concentration changes at both sides of the membrane are ∀ t0 = t w = tM (E1.t from the chart and 2 w then calculate the diffusion coefficient D M . once c 0 ( t ) ⁄ c i is known.6 0.

g.. lymph flow and tissue motion. Clearly.). artificial valves. fluid-structure interactions (FSI) are an integral part of most transport phenomena in biological systems. In this section. filters. while soft tissue may experience finite strain and hence relatively large deformations (see Fig.1. etc. blood flow interacting with implants (e. Typically. More realistic models are introduced and applied in Sect. 1. e.g.5. Other FSI examples are dilute suspension flow across a deflecting cell membrane. 1. In this section.1 Introduction As mentioned. Examples include pulsatile blood flow interacting with moving vessels.4). pumps. strain. 1..5. in conjunction with the conservation laws (Sect. and arterial wall are discussed. However.120 Biofluid Dynamics 1. the state of material interaction at any point in a body is specified by the stress tensor . stent-graft. ranging from solute transfer across cell membranes to heart-valve actions. 1. as a first approximation it is reasonable to model some materials as anisotropic. 5. joint lubrication and deformable cartilage.2 Principal Stresses As reviewed in Sect.4 may be interactively coupled with the material of Sect. under the umbrella of continuum mechanics. and deformation (see Sect. material of Sects.3. valves. heart-lung bypass machines. synthetic grafts.3). elements of fluid kinematics and dynamics of Sects. a pressure due to fluid flow generates a load on a structure which results in wall stress. 1. Continuum mechanics books catering to biomedical engineers include Fung (1994) and Humphrey & Delange (2004). contracting/expanding heart chambers as well as opening/closing heart valves.5. acting on body continua. are made out of layered viscoelastic materials. Most of the biological structures. airflow in flexible bronchioles. blood vessels and airways.). and ultrafine particle deposition in the expanding/contracting alveoli. vessel inserts. prepare to solve FSI problems. stagnant cavity blood. 5. to name a few examples.2.5 BIOMECHANICS REVIEW Fluid dynamics theory applied to biological systems or biomedical devices cannot ignore the frequent coupling of momentum transfer with structural mechanics.4 where nonlinear interactions between blood flow. etc. the basic solid mechanics equations are reviewed which. 1.5. and fluid flow interacting with moving parts of medical devices (e. stents. plasma-cell migration into a moving arterial wall.3 are extended with a review of solid deformation (strain) caused by forces (stress).g. drug delivery systems.4). 1. and low-drag body motion. linearly elastic solids with Hooke’s law as the appropriate constitutive equation.3 and 1.2 and 1.

and α& is the angular acceleration vector. i.e. Example 1. r force times distance.1. where m is the material r r r r mass.b) σ 31 σ 32 σ 33 τ zx τ zy σ z where σ ii are the normal stresses.5. (b) 2-D force components and stresses in Ay . 1.1). As alluded to in Sect.e. This equality of shearing stresses can be r r readily shown in applying the statics conditions ∑ = ma ≡ 0 and F r r ∑ M = Iα& ≡ 0 to the solid cube of Fig.e. a its acceleration.1 Stresses and forces: (a) 3-D solid element with positive stresses. and the shearing stresses σ ij =ˆ τ ij are caused by tangential forces F j acting on surfaces A i with normal vector nˆ i (see Fig.3. generated by forces F i perpendicular to surfaces A i .22). • axial force ∼ ∫ σ ii dA . and . The normal and shearing stresses relate to the following forces and moments (Fig.1. τ ij = τ ji (cf. where σ ii are either tensile or compressive stresses. 1. M = r × F is the applied moment or torque. which try to move (shear) adjacent parts of a solid.1b): (a) y σy (b) τyz τyx y Fy∼σ y F Ay σx τxy τzy ∆y σx Fx∼τyx σz τzx τxz x ∆z ∆x z σy Fig. I is the moment of inertia .. • shear force ∼ ∫ τ ij dA . i.1a. the stress tensor is symmetric.5.Chapter I: Elements of Continuum Mechanics 121 σ 11 σ 12 σ 13 σ x τ xy τ xz σ ij = σ 21 σ 22 σ 23 : = τ yx σ y τ yz (1.1..2..1a.5.5. i.plane.5.

where I is the moment of inertia.g.g. M y = ∫ σ x z dA .122 Biofluid Dynamics • moments (or torques). e. etc. Example 1. M z = – ∫ σ x y dA ..22: Simple Bending and Torsion Formulas for Rods of Linearly Elastic Material Consider a 2-D beam subject to pure bending and then a cylinder subject to simple torsion. A . and τ r θ = Tr ⁄ J ..e. e.. Sketches: (a) Bending x σ x(y) MZ MZ y (b) Torsion τ(r) T r τmax τrθ T R x Assumptions: • The normal stress σ x ( y ) is linear. i. and all other stresses in the beam are zero (pure bending). which may bend a beam. or may twist a body.. where J is the polar moment of inertia. Show that σ x = – M z y ⁄ I . ∫ σx dA = 0. T = ∫ ( τxz y – τxy z ) z dA . • The resultant of the internal force of the beam is zero.

Chapter I: Elements of Continuum Mechanics 123 • The torsional loading does not reshape or reorientate the cross- sectional planes of the cylinder.τ max (E1. τ ( r ) = --. τ max 2 τ max T = ---------.J R R A 4 where J = π R ⁄ 2 is the polar moment of inertia of the cylinder’s cross section. Solution: (a) Bending The moments of the internal forces about the beam’s centerline (i.22-3) R Thus.e. I z = bh ⁄ 12 for A = bh and I z = π R ⁄ 4 for 2 A = πR . • σx = –Mz y ⁄ Iz (E1.g. where k is a constant. we have M z = – k ∫ y dA : = – k I z A where I z is the moment of inertia of the cross section A about the z- 3 4 axis. with k = – M z ⁄ I z .∫ r dA : = ---------.22-4) . normal axis) are equal to the applied moment around the z-axis (see Sketch (a)): • M z = – ∫ y σ x dA (E1... we have • τ ≡ τ r θ = Tr ⁄ J (E1. with τ max ⁄ R = τ ⁄ r . Thus.22-2) (b) Torsion The resultant of the shearing stress distribution being equal to the applied torque yields (see Sketch (b)): r • T = r τ dA . e.22-1) A With σ x = ky . Thus.

5. Example 1. Eq. principal directions.(1. They are normal to three principal planes in which all shearing stresses are zero (see Example 1. known as the Von Mises stress (see HWA in Sect. 5.4 to determine the locations of maximum stresses in both aneurysms and stent-grafts.124 Biofluid Dynamics Returning to Eq.5.7).[ ( σ 1 – σ 2 ) + ( σ 2 – σ 3 ) + ( σ 3 – σ 1 ) ] (1. the loading force N generates a tensile stress σ ξ because of the normal N cos α and a shearing stress τ ξη because of the tangential N sin α . with σ x = N ⁄ A . . due to symmetry the stress tensor has six unknown components. Determine the stress in a plane slanted by α = 35° and verify stress symmetry and the existence of a principal plane.23).3) is used in Sect.2) ∅ σ3 The principal stresses can be combined according to Von Mises’ deformational theory in terms of an effective stress σ e .5. These can be further reduced to three principal stresses which act in mutually perpendicular. Thus.1. (1.5.23: Stresses on an Inclined Plane of a Bar under Uniaxial Tension 3 2 Consider a bar ( A = 10 mm ) subject to a tensile load of N=100kN.1) now reads after the coordinate transformation from σ ij to σ ii : σ1 ∅ σ ij = σ2 (1. 1 2 2 2 1⁄2 σ e = ------.1).3) 2 Equation (1. Thus.5. Sketch: Assumption: η ξ • uniform stress y N α distributions x σξ in/on every τξη plane A Solution: With respect to the incline.

= σ x cos α (E1. For example.23-2) A ⁄ cos α In letting the “cutting angle” vary. 180° ) = σ x = σ max when τ ξη = 0 and 1 τ ξη ( α = 45°.5. Specifically.Chapter I: Elements of Continuum Mechanics 125 N cos α 2 • σ ξ = -------------------. σ ξ ( α = 0°. τ ξη ( θ ) = τ ξη ( θ + 90° ) and at α = 0° & 180° σ ξ = σ max ≡ σ x when τ ξη = 0 With the given data.1)) and principal planes or axes [see Eq.= – σ x sin α cos α (E1. 0° ≤ α ≤ 180° . 135° ) = τ max when σ ξ = ± --. and A τ ξη = – σ x sin θ cos θ = – 47 MPa .11 MPa . i.(1.23-1) A ⁄ cos α and N sin α • τ ξη = – -------------------. σ ξ and τ ξη undergo maxima and minima. σ ξ = σ x cos α = 67.2)] actually exist. it is revealed that both symmetry of the stress tensor (see Eq.. (1. N 2 σ x = ---.σ x 2 Graph: τξη σx/2 0 o o α 45 o 90 135 o 180 -σx/2 Comments: When plotting Eq.e.23-2).5.= 100 MPa . (E1.

+ ---------.126 Biofluid Dynamics 1.5)]. where material properties in terms of constitutive equations correlate forces causing stresses in a given solid with body displacements. where body continuity is everywhere preserved. z (1.5.5.dy dx – τ xy dx + f x dxdy = 0  x ∂x   ∂y  or ∂σ y ∂σ x ∂τ xy y σy + ∂y dy --------.+ f x = 0 ∂τ yx ∂x ∂y τ yx + dy ∂y Similarly.. (1.6b) Clearly..2). stress components vary from point to point in a body. i.e.5) For example.6)] (ii) Stress-Strain Relations (see Sect. all external forces and internal stresses have to be in static equilibrium. (1. (1. employing ∑ F x = 0 [see Eq. Now. i.4)] and can show again that τ xy = τ yx .3 Equilibrium Conditions So far we assumed uniformly distributed stresses across each surface caused by an external load.1. .4). using tensor τ xy ∂x notation with the summation Mz dx x convection of repeated indices.5. j + f i = 0 . nevertheless. (iii) Conditions of Compatibility.5.5.1. we take the moments about the origin [see Eq. τ yx we have σy surface dAz ∂σ Fig. 1. 1.5. This brings us to the three basic principles for solving solid mechanics problems: (i) Conditions of Equilibrium [see Eq.2 Differential 2-D solid element with ---------ij + f i = 0 (1.e. we obtain  σ + ∂σ ∂τ xy  ---------x dx dy – σ x dy +  τ xy + --------- . y. considering a thin 2-D solid element of differential area dA z = dxdy (see Fig.5. three additional equations have to be found. More likely. ∂τ xy ∂σ ∂τ xy Fy τ xy + dx --------y.5.+ --------- .5.+ fy = 0 σx dy ∂x ∂σ ∂y ∂x Fx σ x + x dx In 3-D. i. r r • ∑ M = 0 and F=0 ∑ (1. but. consistent with local strain distributions and deformations.5.6a) ∂ xj varying forces and stresses or σ ij.6) contains six unknown stress components. j = x.(1. given the external forces per unit volume f i .5.5.4. Eq.

shape changes occur. translatory displacement due to an axial force. Equation (1. 1. no stress is induced (see Fluid Statics). to BB' = ds B (see Fig. and rotates. In all these cases. translates. say. 1.5. j + f i in Ω ( t ) (1. Clearly.3.7) with a i = duˆ i ⁄ dt being the acceleration of a material point in the domain Ω at time t.4 Deformation Analysis and Stress-Strain Relationships When a force and/or moment is applied to an object. 1.3. beam bending. b 3 ) the distance to the neighboring point A' ( a 1 + da 1. because of body deformation. i. due to stretching and rotation.e. deformation relates directly to stress when body element distortion..5..5. Eq. Deformation analysis. a line element in a 3-D body. B′ a3 b3 • dsB • r∧ r B A′ ∆r = u dsA • r • r2 A r r1 a2 b2 a1 b1 Fig.18)]: ρ a i = σ ij. a 2.5.Chapter I: Elements of Continuum Mechanics 127 Equation (1. during rigid-body motion.7) is subject to appropriate boundary conditions and necessary stress-strain relations as discussed in the next section [cf. a 3 ) and point B at ( b 1.5. with point A at ( a 1. AA' = ds A . As illustrated in Fung (1994) and other texts.9) or (1. a 3 + da 3 ) is after Pythagoras .g.16)].3). rod twisting.(1.5.3 Line element changes during body deformation (after Fung. In contrast. body elements undergo shape changes. Eq. Thus.6) can be readily extended to dynamic structures where in general [cf. and/or simple shearing. a 2 + da 2.5. stretches. 1994). e.(1. b 2. material deformation may occur in form of body elongation/contraction.

Thus. b) r for every point in the body. as 2 ∂a i ∂a j ds A = δ ij da i da j = δ ij ------. a measure of body deformation. b) ∂a j ∂b j so that Eqs. using the Kronecker delta. (1. we can form ∂b i ∂u i ∂a i ∂u i = + δ ij and = δ ij – (1. and the body is at rest or undergoing rigid-body motion.5.5.8a & b) can be expressed. are then ui = bi – ai (1.da j and da i = ------.--------.9 a & b). a 3 ) or a i = a i ( b 1.5. u 2. we obtain  2 E ij da i da j 2 2  ds B – ds A =  or (1.5.14a & b) into (1. Clearly. 1.14a.5.--------.da l da m (1. (1. implying that E ij = ε ij = 0 .5. if the length of each line element 2 2 stays the same.11a. ds B – ds A = 0 .9a.8b) Now. Using the displacement vector (see Eq.5.5.12a) ∂b l ∂b m and 2 ∂b i ∂b j ds B = δ ij db i db j = δ ij ------. inserting (1. x 2. u 3 ) and b j = ( x 1. b 3 ) (1. determination of any continuous body deformation requires known mapping functions b i = b i ( a 1.10) Given the mapping functions (1.db j (1. b 2. b) ∂ Lj ∂ aj ∂ bj ∂ bj where u i = ( u 1.12b) ∂a l ∂a m Forming the difference. x 3 ) .5.12a & b) and neglecting squares amd products of the deriva- . 2 2 2 2 ( ds B ) = ( db 1 ) + ( db 2 ) + ( db 3 ) (1.128 Biofluid Dynamics 2 2 2 2 ( ds A ) = ( da 1 ) + ( da 2 ) + ( da 3 ) (1.13a.8a) Similarly.10)). The components of the displacement vector u (see Fig.5. a 2. b)   2ε ij db i db j where the symmetric Lagrangian E ij is Green's strain tensor and the Eule- rian ε ij is Cauchy's strain tensor.db l db m (1.5. we can write ∂b i ∂a i db i = ------.5.

5. Sect.13b) reduce to (cf. the shear-rate tensor of Sect. (1.1.e. Eq.15a) 2  ∂ x i ∂ x j or in terms of the shearing strain (i.3. for infinitesimal displacements E ij ≈ ε ij because it is immaterial if the derivatives u i. +  (1. (1.e.15a) can be written as 1 1--- ε x --. σ ij = C ijkl ε kl (1.5.16) reduces to (cf. i.σ kk δ ij (1.5. are needed.. when the properties are identical in all directions.3). 1 γ zx --.5.2): 1  ∂u j ∂u i ε ij = --.e. 1. Sect.. we obtain 1+ν ν ε ij = -----------. A stress-strain relationship describes the mechanical property of a material and is therefore a constitutive equation.16) 4 where the tensor of rank four. 1994).2) ∂u i ∂u j γ ij = + (1.17) out and solving for the six ε -components.5. the 3 = 81 elastic constants (or moduli) can be reduced to 36 (Fung. 1.5. C ijkl . For linear elastic materials.15b) ∂ xj ∂ xi Equation (1. Stress-strain relationships. For an iso- tropic elastic solid. i.Chapter I: Elements of Continuum Mechanics 129 tives of the displacement components u i .γ zy εz 2 2 Clearly.e.15c) 2 2 yz 1--. is symmetric.5.5.γ xy γ 2 2 xz [ ε ij ] = 1--.. instead of 36 material values.5. (1. the Cauchy infinitesimal strain terms in Eq.γ yx ε y 1--- γ (1. i.5.3. for example bones for which Hooke's law holds.2): σ ij = λε kk δ ij + 2µε ij (1. only two.18) E E . j are calculated at the position of a point before or after deformation (see Fig.3.1.σ ij – --.. the Lame constants.5. Writing Eq.17) Clearly.

(1.5.130 Biofluid Dynamics εy εz where E is Young's modulus. i. Clearly.20a) subject to aF ( t = 0 ) = bcu ( t = 0 ) (1.= – ---. Humphrey & Delange. For example. ν = – ---. such as biological soft and hard tissues. Bone Tissue (b) Soft Tissue (c) Rubber σ σ σ εM ε ST εR Fig.1.5.4 Stress-strain behavior of different materials. E G = -------------------. relates the load F to the deflection (or displacement) u(t) as (Fig. mimic some vis- coelastic behaviors. 1.5. 2004). the Kelvin model.5.20b) where a to c are constant system parameters. hystere- sis. rubber. Note: ε M « ε ST « ε R (after Humphrey & Delange.e. i. using nonlinear springs k 1 and k 2 as well as an exotic fluid η in the dashpot and assembling several of these models in series. 2004). E. the other materials exhibit nonlinear relationships and hystereses. compared to metals. Clearly. memory.. i. They are nonlinear viscoelastic. . 1.5..e. ν . 1994.e. Mechanical mod- els.5. and rubber. some bio- logical tissues can be approximated (see Fung.4 depicts qualitatively stress-strain graphs for metal. and G are related. a combination of linear springs and viscous dashpots.5): F + aF· = b ( u + cu· ) (1. an improvement over the Maxwell model and Voigt model. soft the Poisson ratio εx εx lateral strain ----------------------------. shape memory alloys (SMAs).5.(1..19) 2(1 + ν) Actual materials.17)]. Since only two indepen- axial strain dent constants are needed for homogeneous isotropic elastic materials [see Eq. they may exhibit creep. and non-Newtonian fluids. For example.. and other complex behaviors. Fig. do not follow Hooke's law. and G ≡ µ is the shear modulus. the loading and unloading curves differ. (a) Metal.

( ε θ + νε r ) 1–ν 1–ν .e.18).τ xy : = --.5.5 Kelvin viscoelasticity model with creep function due to force pulse F → u ( t ) and relaxation function due to sudden deformation u → F ( t ) .τ xy = ------. 1. σ z = 0 .21c) E 1+ν 1 1 ε xy = -----------.τ yz = ------.( ε r + νε θ ) 1–ν 1–ν (1.( ε x + νε y ) σ r = -------------[ σ z – ν ( σ x + σ y ) ] (1. Eq.γ xy (1.21d) E 2G 2 1+ν 1 1 ε yz = -----------.21a) E 1 ε y = --.21e) E 2G 2 1+ν 1 1 ε zx = -----------.21f) E 2G 2 Clearly.γ yz (1. (1.5.[ σ x – ν ( σ y + σ z ) ] (1.21b) E 1 ε z = --.5.5. i..( ε y + νε x ) σ θ = -------------2.5.22a-d) E E σ y = -------------2.5 Simplifications Hooke's law.τ zx : = --.τ yz : = --.Chapter I: Elements of Continuum Mechanics 131 F u η k1 t t F k2 F u u t t (a) Mechanical Model (b) Step Responses Fig.τ zx = ------. and the remaining normal stresses are: Rectangular Coordinates: Polar Coordinates: E E σ x = -------------2. for plane stress analysis. can be rewritten as: 1 ε x = --.γ zx (1.5. 1.[ σ y – ν ( σ x + σ z ) ] (1.

as indicated in Sect.21a) L1 A1 and Ei where E1 x σ y =σz = 0 • Only normal.dx + ∫ -----------.132 Biofluid Dynamics Provided that Young’s modulus E and Poisson’s ratio ν are known. bone with implanted metal segment). Sketch: Assumptions: Concept: • Segmentally Use of constant Ai Eq.24 and 1.. thus. L1 L N N • u total = . Find the total axial displacement u. ∫ ----------.dx . axial L2 force N E2 • Hooke's law holds N Solution: The reduced governing equations read L σx N • ε x = ----- E and ∆u = ∫ ε x dx . equilibrium equations have to be established first which relate forces to stresses. where σ x = ---- A 0 Specifically. Example 1.5.24: Displacement of a Non-uniform Rod under Axial Stress Consider a vertical. 1. each of these equations contain extra unknowns.e. axially loaded rod with suddenly changing cross section and material property as shown (cf. with u ( x = 0 ) = 0 .25.5. (1.4 and illustrated in Examples 1. L = L1 + L2 A1 E1 A2 E2 0 L1 . A2 i.

Sketches: (a) Axisymmetric Tube Geometry (b) Forces on Thin-Walled Tube Element Tube end: t ∆z • Open σz∆Aring r2 σz=0 r r1 • Closed σθAcut θ z pnet σz≠0 pAcross pAprojected (c) Forces on Thick-Walled Tube Element poutside=-p2 σθ|θ+∆θ σr|r+∆r pin=-p1 σr r2 r1 σθ ∆θ Solution: Thin-walled implies that t ⁄ r 1 < 0.1 and radial stresses can be neglected. the difference between blood pressure and outside pressure.-----.Chapter I: Elements of Continuum Mechanics 133 L1 L2 NL 1 L2 A1 E1 • u total = N  -----------. 1 + ----. gen- erates the load.+ ------------ = -----------. The transmural pres- sure.------  A 1 E 1 A 2 E 2 A1 E1  L 1 A 2 E 2 Example 1. while perfect axisymmetry precludes any shear stresses.25: Blood Vessel Applications Some large blood vessels can be approximated as circular thin-walled or thick-walled tubes where σ z is the axial or longitudinal stress and σ θ is called the circumferential or hoop stress. Thus. a 1-D force balance yields (see Sketch (b)): 2 ( σθ t ∆ z ) – p ( 2 r1 ∆ z ) = 0 and the hoop stress is .

= 0 (E1.25-3) dr r In order to solve this ODE.. du ε r ≡ ------ dr and u ε θ = --- r Thus. i. (E1. we employ appropriate stress-strain relations (see Eq.21)). (E1.25-1) t Also. r 1 ≤ tr 2 .e. -----.+ ----------------.5.d) can be written as σ r = -------------2.= 0 r + ∆r 2 ∆θ ∆θ ∂σ r Taking sin ------.5. (1.≈ ------. expressing the strains in terms of the radial displace- ment u r = u . we need an expression for the hoop stress σ θ ( r ) . in case one vessel end is closed and the pressure force exerts a longitudinal stress.and σ r ≈ σ r + -------. Specifically.22c. Alternatively. we obtain in the limit d σr σr – σθ • -------. we have 2 σ z ( 2π r 1 t ) – p ( π r 1 ) = 0 or pr 1 • σ z = -------.∆ r while neglecting 2 2 r + ∆r ∂r higher-order terms.25-2) 2t Both stress equations were originally derived by Laplace nearly 200 years ago. Most blood vessels are thick-walled.134 Biofluid Dynamics pr 1 • σ θ = -------. Eqs.(1. and hence for an open-ended tube ( σ z = 0 ). a radial force balance on the material ele- ment ( ∆ r θ )∆ r yields (Sketch (c)): σr ( r + ∆ r )∆θ∆ z – σ r r ( ∆θ∆ z ) – 2σ θ ∆ r ∆ z sin ∆θ ------.+ ν --- E du u (E1.25-4a)  dr r 1–ν and .

(E1.b).25-5a) dr r dr r or d. i.3) reads 2 d u -- -------. which are best expressed in terms of the radial stresses.e. 1 – ------ - – --------------.25-7a) 2 r2 – r1 2  r 2  2 r2 – r1 2  r  2 2 2 2 2 p1 r1  r2  p2 r2  r1  • σ θ = -------------- . 1 – ------ - (E1. 1 du u • 2 + . ( ur ) = 0 (E1.e. after invoking the boundary conditions (E1. 1 + ------ - (E1. with the solution of Eq.-----.+ ν ------ E u du (E1.17..( 1 + ν ) C 1 – ----------- 2 C2 1–ν r and E 1 – ν- σ θ = -------------2. i.Chapter I: Elements of Continuum Mechanics 135 σ θ = -------------2.25-6a.25-4b)  dr  1–ν r so that Eq. C2 u ( r ) = C 1 r + ------ r Eqs. we obtain Lamé's relationships for r 1 ≤ r ≤ r 2 : 2 2 2 2 p1 r1  r2  p2 r2  r1  • σ r = -------------- . (E 1.( 1 + ν ) C 1 + ----------- 2 C2 1–ν r So.--1.b)  p out = – p 2 at r=r 2 Thus. --.b) read E 1 – ν- σ r = -------------2.25-4a. (E1.19.5b).25-6a.. 1 + ------ - – --------------.---- d- ---.25-5b) dr r dr subject to two boundary conditions.25-7b) 2 r2 – r1 2  r 2  2 r2 – r1 2  r  2 which are plotted below (see Graph).  p in = – p 1 at r=r 1 σr =  (E1. .– ----2 = 0 (E1.

The first two equations for σ x . i.+ ---------.+ F x = 0 (1.5.e. three equations are necessary to calculate σ x . p x = σ x l + τ xy m (1.5.24a) ∂x ∂y ∂σ y ∂τ xy --------. 1.b) can be derived from the compatibility condition for strain components which directly relate to the displacements u and v (see Sect.136 Biofluid Dynamics Graph: (2p1r21-p2r22-p2r21)/(r22-r21) pout r2 σθ radius σr r1 pin (p1r21+p1r22-2p2r22)/(r22-r21 ) stresses Plane stress analysis.. In case the body force Fz=0.5. and τ xy .5. (1. "thin plate problem"). σ y . subject to surface tractions.24b) ∂y ∂x The third one is the equation of compatibility in terms of stress. i.5.23b) where l = cos ( nˆ . x ) and m = cos ( nˆ . then σ z = τ xz = τ yz = 0 (cf. .+ F y = 0 (1.+ --------- (1.5. Thus. and τ xy are obtained from Eq.  ∂2 ∂  1 ∂ Fx ∂ Fy 2  2 + 2  ( σ x + σ y ) = – -----------.25a) ∂x ∂y  1 – ν ∂x ∂y Equations (1. σ y .6): ∂σ x ∂τ xy --------. y ) are the direction cosines for the normal vector nˆ .5.. --------.25a.e.23a) and p y = τ xy l + σ y m (1.+ ---------.7).

. y ) satisfying the continuity condition (1.5.5. While the general mathematical models. in order to understand the principles of biofluid dynamics and to be able to utilize them.5. strain. one has to first comprehend the fundamental case studies as they apply to biomechanics in man and medical devices.25b) ∂y 2 ∂x 2 ∂x∂y When the body forces Fx and Fy are zero. tubular elasticity. 1. with an engineering math refresher (see App. This is because of the inherent complexities of biological systems in man (Sect.1). 2001. subject to (1. including fluid-particle dynamics and fluid-structure interactions.Chapter I: Elements of Continuum Mechanics 137 2 2 2 ∂ εx ∂ εy ∂ γ xy ---------.+ ---------. σ y ≡ ---------2. A). y ) is introduced where 2 2 2 ∂ Φ ∂ Φ ∂ Φ σ x ≡ ---------2. (1. and developed for an accurate simulation of biological systems in man. with polynomial functions of various degrees (see Ugural & Fenster. Silverthorn.---------∂ Φ- 4 + 2 2 + 4 = 0 (1.23a. (1.= ----------- . what should be done in light of such complexities? First of all. complementary to the continuum mechanics approach).e.b) balance. and material properties. as well as feasible solution techniques (i. and cell deformation plus biological response become important as well. Taking arterial blood flow as an example. and Eq. 2002. are in place. Such fluid mechanics case studies.. a stress function Φ ( x. fluid. specific submodels for closure.24a.---------------- 2∂ Φ. the task of evaluating planar stress. and τ xy ≡ – ------------ ∂y ∂x ∂x∂y so that Eqs. in general one has to consider pulsatile 3-D laminar/turbulent flow of a non-Newtonian fluid interacting with deforming cells in irregular branching tubes with viscoelastic composite walls and wall mass transfer. (1.b). 1995). measured.5.5.5. as well as disease etiologies/effects and tissue restructuring.21a) automatically. 1. As the blood vessels decrease in mean diameter. have to be designed. Similar to the stream function Ψ ( x.6 SUMMARY AND OUTLOOK It is transparent that “biofluid dynamics” deals with rather challenging transport phenomena.b) is less challenging.. 2004. typically in terms of partial differential equations. capillary effects on the blood rheology.26) ∂x ∂x ∂y ∂y Traditionally this biharmonic equation (1.23a. a basic understanding of the human physiology is a prerequisite (see Vander.5.25a) yields 4 4 4 ∂--------- Φ.3. and displacement subject to (1. Then. Fox. Now. among others).26) has been approximately solved. include: .

cell or tissue layers. Physical and mathematical insight gained from thoroughly studying the dynamics of such basic fluid flow. cardiac cycle) generate transient velocity fields.4). etc. 1.7) relate to respiratory airways and blood vessels.5 <nearly>. • Truskey et al.9 plus HPAs in Sect.18-1. behind partial occlusions.17 plus HWAs in Sect.7) are important in joint lubrication and solid surface coating of implants. after bifurcations. • Developing flows (see HPAs in Sect.25 as well as Sect. 5. however when help beyond this textbook is needed. • Stress & strain due to rod bending. prime sources include: • Cengel & Cimbala (2006) or White (2003) for basic fluid flow and Schlichting & Gersten (2000) for boundary-layer and pipe flows.g. Basic solid mechanics case studies include: • Forces on vessel walls and implants (see Example 1. growth factors. 1.23 as well as Sect. • Elastic material displacements (see Examples 1. (2002) for transport phenomena. It is of great importance to solve the HWAs of all chapters independently. • Boundary-layer flow of biofluids containing platelets. 1. etc. 1. • Porous-layer flows (see Examples 1. 1. 1. especially mass transfer. 1.20 plus HWAs in Sect. fluid-particle. • Pulsatile flows (see Example 1.4).5 plus HWAs in Sect. • Curved-duct flows (see HWAs in Sect.5).24 and 1.13-1.7) come into play when considering arterial and respiratory diseases. • Particle-suspension flows (see Examples 1.7) are applicable to convection across membranes. . hormones.4. 5. • Moving boundary flows (see Sect. deforming cells. 1..4) are important in coronary arteries.7).6 plus HPAs in Sect. alveoli.2 and 1. and fluid-structure phenomena in the context of biological systems forms the launching pad to tackle more realistic biomedical engineering problems. etc.7) due to sudden inlet pressure changes or an oscillating input pressure (e. 5. etc.3 and Sect.7) occur in entrance regions. 5.22 amd 1.7 and 1.3. where Poiseuille flow representing simple blood flow in straight vessels and planar (joint) lubrication are the prime examples. 1.1. (2004) and Bird et al. twisting.138 Biofluid Dynamics • Fully-developed flows (see Examples 1. 1.4). interacting with vessel walls (see Sects. • Thin-film flows (see Example 1.. and/or stretching (see Examples 1.

unproven statements. specifically. (d) Figure 1. Note: Whenever possible.5! Recall: c/ =ˆ constant Visualization problems: 1. and/or kidneys are indicated as “boxes. as well as associated solution techniques. it is necessary to come up with creative research work accompanied by state-of-the-art literature contributions. follow the four basic solution steps outlined in Fig. i. 1994.Chapter I: Elements of Continuum Mechanics 139 • Kleinstreuer (1997. 1.1. what are the interactions with fluid flow under normal and pathological conditions? (b) In Fig.1. liver. 1. Sketch and describe distinct cell layers lining the arteries and lung airways. In order to advance beyond these resources. 1.1 shows a single cell. Humphrey & Delange (2004). and Fung (1990). as well as other book series. starting selectively with overviews given in Annual Review of Fluid Mechanics and Annual Review of Biomedical Engineering <http:// www. and HWPs that further help to illustrate the physics of: basic fluid flows. 1997) for continuum mechanics and mechanical properties of biological systems.1 depicts flow regimes and modeling approaches as a function of the global Knudsen number. or omitted derivations in the text. 1997 and Clift et al.” Select your favorite organ and show detailed fluid pathways and organ functions. 2003) for momentum transfer and convection heat transfer as well as two-phase flow (see also Crowe et al..AnnualReviews. (2002). and fluid-structure interactions. • Fung (1990. suspension flows. (c) Concerning Fig. typically published in top biomedical engineering and related journals.2 vital organs such as the brain. A few HWPs were inspired by assignments given in Truskey et al... sketch the alveolar region in more detail and describe the local air/blood flow with gas exchange.e.2. assignments that focus on open questions.3. 1. 1977). • Humphrey & Delange (2004) for introductory>. Incorporate the Reynolds number .7 HOMEWORK ASSIGNMENTS This section consists of two groups of homework problems (HWPs). lungs. 1.1 Figure enhancements and new graphs: (a) Fig.

7d). external flows.4b).2 lists real fluid flow characteristics.24) and (1.2. and diffusivity of D1 and .4.7f). (k) Concerning Sect.4. respectively. The solute concentration in the lumen and the outside wall (r=R0) are Ci and C0. (1. (j) Considering turbulent boundary-layer flow. (f) Contrast the Lagrangian vs.1. (1.3). respectively.3. (h) Sketch boundary-layer developments and associated axial velocity profiles for steady laminar incompress- ible flow over a horizontal flat plate and in the entrance region of a circular tube and compare. respectively. open systems with associated variables and parameters helpful in deriving/explain- ing Eq.3.2.2) and Eq.2 Find the effective diffusion coefficient for diffusion through multiple layers of tissue. and outer layer are Ri. and R1-Ri.4. (a) Consider a two-layer model of an artery. Sect. Discuss differences between gases and liquids. depict and state mathematically all possible initial and boundary conditions for both cases. (1. thickness of inner layer. The radius of the cylindrical lumen. sketch coherent structures and indicate random phenomena in a ∆ x -slice for 0 ≤ y ≤ δ . illustrate all terms of Eq. (b) Consider a rectangular laminate consisting of two layers with thickness of L1 and L2.4. (1. (1. and (1. (1. 1. (i) Provide sketches for Eqs. The diffusivity in the inner and outer layers are Di and D0.4.g.. (1. (1.1 problems: 1. Then let the object rotate and add angular momentum terms in your description. Re L Kn = u λ / ν ) and rework the figure.4. R0-R1.2c). the Eulerian approach by sketching closed vs. (l) Develop illustrative graphs explaining the (physical) meaning of Eqs.3.3. (g) Starting with a fixed and non-inertial coordinate system for an accelerating object. (e) Figure 1.27) and explain all terms. depict in two bar-diagrams the relative time scales and length scales of turbulent air as well as water flow past a flat plate. Focusing on internal vs.140 Biofluid Dynamics Re L = u L / ν (e.1).6).3.3.

a is the red cell radius. assess the effect of red cells on the diffusion of O2. Use this result to evaluate the accuracy of quasi-steady analysis. its effective diffusivity (DA). bulk. derive the permeability coefficient (P) of this molecule as a function of L. and γ& is the shear rate. 1. Also determine conditions when the two-layer model behaves as an effective one-layer model. Given a constant species inlet. protein.7 Derive the mass balance equation for steady one- dimensional diffusion through a funnel of varying cross . and platelet. 1.6 The diffusivity of platelets in blood can be enhanced by increasing mixing due to the local fluid motion generated by individual red cell rotation.5 Revisiting Example 1. Following Keller (1971). P is defined as the ratio of solute flux to the concentration difference across the vessel wall.4 Considering one-dimensional diffusion of a nonelectrolyte molecule A in a membrane with thickness L. show that Dp may be written as D p = Ca 2γ& where C is a constant.. determine an expression for the flux in terms of bulk fluid concentrations on either side of the membrane. Thus.8. the effective diffusivity can be the sum of Brownian molecular diffusivity (D) and the “rotation” induced diffusivity (Dp).3 Species boundary-layer flow: Consider flow of a fluid over a flat membrane with permeability Pi. concentration C0. (b) Consider that the side x=0 with concentration c0 is connected to a tissue (chamber) with volume V1. now assume that the concentration at x=h is maintained at zero.e. Assume that the c1=c0 at t=0. and the available volume fraction of A in the membrane (KA). 1.Chapter I: Elements of Continuum Mechanics 141 D2. Using dimensional analysis. i. (a) Assess the time required for the membrane to reach steady state and compare this result with that obtained from quasi-steady analysis. 1. find the concentration c1(t) in the chamber. 1.

1. Using Fig. a solute build-up can be flux observed very near the membrane. (b) Microcirculation of blood in tissue. Take c0=2 mol/litre.e. Sect. Water i.e. The material concentrations at the inlet and outlet are c0 and cL. Find c(x.t) can penerate easily but sol. 1.e. including transport across cell membranes. i.10 In Section 1. three basic questions (i)–(iii) (page 17) are raised and some flow assumptions are listed to help to identify and categorize a fluid dynamics problem. arterial wall or lung.9 Consider a rotated and deformed fluid element at time r t + ∆t (cf. Find the distribution of mass flux along the funnel whose radius varies linearly with distance following the formula  x r ( x) = r0 1 +   L where L is the length of the funnel.2.2. (c) Lymph transport in tissue and vessels.8 Consider a selective semi- permeable membrane of c negligible thickness. Fig. set up the following problems (cf. 1. δ tions. c0 ute is somewhat hindered.1.1 as a guide.g.3.t) as well as the region δ of ele.5) for future solutions: (a) Species diffusion and/or convection in and around cells.2. 1.142 Biofluid Dynamics section (i. in organs such as liver and kidney..1. x { vated solute concentration under steady-state condi.2.. fluid flow and species transport. Derive the vorticity vector ζ and the rr rate-of-deformation tensor ε . respectively. DAB=2 ×10-10 m2/s. (d) Perfusion. Water c(x...3).2 and Table 1.2 problems: 1. A=A(x)). e. Fig. . vwater=30×10-9m/s.

(h) Pulsatile flow in an arterial segment. into a glass. (f) When bringing a spoon near a jet. perforated shunt. (g) A snow storm leaves a cavity in front of a pole or tree and deposits snow behind the vertical cylinder. (b) Under otherwise identical conditions. faucet stream..  l Find the total acceleration in the x-direction: . (i) Very high winds. (c) Certain non-Newtonian fluids when stirred in a cylinder climb up the rotating rod. answering the three questions (i)– (iii) and listing key assumptions with their mathematical consequences. e.12 Consider steady 1-D (plug) flow in a slightly converging nozzle where  x A(x ) = A0 1 +  and u (x = 0 ) = u 0 . (d) Chunks of metal are torn out from ship propellers at high speeds. (e) When pouring a “heavy” beer.. (k) 3-D effects in river bends create unusual (axial) velocity profiles right after the bend and lateral material transport results in shifting riverbeds. (j) Particles of a suspension settle faster in a tilted container than in a vertical one. (i) Airflow in a lung segment. leaky pipe. Note: Set up initial problem solution steps by providing a detailed system sketch. (h) Flapping butterfly wings in China may cause a monsoon in India.g. it gets sucked into the jet. 1.Chapter I: Elements of Continuum Mechanics 143 (e) Laminar flow through a porous tube such as a hollow filter.11 Explain with sketches and equations/proportionalities the following counterintuitive flow phenomena: (a) Keeping the tailgate of a pick-up truck up reduces aerodynamic drag and hence saves gasoline.. a Guinness stout. e. it is easy to blow out a candle but nearly impossible to suck it out. small bubbles float down along the glass wall. (j) A flow system of your choice.g. (g) Wire-coating in a die or sheet-coating with a blade. (f) Droplet formation after jet break-up. tornados. rip certain roofs off houses.g. e. etc. 1.

1.18 Starting with Eq. (1.1. derive Eqs.15 Three-step derivations: (a) Starting with Eqs. derive r the continuity equation.3. derive the Reynolds.1) using an open system. (1.144 Biofluid Dynamics (a) a x (x ) (Euler) (b) a x (t ) (Lagrange) Sect.16 Consider a cylindrical control volume ∆r ⋅ r∆θ ⋅ ∆z . derive Eq.3.3.21). 1. Stokes.1.23).3. and . v = −2 xy . (1. 1. and Womersley numbers. do the Navier-Stokes equations hold for incompressible turbulent flow as well? Explain! r r r r 1.3. does this equation hold for transient flows as well? 1. and (b) Eq.13 Derive: (a) = + (v ⋅ ∇ )v = a total = a local + a convection Dt ∂t r r starting with the total differential dv =ˆ Dv .19 Use a sketch similar to Fig.5).33).3. ∇ ⋅ v = 0 . (1.20). (1. derive Eqs.14 Using “scale analysis” (see Sect.17 Derive the angular momentum equation rr r ∂ r (r × ρvr ) + ∇ ⋅ (rr × ρvr vr ) = ∇ ⋅  rr × T  + rr × ρf B ∂t   1. (1. Strouhal. Peclet.35a).3.1 and a differential mass- balance approach to derive Eq.25). (1.3.22 (a) Given: u = x 2 − y 2 .3. 1.5.36). If ρ =¢. (1.20 Considering direct numerical simulation (DNS).21 Show that ∇ ⋅ v ≠ v ⋅ ∇ and ∇v ≠ v∇ .3. 1.24) and (1. (b) Starting with Eq. 1.3.3 problems: r r Dv ∂v r r r r r 1. 1.

(1.e. to find the velocity profile and flow rate per unit depth.23 Continue Example 1... η=η(y.. y ) and v(x ..25 Solve the Blasius problem [Recall: Steady laminar flat plate flow with u ∞ = u outer=¢.]. ∂ p ⁄ ∂ x = 0 ]: (a) using u = u ∞ sin(ay / δ ) for the integral method. y ) and plot u ( y ) at two x-locations and v(x ) along the B-L edge. and (b) employing η = y / δ using similarity theory (see App.= --. Avoiding initial and end effects.26 Consider draining of a liquid film on a vertical plate. and plot the results. i. and computing load L for an “optimal” case. Obtain u (x .24 Consider a smooth steady liquid film of quasi-constant thickness δ on an inclined wall. The viscosity changes normal to the surface as µ = µ 0 exp[− α ( y / δ )] where α is a constant. find the streamline equations dy v [Recall: Streamline equation to be obtained from -----.20) in cylindrical coordinates and discuss biofluids applications. δ ( x ) . Find the velocity distribution and the flow rate per unit depth. Characterize the flows. 1. . 1. or from the stream function ψ ( x. ∂y ∂x Comment! (b) Derive Eq. 1. 1997).Chapter I: Elements of Continuum Mechanics 145 v r = − A / r . dx u etc.t). 1.3. y ) =¢. show that the film thickness 1/ 2  νy  δ ( y . A and /or page 154 in Kleinstreuer. vθ = − B / r . t ) =    gt  Use a combined variable. where ∂ψ ∂ψ u ≡ ------. etc.and v ≡ – ------.5 by finding p(x ) . plotting p(x ) for different design parameters.

i.29 Consider the history of “drop-spreading” on a horizontal surface. (b) Can the RTT be used to solve a shrinking. Estimate the “radius of influence.e. vaporizing droplet? Set up (the) problem and find a macroscale solution. µ) in a parallel-plate device (graph) where one wall is lined with endothelial cells to demonstrate elongated cell alignment with substantial 3 3 –3 shear flow.146 Biofluid Dynamics 1.” 1. . 50. show that vθ = r02ω 0 / r . calculate the operational conditions. t ) . i.30 A balloon is being filled through inlet area Ain with a fluid velocity v in of density ρ . Find v θ (r . 1. æ<1.28 Repeat Problem 1. (ii) translating with u0=¢. 1. ( ) − σ R1−1 + R 2−1 + τ nn inside = τ nn outside Perform a dimensional analysis of this problem and find the key (dimensionless) system parameters. Hint: Solve for Case (iii) where the other ones are subsets.e. assuming a log-law with “radial” coordinate y = r − r0 .t. and (iii) translating and rotating with u0 and ω0. (b) Suddenly. consider three cases w. where approximately. 100 dyn/cm2. µ water = 10 Νs/m2 and h=250µm as well as measued τwater=10. where b=æa.27 (a) for turbulent fluid flow. Given ρ water = 10 kg/m .31 For Poiseuille-type flow (ρ. dp/dx=¢) in an airway (radius a) with a concentrically-placed cylindrical catheter (radius b).. (a) Under steady-state conditions. 1. (a) Find an expression for the rate-of-change of system mass within the balloon at that instant.. including the basic Poiseuille case (b=0). the cylinder is stopped.27 Consider a long very slender cylinder of radius r0 rotating at ω 0 = ¢ in a large container filled with a liquid of viscosity ν .r.32 Consider steady laminar flow (ρ. the catheter: (i) stationary. 1. µ.

0≤r≤(R-δ) with µm. find suitable fluid property expressions/values and derive the velocity profiles. Employ the Reynolds Transport Theorem to find Rb(t.e. ds. 77-83. balloon angioplasty. etc.33 Red blood cells (RBCs) in tubular flow appear to congregate in the tube’s central core. leaving a (nearly) cell- free plasma layer along the vessel wall. volumetric flow rates.34 Due to tissue/muscle actions or near vessel junctions. dp/dz=¢) and the wall shear stress at any point (xi. 1. i. As a first approximation. (1993) in Annals of Biomedical Engineering. b.37 In the deep lung region. vs). 1. and a homogeneous mixture region. and (b) recover the basic Poiseuille flow solution. metered medical dose delivery. Assuming a constant layer thickness δ characterized by a plasma viscosity µp. and vs is the average fluid velocity in the syringe. . y=h Reh u=u(x. ds is the syringe diameter. and then: r (a) calculate the wall shear stress vector τ w ..y) u=u(y) y e-cells x y=0 0≤ x ≤ Le x>Le Entrance Region Fully-Developed Flow 1. yi) on the elliptic boundary. 1. where Rb(t) is the balloon radius. i.. assume an elliptic cross section and find the volumetric flow rate Q(a.35 Analyze the 3-D parallel plate solution of Usami et al.Chapter I: Elements of Continuum Mechanics 147 associated Reynolds numbers (check if Re max < Re critical ≈ 2000 ) and entrance length (for cell placement in the fully-developed flow region). Vol. blood vessels are noncircular. pp.36 Consider a syringe-balloon system applicable for temporary vessel occlusion. piston velocity.e. 22. vz=vz(y). µ. and shear stress distributions. 1. O2 from the inhaled air diffuses from the alveolus across the alveolar wall into the blood.

1. RBCs. 1.38 Some blood vessels are tapered. where r(x=0)=R1>R0= r(x=L). find u(x. w.3).2 { 10µ AIR m O2 Alveolus ρ.64 6. In turn. µ Air O2 Air CO2 v CO2 Epithelium Endothelium 1. Compare the theoretical results with experimental observa- tions where D1=260×10-6m and D0=78×10-6m Q [cm 3 /s]×10 4 15. Graph: Capillary 5µm ∼0.148 Biofluid Dynamics i. find the oscillatory. filled with a viscous liquid of density ρ and dynamic viscosity µ.e./ RBCs 0. u=u(x..” Of interest are the pressure drop across the stenosis and the centerline velocity as a function of inlet Reynolds number.57 3. the movement of a peripheral plasma layer should be considered. i. moving in parallel-plate like capillaries (see Graph). .. Focusing on the pressure drop for blood flow in one of the capillaries.1. 1. CO2 diffuses into the alveolus and is exhaled (see Fig. Using Example 1.40 Oscillatory blood flow in straight vessels: Consider a circular pipe of radius R.39 Design a laboratory set-up and develop a mathematical model to test “flow through a stenotic tube. r) rather than just u(r) as in Poiseuille flow.58 9. develop dimensionless groups on which ∆p/ρ depends.e.5 as a guide. r). Q and τwall for a slightly tapered tube R1 ≤ r ≤ R0.175 τ wall [dyn/cm 2 ]×10 -3 541 330 225 120 Note: For microcirculation.

Assume that axial pressure gradient term can be decomposed into temporal harmonics of the form ∂p ∞ = ∑ An e inωt and the velocity can be decomposed as ∂z n =0 ∞ u (r .e.4 sec). it is postulated that .2 0. and calculate the volumetric flow rate Q(r. and that the density of blood is 1.8 1 Time (sec) Calculate and plot the fully developed velocity profile as a function of radial position for the following time levels of the waveform: (i) early acceleration (t=0).t) (Womersley. (iii) mid deceleration (t=0. the dynamic viscosity of blood is 0.4 0. where i = − 1 is the imaginary n =0 number.2 sec).41 The following graph is a representation of a carotid artery inlet flow waveform in terms of the centerline velocity.e. z ) . 1.035 g/(cm-sec). i. fully developed velocity profile as a function of radial poistion (r) and time (t). Assume that the tube radius is 0. (1984).5236) [cm/s] + 4 sin(3ωt + 2. The waveform can be represented by a Fourier decomposition: uin (t ) = 40 + 20 sin(ωt ) + 18 sin( 2ωt − 0. u(r. The waveform repeats with a frequency of 1 Hz (pulse rate 60 beats/min).05 g/cm3.4cm. 1.. following Sdougos et al.3 sec). 1955).t).42 Cone-and-plate viscometer: Derive a more accurate solution for v θ (r . t ) = ∑ u n (r )e inωt . and (iv) end deceleration (t=0.618) 100 80 Velocity (cm/sec) 60 Flow Waveform 40 20 0 0 0..Chapter I: Elements of Continuum Mechanics 149 laminar. (ii) peak systole (t = 0.6 0. i.

44 Draw two time-sequence profiles of fluid flow start-up in simple Couette flow for: (a) Newtonian fluids.4 problems: 1. (c) Describe BME applications of this device. α=3°.01g / cm ⋅ s . 2 2 (a) Derive the circumferential velocity and interpret ε. µ = 0. 1.43 A bed of capillaries in a tissue can be approximately represented as a repetitive arrangement of capillaries surrounded by a cylindrical layer of tissue (see graph). (b) Find an expression for the wall shear stress and calculate ω and τzθ (z=0) for R=30cm. . such that the shear stress does not differ by more than 10% between location R1=10cm and R2=20cm. The consumption (or production) of the solute by the cells within the tissue space is the driving force for diffusion of the solute. 1. and (b) power-law fluids. Sect. equation) to describe the relationship between the velocity and pressure gradient for oscillatory flow of a Casson fluid (blood) in a rigid cylindrical tube..e. Assume that the metabolic consumption of the solute is a constant (R0) and the combined resistance of the fluid flowing through the capillary and the permeability of the solute in the capillary wall can be represented by an overall mass transfer coefficient (K0). Calculate the solute concentration distribution in the tissue space. and ρ=1g/cm3.150 Biofluid Dynamics  z    z   z  4   v θ = ωr   + ε  A 2  + B  + L   rα    rα   rα    where α is the plate-cone angle and ε = ρr α ω / µ << 1 . Assume that the blood Tissue flows through the Capillary cylinder capillary with a uniform velocity U and an inlet solute concentration C0. which can be described by an effective diffusivity DT. Justify your graphs with physical/ mathematical arguments.45 Derive an expression (i.1.

48 The suspended leukocytes in a tissue culture medium may flow through a parallel-plate channel and adhere to the endothelial cells which make up the lower surface of the channel. respectively).26)). Assume that the airflow is U θ uniform with a velocity of U and δ<<R. The deposition r0 probability is the ratio of δ and 2r0. the additional drag force due to the presence of surface should be considered besides the Stokes drag force (Eq. respectively. develop the expressions for ratio of λ/λn and τ/τn as a function of δh/R (where λ and τ are the resistance and wall shear stress in the presence of the stenosis. fully developed blood flow in an artery with mild stenosis (δh<<R with δh being the maximum height of the stenosis and R being the radius of the normal artery). Consider the blood to be a power-law fluid. The additional drag force is tabulated by Goldman et al. When the spherical particles move near a solid surface. laminar. find the effects of stenosis on resistance to flow and wall shear stress. 1. where the effect of flow behavior index n should be discussed as well.009 gcm-1s-1.47 Micro-particle deposition in r a 2-D bend: Consider a vp spherical aerosol flowing in Particle }δ a two-dimensional bend with •ٛ θ in radians. length is 7. as shown in the sketch. That is. width is 2cm. Following Young (1968) and considering steady. (a) Calculate the drag force acting on a leukocyte of diameter (d) 12µm that adheres to the endothelium.07gcm-3. . density is 1gcm-3. The density of leukocyte is 1. the fluid viscosity is 0.Chapter I: Elements of Continuum Mechanics 151 1.4. and the average velocity is 2000 µms-1. 1. and λn and τn are the resistance and wall shear stress in the case of no stenosis. Derive an expression to calculate the deposition R probability of the particle. (1.46 Non-Newtonian flow in a stenosed artery: Assume that a stenosis develops in the arterial wall in an axially symmetric manner. Assume that the channel height is 250µm.5cm.

20.49 Calculation of the respiratory mass transfer coefficient. Assume that the concentration distributions of vapor or nanoparticles are known. and the fiber length per unit volume of the porous region is l. 1. 1.152 Biofluid Dynamics (1967) as a function of H/d with H being the height of the particle above the surface. (b) Find the distance the above leukocyte travels through the channel before it adheres to the endothelial cells. Sketch the velocity profiles in the channel and compare to those in a channel with a solid wall. Assume that such a porous medium consists of n parallel tubes per unit surface area (see Sketch in Example 1.18) where each tube has length of h and diameter of d. (1. now consider steady laminar flow in a channel of length L and height 2h with porous walls. hm.42)). Considering an airway segment with length L and diameter D. (a) Find the porosity of the clefts and the hydraulic conductivity of vessel wall using the Carmen-Kozeny theory (cf. Assume that diffusion through each tube is identical and the solute size is much less than d. (b) Find the fluid flux across the capillary wall assuming that the transmural pressure difference is ∆p and the wall thickness is h. d. and D (the diffusion coefficient in the pore). is helpful in quantitatively predicting the regional uptake of inhaled vapor or nanoparticles.. Assume that the area fraction of such porous clefts on the endothelial surface is Ap. Find the mass flow into the wall. deposition efficiency) of vapor or nanoparticles from hm with the assumption of zero wall concentration. 1. h. 1. the channel length is h.52 Similar to Example 1. the average diameter of fiber is df. discuss how to calculate hm. Eq. .51 Tissue as a porous medium surrounding capillaries: The medium between endothelial cells in the capillary wall could be considered as channels with random cylindrical fibers. Also derive an equation to calculate the uptake coefficient (i.50 The capillary wall can be modeled as a porous membrane.4.e. Find the vascular permeability coefficient P as a function of n.

The diffusion coefficient of cholesterol in the tissue is D. Find the steady distribution of fluid velocity. there are axes where all shear stresses vanish..7mm.e.Chapter I: Elements of Continuum Mechanics 153 1. and cholesterol concentration in the sub-endothelial space. indeed. respectively.53 As part of atherogenesis. while they are zero at the outside wall of the sub-endothelial space (i. The stress and strain are assumed to be uniform.55 Consider a 2-D loaded body with planar element ∆ x . ∆ y (see Graph).1. while the nominal stress is forces over the undeformed oriented area Ao. the cholesterol and plasma may pass across the damaged region in the endothelial layer to the sub-endothelial space. r=a) is c0 and p0. Recall: The true stress is a measure of forces acting over a deformed oriented area A. Considering such a soft tissue with original length L=10mm. compare the variations of true and nominal stresses for the current lengths (l) of tissue from 10.001 to 10.. Show that. The cholesterol concentration and the fluid pressure at the wall of the hole (i. pressure. 1.e.5 problems: 1.54 Many soft tissues often conserve the volume when deforming. Graph: Loads y y σy η ση σξη σyx σηξ σxy σx σξ x ξ σyx θ x (a) Basic Material Element (b) Arbitrary Element Rotation . Sect. r → ∞ ). The fluid and cholesterol transport radially and symmetrically about the center of the hole. Assume that the damaged region is a circular hole with radius a and the sub-endothelial space is a semi-infinite porous medium with hydraulic conductivity K.

and 2. (a) Show that the inflation pressure p for a spherical balloon can be given by 2h µ ( ) p = 0 λk −3 − λ− 2 k −3 R0 where λ is the circumferential extension.56 Given the state of stress in Problem 1. 1. (b) Demonstrate the possibility of blow out for a rubber balloon with k =2 and show that a ventricle will not blow out ( k ≈ 18 for myocardium). and h0 and R0 are the initial (zero-pressure) balloon thickness and radius. The calculation of Ex can be found in Humphrey & Delange (2004).55. Compute the error introduced by the linearization in each case and evulate the values of Λ for obtaining the reasonable approximation. 1. 1. σy=-10 MPa. σy=150 kPa. calculate the values of ση for all values of θ from 0° to 90° and plot as a function of θ. and isotropic behaviors where E=16 GPa and ν=0. elastic.4): the component of a 2-D displacement vector are given by ux=(Λ-1)a1 and uy=0. 1.154 Biofluid Dynamics 1. 1.60 The strain energy function of a material can be µ φ= k ( ) λ1k + λk2 + λ3k − 3 where λ1 . λ2 .1. find the values of the principal stresses and denote them on an infinitesimal element with orientation given by σp.1.59 Given σx=20 MPa.5. Calculate and compare the exact (Ex) and the approximate/linearized (εx) strains for Λ=1. find the principal stresses and principal strains with linear. σx=120 kPa. 1. λ3 are the principal extensions and µ and k are material constants.325. (c) Plot the pressure as a function of k and comment. homogeneous.57 Assuming that σx=0.001. and σxy=0 kPa.0.5. and σxy=σyx=5 MPa.58 Material deformation analyis (see Sect. and σxy=-20 MPa. σy=0. 1. respectively.01. .

” Prentice Hall. A.. at an intracranial location. H. (2002) Int. Roberts. assuming that its radius R=2. Sci. where the constant n can be varied. New York. Raff. How much does the aorta need to thicken to restore σ θ back to its original value? 1.. B. (1998) "Computational Fluid Mechanics and Heat Transfer. J. V.25-1) derived for a thin-walled cylinder also provides a reasonable estimate for the mean circumferential stress in a thick-walled tube. R. G.. H. (b) In hypertension. S... Ab. e.Heat & Mass Transfer. calculate the maximum mean pressure when rupture may occur..” Second Edition.. New York. 41: 545. Kao. S. r1 increases) and the wall thins (i. Tannehill. for bone and prosthesis. C.e. (1997) Chem.Chapter I: Elements of Continuum Mechanics 155 1. and Walter. Consider a luminal pressure of pL=np.. K. Upper Saddle River.. J. A. (2002) “Molecular Biology of the Cell. G. 1.. Johnson. D.25 revisited: (a) Show that Eq. P. (1970) J. i.e. A.e.. References Alazmi B. J." McGraw-Hill. Eng. and different but uniform properties..g.5mm. New York. Batchelor. Astarita. 45: 3071-87. NY. and Selamet. Lewis. and Vafai K. and the luminal radius returns to r1 due to smooth muscle contraction and a shear-stress-mediated vasoconstriction. τ decrease) so that σ θ increases dramatically. i. Arpaci. respectively.. εxb=εxp. and Pletcher. A. (E1. (1999) “Introduction to Heat Transfer. Fluid Mech. and a varying mean net blood pressure p≥120 mmHg.61 Example 1.62 Estimate the axial. Anderson. Setting the aneurysm rupture stress at σcritical ≈2MPa. M. Alberts. NY.63 Consider a spherical thin-walled aneurysm.” Garland Publishing. ..e... NJ. Ep. wall thickness h=15µm. Eb and Ap. the aorta distends (i.. Bejan. (1995) “Convection Heat Transfer. 52 (24): 4681-98. partitioned stress in a cylindrical bone with a concentrically implanted prosthesis subjected to different loads N=Nbone+Nprosthesis assuming equal axial strain components. John Wiley & Sons. K.

. 10: 44.. Crowe.. (1988) "Viscous Flows: The Practical Use of Theory. S. W. J.” Wiley-Interscience. Soldaini. Cengel. 120: 446-54. J. MA. J. Fung. New York. Grace.” John Wiley. Chandler. 13 (5): 989.. Boston. C. New York. L.. C. NY. B. Brenner. (1960) Energia Nucl. and Zavattarelli. (2002) “Transport Phenomena. New York. Bird.. M. Drops. and Lightfoot. New York. Lombardi. P. M. E. R. H. W. Boussinesq." Butterworth-Heinemann. Buchanan. C. Ferziger.... Fung. C. New York. J. H. Fung. Flow. Y.. (1993) “Macrotransport Processes. New York.” Academic Press. and Hassager. D. Math. (1998) “Principles of Gas-Solid Flows. E. and Peric.T. R. Clift.” Springer-Verlag. (1978) “Bubbles.” Prentice Hall. G. (1987) “Dynamics of Polymetric Liquids.. R. New York. Drew. New York. FL. NJ. and Growth. Boston. (1987) “Introduction to Modern Statistical Mechanics.” Butterworth-Heinemann. and Weber. Y.” Springer-Verlag. New York. MA. Biomech. NY.. Stewart. M. New York. Boston.. (1978) “Nonlinear Analysis in Chemical Engineering.. Churchill.” Cambridge University Press. B. A. (1994) “A First Course in Continuum Mechanics: For Physical and Biological Engineers and Scientists. (1910) Z. (1997) “Biomechanics: Circulation. L. S. G. NY. NY.” CRC Press. New York. and Tsuji. H.156 Biofluid Dynamics Bird. A. G.” McGraw-Hill..” Oxford University Press. S. 58: 90. M. (1967) AIChE J. NY. A. and Zhu. NY Fung. Y.. A. S. R. Wicks. R. C. Fan. N. O.. D. New York. and Edwards. 23: 46. NY. (1998) “Multiphase Flows with Droplets and Particles. and Cleveland. (1964) AIChE J. J. Armstrong.. C. Acad. Y. W. NY.. D. New York. (1999) “Theory of Multicomponent Fluids. (2006) “Fluid Mechanics: Fundamentals and Applications. Finlayson. C. (1993) “Biomechanics: Mechanical Properties of Living Tissues. and Particles. Stress. T. MA. Silvestri.” Oxford University Press. Englewood Cliffs. R.” Wiley.. NY. C.51. McDonald. E.. H. and Pritchard. (1990) “Biomechanics: Motion. Fox. R.. A. NY.. Dukler. R. A. Bird. Blasius... NY. NY. New York. (2004) “Introduction to Fluid Mechanics. C.M. R. S. Y. Sci. M.” Springer-Verlag. (1994) “Molecular Gas Dynamics and the Direct Simulation of Gas Flows. I. .” Springer. (2002) “Human Physiology. NY. E. NY. Phys. Y. B. Cicchitti. 7: 407-525.” McGraw-Hill. (1998) ASME J. Eng. and Kleinstreuer.. Boca Raton. and Passman. (1877) Mem. A. Sommerfeld. and Cimbala.. (1999) “Computational Methods for Fluid Dynamics.” Springer-Verlag. J. Chen. Fox.A.. and Middleman. Pres.” McGraw-Hill.

D. Soc. Greenberg. Hoffman. J. J.. New York. W. J. and Brenner. NY. A. A." Martinus Nijhoff Publ. S. and Crawford. (1988) “Buoyancy-Induced Flows and Trasport. Irvine. Inc. and Ioanuides. A. New York. Minneapolis. Katz. NY. Keller. 30: 1591. Eng. H. CA. B.” Hemisphere Publishing Corp. (1999) J. B. Lond. Kleinstreuer. and Agarwal.H. (1993) “Convective Heat and Mass Transfer. (1999) J. Los Alamos.” Los Alamos National Labs. NY. Report UMSI 2001/26. (2004) “An Introduction to Biomechanics.. NJ.” Marcel Dekker. Cox. (1998) “Advanced Engineering Mathematics.. (1981) AIAA Paper. H. NY. NM. New York. Kleinstreuer... New York.. H. Hansen.. E. and Ahmadi. and Delange. L. NJ. C... (1987) Int.G. K." MAE Dept. 121: 5-12. NY. Ser.M. A. R.” McGraw-Hill.” Prentice Hall. R. M.. Happel. S. I. H. M. (2000) “Toward a General Theory for Multiphase Turbulence: Part I. (2002) “Principles of Heat Transfer.Liquid Flow.. 22: 637-651..J. Kays... University of Minnesota. Elghobashi. S. B. S. G. G. Y. (2003) “Two-Phase Flow: Theory and Applications. Kaviany. J. R... and Sirignano. and Da Cruz. Joseph. University of California. D. 30: 173-83. I. (1967) Chemical Engineering Science. Gosman. A. and Van der Heyden. and Sammakice. Boston.” Res. Englewood Cliffs.. 81-03223. J. Moy.. and Brenner. Trans. S. D. (2001) “Numerical Methods for Engineers & Scientist.” Springer. (1977) Phil. B. T. C." Cambridge University Press. M. C. C...L. M. T. (1971) Fed. Kleinstreuer. (1997) "Engineering Fluid Dynamics: An Interdisciplinary Systems Approach. Jalurice. 3: 361-5. W. NY.” Taylor & Francis. Humphery. D. Fluids Eng. and Newman J. Sci. 46 (10): 2076-2085. B. 279 (967): 413. Goldman. (1998) "On the Equation for Spherical-Particle Motion: Effect of Reynolds and Acceleration Numbers. Isbin. 30 (6): 739-58.. (1964) “Similarity Analyses of Boundary Value Problems in Engineering.D. Gebhart. (1957) AIChE J. New York. C.. New York. W. A.. (2000) AICHE J. (1999) ASME J. MA. Kashiwa. (1983) "Low Reynolds Number Hydrodynamics with Special Applications to Particulate Media. and Goldsmith.. MN... C.. Karino. Proc. (2001) “Interrogations of Direct Numerical Simulation of Solid . H. B. Upper Saddle River. Gates. M. M. Mahajae.” Wiley-Interscience. R.Chapter I: Elements of Continuum Mechanics 157 Gad-el-Hak. Kim. J. . New York. E. D. He.. C. E. 25 (5): 597- 607. Aerosol Sci..” Prentice Hall. Aerosol Sci.. and Martonen.

Heat & Mass Transfer.. 119 (2): 233-47. Kolmogorov. E.. P. Glowinski. K. New York. D.” Wiley-Interscience. FL. Experimental and Clinical Principles.. G. W. J. E. H. S. 96: 427-443. (1994) "Rheology: Principles. C. L. Patankar. D. R. and O’Rourke. F.S.. Nauk S. D.” CRC Press. and Zaitsev.. W. W.158 Biofluid Dynamics Kleinstreuer. Panton. (2001) Crit. and Cybyk. J.. (1998) Ann.. Z. Li. PA. T. (2002) Int. Rev.. Trans. Michel C.” (3rd edition) Wiley.. N. C. G. Hyun. 30: 403-41. Macosko. F. New York. 45 (25): 4949-4955. Non-Newtonian Fluid Mechanics. C. Nield. M. Ser.. Longest. 82: 1-30. N.. National Academy of Science. Middleman. A. (1993) “Heat Conduction. NY. 26 (4): 863-88. Kuznetsov A.” Springer. Tech. 23 (2): 201-23. and Xu G. FL. ASME.D. N. J. Akad.” Lea and Febiger. C.. New York. Na. FL. Nield. G.S. and Pan. (1992) J. P. NY. A. Boca Raton. Archie. (2005) “Incompressible Flow.. 29 (1): 1-64.. NY. and Gutfinger. T. D. Pnueli.. W. E. D. and Truskey. Sugiyama M. E.. and Ahmadi. Buchanan. Fluids Eng. N. (1942) Trans.. Fiz. Loth. (2000) Int. A.V. Kuwahara F.” CRC Press... S.” Academic Press. New York. Boca Raton. Philadelphia. (1999) “Convection in Porous Media.R. Rev.L. ASME.. and Ahmadi. 6: 56. (1995) J.” CRC Press. Cambridge.. H. Rev.. Singh. A. McAdams. Tech. A.. 44 (22): 4375-4379. NY. and Bejan. Energy Combust. Özisik. (1997) Exp. J. A.” Cambridge University Press. G. R. Michaelides. Heat & Mass Transfer. J.... (1979) “Computational Methods in Engineering Boundary Value Problems. Naterer. E. (2003) “Heat Transfer in Single and Multiphase Systems. C. Physiol. P. Sci. A.” John Wiley. D. (1998) “McDonald’s Blood Flow in Arteries: Theoretical. P.. New York. Aerosol Sci. R. F. J. Measurements and Applications. Polyanin. 26 (3):161-223 Murray. (2000) Prog. Merrill. B." VCH Publ. 16: 209-26.A. (1926) Proc. C. Boca Raton. A. W. 12: 207-14. Eng.. (2001) Int. NY. Li. W. Joseph.. . Polyanin. 26 (9): 1509-24.A. Nichols.. W. and Zaitsev. and Hu. (1995a) “Exact Solutions for Ordinary Differential Equations. (1995b) “Handbook of Solutions for Partial Differential Equations.. (2001) J. A. (1998) “An Introduction to Fluid Dynamics. D. S. (1942) Izv. (1997) J. Patankar. Biomed. Aerosol Sci. (1992) “Fluid Mechanics.. New York. Y. H. V. V.. Multiphase Flow. Fluid Mech. Physiol. NY. and Xiong M. 64: 193. Oh. (1968) J.. UK. Oran. Nakayama A.

(1971) “Advanced Mathematic for Engineering & Scientists. Dewey. S. (1997) “Mechanics of Fluids. (1998a) J. McGraw-Hill. New York. M..P. J. New York." Taylor & Francis. MA.” McGraw-Hill. (2003) “Fluid Mechanics. Pacific Grove. 36 (5): 696-702. P.. Washington. B. NY. ASME. D. (1984) J. 28: 455-57. R. (1999) “Fluid Dynamics. The Netherlands.” Schrum’s Outline Press. Oxford. and Katz. Cambridge. D. (1990) "Multi-Phase Fluid Dynamics. and Luciano. W. La Canada.. Fluids Eng.. F. C. Fluid Mechanics. J. M. New York. and Gersten. B. Ser. C. Boca Raton. FL. F.. Tanner.. G.. (2000) “A Physical Introduction to Fluid Mechanics. Upper Saddle River. Upper Saddle River. R. J. Sabersky. S. D.” McGraw-Hill. Upper Saddle River. (1985) "Engineering Rheology. Bussolari. A.” Prentice Hall. A. (2001) “Human Physiology: The Mechanisms of Body Function. Prandtl. (2000) “Turbulent Flows.C. D. Spalding. Vander.” DCW Industries. Tannehill. Germany.Chapter I: Elements of Continuum Mechanics 159 Pope. White. A. 138: 41. I. Smits. 138: 379-404. . R. R. Ugural. and Wiggert. NJ.. C. J. C. 452... D. M. Schlichting. and Fenster.. (2000) “Boundary-Layer Theory. and Gates. Z. Soo. Boston. H. (1998) “Fluid Flow. Yuan. H. G. MA. F. Rodi. MA. Mech. A. Roache.. E. (1997) "Computational Fluid Mechanics and Heat Transfer. G. L. Warsi.” IAHR. and 1928) “Über Flüssigkeitsbewegung bei sehr kleiner Reibung. Spiegel. M.. D. Appl.” John Wiley. CA. V. H.. R. Hamptmann. P. Sdougos.R. Soc.” McGraw-Hill. Delft. NJ." Science Press. and Gower Technical. S. Acosta. Roache. I. Wilcox.F..F. NY. CA. A. A. (1932) Proc. H. (1991) “Viscous Fluid Flow. E. P. Beijing. White. Leipzig and NACA Memo N.U. S. L.” McGraw-Hill.. San Francisco. UK. G. Sherman. Boston. U. Taylor.” Teubner Verlag. (1961) J. J. UK. Trans." Oxford University Press. (1995) “Advanced Strength and Applied Elasticity. D.C. G.R.” Benjamin Cummings. New York. C. M. DC. and Pletcher. (2004) “Human Physiology: An Integrated Approach.” Prentice Hall.” Springer-Verlag. A. J. (1980) “Turbulence Models and Flow Applications in Hydraulics. 120 (3): 635.” CRC Press. Potter. A. (1998) “Turbulence Modeling for CFD. NY. Anderson. (1969) “One Dimensional Two-Phase Flow.” Cambridge University Press. (2004) “Transport Phenomena in Biological Systems. NY.K. CA. Wallis.” Brooks/ Cole.” Pearson Prentice Hall. Berlin. NJ. Silverthorn.. Truskey. Brookfield. (1998b) AIAA J. (1904.

NY. 1: 3-51.. and Orszag.. O. Z. C.. Physiology. Sci. Industry.” AIP Press. New York. New York. Young. (1998) "Dynamics of Bubbles.. J. Zhang. 90: 248-254. Fluids. Eng. Yakhot. (1955) J. Upper Saddle River. Z. . Springer- Verlag. A. 127:553-563. (2002) Phys. and Rigid Particles." Kluwer Academic Publ. Womersley.160 Biofluid Dynamics Wilkes. M.F. NJ. 14 (2): 862-80. (1999) “Fluid Mechanics for Chemical Engineers. S. (1968) ASME J. and Kleinstreuer. S. J. R. Zapryanov. NY. (1986) J. Drops.” Prentice Hall. D. V. (2000) “The Physics of Pulsatile Flow. Computing. Zamir. and Tobakova.

For the metabolic needs of the cells at th e tissue and or gan level. water. or gan. Once that has been accomplished. solving the c onceptual a nd mathematical complexities of fluid mechanics is most challenging. tiss ue. Charney. ions. Chapter I . the scalar transport phenomena of living s tructures and biomedical devices typically encountered can be readily simulate d and analyzed.1. and active transport (see Pollack. 20 02. or Weiss. a re driven ac ross the cell (or plas ma) membrane (Fig. as well as other more complex trans port mechanisms.2. proteins. re viewed the essentials of tra nsport phenomena. heat transfer and mass transfer in mammalian systems as well as in medical devi ces. and whole-body levels (T ruskey et al. 1996). 2001. in conj unction with Appendices A-C. Ber ger et al . such as ca rrier-mediated. As discussed in Sect. in cluding fluid-structure interaction (FSI). solutes. in order to establish a sufficient background for stu dying biofluid dy namics top ics of the remaining chapters. Datta..1) by passive diffusion. such transport and FSI processes oc cur at the cellular . and provided standard exercise s (see Sect. relying on effective air and blood flow on the macro-level as well as diffusive oxygen and spec ies transfer on the mic ro-level. 1. summarized basic solution tec hniques. At the whole-body le vel. hy perthermia or cryogenics can be employed to de stroy diseased tissue or deac tivate cells and body parts. 1. For decoupled flu id-structure an d/or flu id-particle systems. thermoregulation via body movement and sweating is necessary to achieve a constant (body) tempera ture.. 2 004. Fo r th ermal therapy. Focusing on the human body. etc. oxygen and nutrients are provided by the respiratory and circulatory systems. and osmotic pressure dif ferentials. 1992.1. hydraulic. among others). 19 96. Chapter II BIOFLUID DYNAMICS CONCEPTS Biofluid dy namics encapsulates fluid flow . passive-ion.7). pr ovided that initial/ 161 . 1.

1. 2. mass transfer . and electric charges lead to Heat & Mass Transfer Heat Transfer Modes Mass Transfer Modes C onduction Radiation Diffusion (Thermal Diffusion) (E-M Waves) or Dispersion Fluid Mechanics Convection Convection (forced & free conv.1 Concepts of transfer processes. and regulatory organs (see Ch. 2. as well as relia ble data for material properties. transfer pa rameters.162 Biofluid Dynamics boundary conditions. this is achieved with the controlled pumping action of the heart a nd the viscoelasticity of the blood vessels re acting locally to variations in blood pressure. a healthy bo dy regu lates flu id fl ow. filtering. Chapter II deals with a representative example of biotransport phenomena as w ell as biofluids a spects of the cardiovascular system.) Fig. differences in temperatures. Mass: Phase change.2. i. the local Reynolds number drops rapidly because after the trachea the airw ays bifurcate swiftly into smaller and smaller bra nches.2 inluding DARCY flow Sinks & Sources Heat: Friction. 1. pressures. blo od flo w and air flow a re la minar e xcept in severely stenosed arteries and in the largest airways during rapid breathing. Fig. IV). concentrations. For examp le.e. chemical reactions. sa y. . 34°C at extre mities and 37 °C in the core due to the thermo-regulatory activities of the blood and skin. Non-equilibrium Conditions.. III) as well as that of respiratory . exo/endothermal reactions. and hea t exc hange in such a way that “e xtremes” are avoided. Now.1 Transport Phenomena Through various mechan isms. The body te mperature varies only between. and computer model validation are available. For blood flow. etc. etc. In the lung. Chapter II material functions as a stepping ston e to understand the mechanisms and implications of arterial diseases (C h.

Al ternatively. • Heat and/or mass tran sfer of the human bod y in a polluted environment. Needless to say .e . 4.5). They are empl oyed when track ing a chemical thro ugh t he bo dy from intake to dischar ge. 1 . w hich h as u niform inlet and out let streams carrying species. 1972. Thes e topics a re further elucidated in subsequent sections when applied to basic living structures. beyond diffusional process es in and around c ells. i. plus lung volume which depends on the surrounding tissue. the se basic units can be combined into dif ferent formations. to xins. th e pl eural p ressure p s (s ee Se cts. As mentioned.1. relying on coupled first-order rate equations or greatly simplified transient diffusion/dispersion equations (cf.4) and fluid-structure intera ctions (Sect. transport phenomena were tr aditionally analyzed with a lumped- parameter (or body-compartment) approa ch. nu trients. loop. representing the alveolar region (see Fig. 2. 1.Chapter II: Biofluid Dynamics Concepts 163 Figure 2. a h omogeneous.2.2). or body p art. a scenario which could be modeled as “a cylinde r in cross flow w ith uniform point inhalation” (see Hyun & Kleinstreuer. fundamental to heat and mass transfer.g. Middleman. or dru gs of d ifferent concentrations. for ex ample.1 and 4. CO 2.3).. organ. pe. with an emphasis on its c onversion and deposition in a specific or gan (see Finlay. among others).. • The human respiratory system could be lumped into a “collapsible- tube model” consisting of a tube with axially varying cross-section and material properties as well as a balloon-like end. ion s.. 1. An inlet pressure wave pin(t) changes the lung press ure.g.2 and Sect.3).1 Biofluid-compartment Models Because of the c omplexities invo lved in solving biofluid flow problems and modeling micro-sc ale to macro-scale interact ions (say . is mome ntum transfer (Fig. 1. 2. inert tracers could be injecte d and their transient conce ntrations could be meas ured in orde r to evaluate mass transfer coefficients or volumetric flow rates with an appropriate compartme ntal tran sport model (see Fig. isotropic tissue regi on or capillary bed. and parallel arrangements. membrane → cell → tissue → vessel → organ → body-segments → whole- body).2). O2.” e. e.1. including fluid-particle dynamics (Sect. 20 01. a compartment is a “well- mixed box. a workplace.. Examples include: • Single or multiple body compartmen ts in various seri es. i.e .1. It should be noted that heat convection is not a separate energy tran sfer mo de but actually conduction an d possibl y thermal radiation in a mo ving fluid. 2.1.1 lists key driving forces and concepts associated with heat and mass transfer. Of interest is the time -rate-of change of spec ies concentrations inside the box as well as species deposition and conversion. 2001). . on physiologically based p harmaco- kinetics modeling). In general.

2. R1 i=1 Qin =Q1 =Q2 =…ٛ c1 Q cv Rtotal=R1 +R2 +…ٛ ca i=2 c2 Q2 ∀2 . which makes up 60% of the body weight in adult males.e.164 Biofluid Dynamics • The weight of a human body.1.4 de picts the distribution of the total body flui ds.2 Fig.2 Single. i. basically water. depending on age and sex.1. consisting of: (i) fat. (a) Single Compartment Model with Blood-Tissue Interaction Notation: ca • c a =ˆ artery injected tracer concentration Blood Qout • Qin =ˆ arterial blood stream •ٛVB cB Qin cv • Qout =ˆ venous blood stream VT c T • Qout = Qin = Q . could be subdivided into three compartments.and multiple-compartment models.1. 2. . R2 cv. (ii) extra- cellular mass wi th extra-cellular wa ter (ECW)..3 The lung as an elastic-tube model.1 ∀1 . and (iii) body cell mass with intra-cellular wate r (IC W). R =ˆ flow resistance Species •ٛ mass • ∀ = VB + VT =ˆ total compartment volume Tissue transfer • cv =ˆ venous tracer concentration (see K) • K =ˆ diffusion-convection coefficient (b) Two Compartments in Parallel (c) Compartments in Series –1 –1 –1 ( Q = Q1 + Q 2 . Surrounding Tissue Increase in wall elasticity ps Lumped Lung Resistance Alveolar pe x Mouth •ٛ Region pin (t) C onducting Zone (oral airways + bronchial tree) Fig. Figure 2. R total = R 1 + R 2 ) Qin c0 Q1 cv.

2d) where k ij are the rate constants.1.4 Example of total body-fluid distribution. 2.1.. Thus. qij is the rate of material being transferred from compartment i to compartment j. the material conserva tion law for compartments in series (see Fig.1.1.1) dt j =1 where ci is the amount of material per unit volume accumulating in compartment i. i.Chapter II: Biofluid Dynamics Concepts 165 ICF (25 ltr) Notation: Blood • Plasma (3 ltr) • TBF =ˆ Total body fluids TBF • RBCs (2 ltr) (40 ltr) • ICF =ˆ Intracellular fluids ECF (15 ltr) • ECF =ˆ Extracellular fluids Interstitial Fluid (12 ltr) Fig.1) can be written for each compartment i as: .e.2b) For c being positive and much less than unity.2c) and hence qij = k ij ci (2. f(c) can be linearized to f (c) ≈ a1c (2. A constitutive relationship is necessary to express the transfer rates qij with the principal unknown ci . qij = f (ci ) ( where a Taylor series expansion around point c yields f (c) = a o + a1 c + a 2 c 2 + L (2.1.1. and S i (t ) is the rate of material injected from the system’s exterior (or genera ted due to ch emical re action in s ide compartment i). for linear compartmental models with constant coefficients. (2. In general.2c) can be expressed as: N d ci = ∑ (q ji − qij ) + S i (t ) (2.1. Eq.

166 Biofluid Dynamics N dci = ∑ (k ij c j − k ji ci ) + S i (t ) (2. k 21 = 1.3) dt j =1 Equation (2.3) reads for i=1: dc1 = 0 = − (k 21 + k31 ) c1 + k12 c2 + k13 c 3 (E2. ci (t = 0) . Linearity of Eq.3) has a homogeneous solution of the form n ci (t ) = ∑ Ak e −α k t (2.1-2a-c) .1. k 32 = 2. (2. and S i (t ) .1. (2. 2001.1: S teady-state Solution for Three Compartments in a L oop Configuration Take k12 = 1. while initially c1 = (0) = 1. k 23 = 1. and S i (t ).1-1) dt or with the k-values for i=1. ci (t = 0).3) allows one to compute ci (t ) in steps a nd then sum up all the (independent) solutions (see Hoffman.3: − 5c1 + c2 + 2 c3 = 0 c1 − 3 c 2 + c3 = 0 (E2. c 2 (0) = 0. Example 2. The particular solution f or ci (t ) depends on the information for k ij . and c 3 (0) = 0. k 31 = 4.1. while the eigenfrequencies α k ~ k ij .1. among others). and k 13 = 2. Sketch: Assumptions: Concepts: Q c0 =0 • Well-mixed • Lumped 1 k21 k12 compartments parameter k31 2 k13 • Steady-state approach k32 k23 • No external input • First-order rate 3 nor sinks/sources equation Solution: W ith dci / dt = 0 and S i (t ) = 0 and using the k ij values.2.1.4) k =1 where the coefficients Ak are proportional to k ij . Eq.

2: Single-Compartment Model Returning to the single-compartment model as depicted in Fig. • Lumped u Ca mixed parameter blood (B) reservoirs approach x within single • First-order tissue (T) compartment rate • Constant flow equation Cv (t) rate and system parameters To record the cha nges in concentration of a tra cer C B in blood an d tissue.1-3a) Now. it i s of interest to find an eq uation describing the outlet species concentration Cv(t) for a given inlet tracer concentration Ca. 2. Sketch: Assumptions: Concepts: • Dual well.0. The total s pecies mass per unit volume in the system was at time t = 0: total cinitial ≡ c1 (0) + c 2 (0) + c3 (0) = 1.1-3b) Hence. mass conservation requires that at all times 3 ∑ c (t ) = c i =1 i total initial = 1 .1-1) ) favors influx to compartment for which k13=2.25. (E2.Chapter II: Biofluid Dynamics Concepts 167 4c1 + 2c 2 − 3c3 = 0 This matrix has the solution c1 = c 2 = 0.0 while c2 and c3 were at zero concentration. the steady-state values of ci in the three compartments are: c1 = 0.25. we assume the convective st ream to be the average veloc ity .0 (E2. c2 = 0.0 (E2.2a. The strong transfer out of compa rtment (see Eq. and c3 = 0.5 c3 Thus. Example 2.5 Comments: At ste ady-state an e quilibrium has been reached after initially c1 was equal to 1. a se cond e quation is needed.1.

2-3) Co where C o = C v ( t = 0 ) and A1. We assume that C = C(x. (2002).36) for species transport in blood and tissue. axial diffusion is negligible.t) on ly. C v ≈ C B . we consider a representative capillary with a thin membrane wall. 2. and the tra nsfer coef ficient B ≡ PS/∀ B is constant. we have: ∂C B ∂C B ∂ 2C B m& +u = DB 2 − (E2.2-2b) dt Assuming a well-mixed blood flow zone.e . ∀ B. and K (the dif fusion- convection mas s trans fer coef ficient). and α2 are functions of Q.. (1. is discusse d in Bird et al.3.T a re the blood/tissue volumes.and Q are constant. m& is the species mass flow rate ac ross the blood-tiss ue interfac e. For a somewhat more detailed compartmental transport analysis of the microcirculation region. 2. where P i s the membrane permeabi lity.2-2a) dt and dCT ∀T = Κ (C B − CT ) (E2..3). m& = K (C B − CT ) . A de tailed solution and graphing of Eqs.e.b) is of the form C v (t ) = A1e −α1t + (1 − A1 )e −α 2t (E2. surrounded b y a h omogeneous so ft tis sue of co nstant concentration C T (see Fig.2-1b) ∂t ∂x ∀T Neglecting diffusion and bio chemical reactio n and assu ming t hat K. D B. the simplified equations for CB(t) and CT(t) are then dC B ∀B = −Q (C v − C a ) − Κ (C B − CT ) (E2.2-1a) ∂t ∂x ∂x ∀B and ∂CT ∂ 2 CT m& = DT 2 + −R (E2. and R is the rate of species consumption by metabolism.2-2a. speci fying Eq. Thus. α1. Ca. VB.5).T are the axial blood/tissue diffusion coefficients. (E2.b) are left as an HWA (see Sect. (E2. i. S = dπl is th e surface area of species mass . i. Q/A = u is the area-averaged constant blood velocity.2-a.168 Biofluid Dynamics u = Q / A.1. the solution of Eq. and κ (the partition or solubility co efficient).

Chapter II: Biofluid Dynamics Concepts 169 transfer between the cap illary and tissue.g. and ∀B is the blood volume.8) Q2 Q where k i = (κQ/∀) i .1.5) ∂t ∂x subject to appropriate initial and boundary conditions (cf. (2.5) can be numerically solved using readily available mat h softwa re such as Matlab.1. etc. Femla b.2).7) Using this solution to Eq. tissue volume ∀. 2. i = 1.e.1.2-1a) with m& = 0.3: Single Mass Transfer Compartment Consider a homogene ous tissue re gion p erfused by a ste ady blo od stream. A fluid mass balance yields: Q = Q1 + Q2 (2. 2.2).2.. K ≡ 0 describing tracer depletion in a single compartment.1.t) d cin cout Membrane Tissue Fig. (E2. Mea suring Cv(t). 2. t wo compartments in parallel may form an appropriate model (see Fig.1. Then. B.8) can be employed to find ki or Qi/Q. and CT known. (2. In case a more complex tissue region or an organ (e. Eq. the governing equation can be written as ∂C ∂C +u = B(C − CT ) (2. Fig. Eq. Mathcad.1.1.5 Mass transfer between tissue and embedded capillary of diameter d and length 0≤x≤l. With the “constants” u.1. (2. i.6) and a species mass balance dictates: QC v = Q1 C1 + Q2 C 2 (2. Example 2. a kidney) has to be subdivided. where κ is the partition coefficient.1. Tissue: cT = c/ Qin Qout x c(x. Given the volume tric flow ra te Q.. Eq. FlexPDE. and partition (or solubili ty) coefficient κ for tracer C in the blood stre am.1.7) can be written as Q –k t Q –k t C v ( t ) = ------1 C 1 e 1 + ------2 C 2 e 2 (2. find the outflow concentration C v (t ) for Case (a): constant tracer .

where C v (t = 0) = C 0 . cv equation c(t) being the ∀- averaged tracer concentration Solution: d∀ Fluid flow: − Qin + Qout = : = 0 from which follows dt Qin = Qout = Q (E2.3-4a.e. when C v (t = 0) = 0.3-1) dC Species mass balance: Q(C v − C a ) = −∀ (E2. a nd C ase (b): trac er depletion.3-2a) dt or ∀ dC v C v (t ) = − + Ca (E2. i..170 Biofluid Dynamics inflow concentr ation Ca.3-3) Case (b): With C v (t = 0) = C 0 and C a = 0.3-5) . Q • compartmental transport • ∀ B << ∀ T ≈ ∀ T B Qin (venous Perfused (arterial side) tissue side) • C v = κC .3-2b) κQ dt Case (a): With C v (t = 0) = 0 and k = κQ / ∀. Sketch: Assumptions: Concepts: ca = c/ ∀ = V + V • constant C a .b) dt k dt which has the solution C v (t ) = C 0 e − kt (E2. C v (t ) = C a (1 − e − kt ) (E2. dC 1 dC v QC v = −∀ or C v = − (E2. C • first-order ra te Qout. we can simulate tracer depletion with time.

4 0. i.3-3) (b) Eq. (E2. and pulmonary vein (see Fig. mass transfer Assume: W ell-mixed compartments.3-5) Comments: Clearly. CO 2. volumes ∀. .2). pulmonary artery. Example 2. and reaction rate coefficients kj.e. constant vo lumetric b lood streams Q. 1. Sketch: Lung I Lung II Vena Cava Venous Capillary Capillary Return Qin Q3 Q2 3 4 Atrium Q4 Left Heart Tricuspid LI 5 valve 1 Q1 Ventricle PA PV Q7 2 LC 6 7 Note: blood flow.2 0 0 0 2 4 6 0 2 4 6 kt kt (a) Eq.4 0.g .6 cv(t)/ca cv(t)/c0 0.8 0. (E2. is determined by k=κQ/∀. O2.1.4: Compartmental T ransport M odel for Circulatory System Develop a set of coupled first-orde r rate equations for species mass transfer (e. etc.) betwe en the right heart chambers..8 0.. permeabilities p i. mass transfer areas A. Cv → Co in Case (a) and Cv →0 in Case (b).Chapter II: Biofluid Dynamics Concepts 171 Graphs: 1 1 0.2 0. How rapidly steady-state is reached.6 0. lungs. when comparing Case (a) with Case (b) it is apparent that a system’s initia l/inlet conditions greatly dete rmine the tracer output Cv(t).

4-2a.b) allows us to observe the tracer dist ribution C i(t) in various body compartments (see Graphs). a Runge-Kutta routine can be used to solve the c oupled set of rate equations. subject to some type of initial conditions. For example. . ∞  2λ  δ (t ) = lim  exp(−λ2 t 2 ). where ∫ δ (t )dt = 1 λ→∞ π  0 (E2. a sudden tracer pulse injection C in (t ) in form of a Dirac delta function.4-1a-g) dt dC 2 Pulmonary artery: V2 = Q1C1 − Q2 C 2 dt dC 3 Lung I capillary: V3 = (CQ ) 2 − (CQ ) 3 + ( pA) 3 (C 5 − C 3 ) dt dC 4 Lung II capillary: V4 = (CQ ) 3 − (CQ ) 4 + ( pA) 4 (C 5 − C 4 ) dt Lung interstitial: dC 5 V5 = ( pA) 3 (C 3 − C 5 ) + ( pA) 4 (C 4 − C 5 ) + dt ( pA) 6 (C 6 − C 5 ) dC 6 Lung cellular: V6 = ( pA) 6 (C 5 − C 6 ) − k 6 C 6 dt dC 7 Pulmonary vein: V7 = (CQ ) 4 − (CQ ) 7 dt With a given set of system parameters and coefficients.172 Biofluid Dynamics Approach: Coupled rate equations for multi-compartment system in series Solution: dC1 Right heart: V1 = (CQ ) in − Q1C1 (E2.

V 2. (pA)3=(pA)4=(pA)6=104 ml/s.1. and lymph flow in the lymphatic network.3 i=2 i=5 i=7 ci/cin. V6.0 0. 0 δ ( t ) .3..V5. 50. Cooney (1976). .2.2 Tissue Heat and Mass Transfer Transport phenomena associated with tissue include blood flow with heat transfer in the circulatory system (see Fig. biofluid compartment model s are employed. the species reaches the left heart after about 60 seconds. 50. i. The processes ca n b e ca tegorized in to . as well as dif fusion-dominated mass transfer on the cellular/capillary level when oxygen and nutrients carried by the blood are exchanged with carb on dio xide an d o ther waste p roducts (see Fig.e . (2004).4 0. V4.2 0. Q1 = Q2= Q3= Q4 = Q7 = 100 ml/s. 250. Sect. (2002). 1000. and Chapter IV ). 0 δ ( t ) . and . 200. i.. experience a time lag. Diminished by tissue uptake. Bird et al. C in ( t ) = C in. 1. V 3. for the heart chambers (see Compartment ) and the noted system parameters. As a result of species dispersion and net efflux. and Truskey et al. V 1. 1. 400 ml. the maximum concentrations in ea ch chamber continuous ly reduce a nd s hift. the three graphs depict the transient species conc entrations in Compartments . 2. Comment: Given the inlet condition. Most of the remaining exampl es given in Chapter II rely on the lumped-parameter approac h. C i(0)= 0.1. k6 = 1000ml/s.1 0 0 20 40 60 t [s] Note: The paramete rs used are: C in ( t ) = C in. V7 = 125. Additional case studies us ing the compartmental transport approach may be found in UG hea t/mass transfer textbooks as well as in Middleman (1972). respectively.Chapter II: Biofluid Dynamics Concepts 173 Graphs: 0.e.

(a) Schematic of blood flow in the circulatory system Vein Artery Capillary Bed Arteriole Venule .3..2 and Ch. i. Heat transfer examples: • The blood temperature in the heart’s ventricles and the major arteries remains e ssentially co nstant.1. (2004) and Bird et al.1 . whe n body pa rts are su ddenly being overheated o r su bcooled.3.. capillaries. • Pressure-driven blood filtration occurs from the micro-circulation into the interstitial tissue space wh ere the fl uid (lymph) is collected by the lymphatic network. or flow rate Q = vA .1 and 1. Most species transport process es on the cellular and tissue leve ls are diffusion driv en (see Sect. IV ). communicating with each o ther via the microcirculatory bed’s arterioles.1. called lymphatics. arteries and veins. Sects. indicated by th e local veloci ty vector v . Convection mass transfer examples: • Blood flow takes place in two sets of vessels. i.e.3.6b).5 provide the mathematical framewor k to model and simulate these momentum and m ass transfer pro cesses. (2002). while excess fluid and proteins migrate to the blood via the thoratic duct. or mass flow rate m· = ρ Q . 2.3.2).174 Biofluid Dynamics convection-diffusion mass transfer and convection-conduction heat transfer (see Sects. one has to distinguish between: • species (or tracer) mass transfer expressed as concentration C. Both local blood and tissue temperatures are the same until blood mixes at v arious confluences as well as in the vena cava and the heart’s right atrium. or expressed as bulk flow in terms of average velocity v . e. and venules (see Fig. 1.6a). 2. Clearly. • Muscle tissue motion and lymp hatic pumping cause lymph flow ultimately to the vena cava. Su pplementary references on mass transfer include Truskey et al. Clearly. and r • fluid-mass flow . tissue temperature equi libration occurs a s the b lood p asses thro ugh th e sma ller arteries (see Fig. 3 to 1. 1. 1.

i = H. Tissue heat transfer equation. and S ∀ b T represents a heat generation rate due to metabolic processes. Nevertheless.” It s un derlying assum ptions in clude constant material properties.35) can be rewritten as (Pennes. (1.6 Schematics of b lood fl ow: (a) i n th e ci rculatory system .3. s harply vary ing material property values. and (b) wit h body heat transfer included.9) is known as the “bioheat equation for mammalian tissue. 2. metabolic activities. uniform distribution of blood capillaries in the tissue volume.1.1. and const ant arterial . constant me tabolic heat generation. C (after Datta 2002) Temperature Excessive Heating Tissue temperature TH cooling TA (x) up TA@equilibrium TA=32° C TA (x) down heating Tissue temperature TC Excessive Body Cooling x Aorta to Blood Flow Path Arteries Branches Capillaries (6 -1 mm) to Arterioles Venules (300 -10µm) (≈ 4µm) to Small Veins to Veins Vena Cava (15-750µm) (3-6. T A is the arterial blood tempera ture. • A mathematical description of the thermal exchange in tissue is complicated by two s ets of bloo d vessels i n the millimeter to micrometer range.3 mm) Fig. Eq.Chapter II: Biofluid Dynamics Concepts 175 (b) Schematic of arterial temperature (TA) adjust ments to sudden tissue temperature changes (Ti – TA). etc.1. & is the volumetric flow rate of blood per unit volume of tissu e.9) conduction convection heat source accumul. Equation (2. 1948): ∂T 2 & (T − T ) + ρc = k1∇23 T + ρ b cb ∀ S{T 123∂t 144b24 A 43 (2. geometric irregularities. where T is the tissue te mperature.

dif ferent forms of bioheat equations have to be applied to specific soft-tissue regions as well as small and large blood vessels..9). in order to capture the ef fects of bl ood perfusio n with heat t ransfer th rough t issue. as reviewed by Charney (1992). 2002. cryosurgery or frost bites). Example 2. su rface heating (e. The blood ( ρ . C hen & H olmes (19 80) p ostulated an ad vancement ove r Eq. i. because of the lack of our knowledg e in terms of local blood (or lymph) velocity fields.9) has been used to predict the tissue temperature in space and time due to excessive bo dy surface co oling (e.1. skin burning or hyperthermia).g. ∂T ρ t ct t + ρ b cb u ∇Tt = ∇(k t ∇Tt ) + ∇(k c ∇Tt ) + ∂t 1424 3 144424443 1424 3 Heat convection Thermal diffusion in Heat in pervious tissue tissue and capillaries accumulation & (T − T ) + ρ b cb ∀ q& s 14424 j 43t { Heat transfer from Internal blood vessel j heat source (2..1. material propertie s and boundary co nditions. constant flow rate ∀ . a more realistic description of tissue heat transfer may be gained. Improvements of Penn es’ mod el and special applications have been provided by several researchers. Eq.. (2. c p ) enters the tissue with a & and temperature T A < T1 < T2 .176 Biofluid Dynamics blood temperature.5: Application of the Bioheat Equation Consider blood perfusion of a tissue layer of thickness h where at the fat-tissue interface T = T ( x = 0) = T1 and at the tissue-core interface T = T ( x = h) = T2 .1. Assuming such an idealized tissue volume. and whole body freezing (see Datta. For example. among others).e. with the increase in the number of equation terms and necessary parameters. however. (2. which implies that thermally significant blood vessels are distrib uted throughout the ti ssue.g. with it comes the need for more relevant physiological information and powerful computational tools. Clearly.10) As always.

5-1) k − (c p ∀ dx 2 b A & / k ≡ m2 .9) can be reduced to: d 2T & ) (T − T ) = 0 (E2.46 m=1.6 Θ 1. (2. Eq.47 1.4 0. where T(x) is the local tissu e temperature. Introducing c p ∀ we have to solve an ODE of the form: T " − m 2 (T − T A ) = 0 (E2.8 1 x/h .Chapter II: Biofluid Dynamics Concepts 177 Sketch: Assumptions: Approach: Fat Tissue Core • Steady 1-D • Reduced & uniform flow Bioheat ∀ • Negligible Equation metabolic T1 T(x) T2 rate • Direct • Constant integration q& s & ∀ properties x x=0 x=h Solution: Based on the stated assumptions.19 1.5-3) cosh(mx) − coth(mh) sinh( mx) Graph: 2 m=0.6 0.8 m=0.2 0.5-2) subject to T (0) = T1 and T ( h) = T2 .2 1 0 0. 1993): T − TA T2 − TA sinh( mx) Θ≡ = + T1 − TA T1 − TA sinh( mh) (E2.1.4 1. The analytic solution in terms of the dimensionless temperature is (Özisik.

Assuming the cell to have a very thin spherical membrane. 2400. & =400.6×103J/(Kg⋅K). the rate of thermo-chemica l reactions (i. say . 24000 ml/ ∀ min (i. Comments: Surprisingly. Ex posing cells to a medium of elevated tem perature. Sketch: Assumptions: Approach: • Transient radial • Reduced heat conduction only equation in T0 T∞ٛ • Negligible spherical R membrane thickness coordinates r and heat production • Infinite series q& s h solution rate q& s • Constant properties Solution: With the assumption that T = T(r.1. the spatial temper ature exhibits in this case approximately linear variations.178 Biofluid Dynamics Note: The parameters used for calculation are: h=1cm. Yang. metabolism) in a cell i ncreases with temperature up to a point. (2. T1=34.9) can be reduced to (see Appendix B for spherical coordinates): ∂T α ∂ 2 ∂T = (r ) (E2. Example 2.19. Then. 1.47). c p=3.. 046.e. m=0. the rate of activities declines and in case of excessive hyperthermia the cell dies (cf. 1989). e.g. k=0. 10 º C ≤ Tcell ≤ 45 º C . Fo r example.6-1a) ∂t r 2 ∂r ∂r .e. outside the biokinetic zone. TA=32°C.67 W/(m⋅K).6: Cell Hyperthermia Temperature variations have a pro found effect on cell functions.t) on ly and ign oring an y fluid convection or heat source.5°C. The reasons are that the blood flow rate has only a minor effect on the variations in temperature T(x) under normal physiologi cal conditions when assuming steady 1-D heat transfer across a rather thin tissue slice. t ) su bject to appropriate initial and boundary conditions. T2=37°C . Eq. causes cell-death by hyperthermia. . develop a heat transfer equatio n an d so lve it for T (r . T∞ = 50 o C .

6 H=1000 0. The rate of temperature rise in the cell increases with larger H. = 0.and smaller R.4 0.6-1b-d) ∂T k = h[(T ∞ −T ( R)] ∂r r = R With h / k ≡ H .8 0.6 0. and ∂r r =0 (E2.6-1).the infinite T∞ – T0 series solu tion to (E2.8 1 r/R t (s) Notes: Fo r graph (a). R<1mm) can be negligible.2 H=10 0. based on the separati on-of-variables approach. R=1mm. t = 0) = T0 . is given by Carslaw & Jaeger (1959) or Özisik (1993). where h is the convection heat transfer coefficient and T∞ – T k is the thermal conductivity.6-1a) has to be solved subject to: ∂T T (r .1 t=1 t=2 t=10 0.values.6 Θ(r=0) t=20 Θ 0.43×10−7m2/s. H=100 m-1.6-2b) Graphs: (a) (b) 1 1 0. and α=1. and defining Θ = ----------------.Chapter II: Biofluid Dynamics Concepts 179 where α ≡ k /( ρc p ) is the thermal dif fusivity.4 H=100 0. an d fo r graph (b).2 0. R= 10 µm.2 t=50 t=100 0 0 10 -4 10 -3 10 -2 10 -1 0 10 1 10 2 10 0 0. and α=1.43×10−7m2/s. Comments: The temperature variation inside the sphere with a small radius (say . Equat ion (E2. H ∞ (β n R ) 2 + ( R ⋅ H − 1) 2 ∑ 2 Θ=2 2 2 sin (β n R) sin (rβ n ) ⋅ e −αβ n t r n =1 β n [(β n R ) + H ⋅ R ( R ⋅ H − 1)] (E2.2…) are the roots of βR cot ( βR) + R ⋅ H − 1 = 0 (E2.4 0. .8 t=0.6-2a) where βn (n = 1.

1b) As where. Although the microv essel wall. is a com posite.4 and Table 2.180 Biofluid Dynamics Transvascular transport. actual values for these phenomenological parameters L p and σ s have to be determined experimentally on a case by case bas is.4. ∆p is the static pressure difference.1). which depends o n th e ty pes of solute and membrane structure.4 and 1. end othelium.1. 1. and drug s into the m icrocirculation of ind ividual or gans. and σ s is the osmotic reflection coef ficient. Eq. which feature high porous surface areas. is characte rized by the hydraulic conductivity and per meability coef ficient for specifi c solutes (see Sect. (2.4. nutrients. fo r transvascular transport calculations it is con sidered to be a si ngle p orous membrane. Th e systemic blood circulation delivers fluids. consistin g of the glycocalyx.1. The material transp ort across such capillaries. 1999). Equation (2.11a) where L p is the effective hydraulic conductivity of t he membrane. kg mˆ =ˆ mass flux [ ] cm 2 ⋅ s p c =ˆ hydrostatic capillary blood pressure π c =ˆ osmotic pressure of the plasma proteins pi =ˆ interstitial pressure outside capillary wall π i =ˆ osmotic pressure of proteins in interstial fluid k =ˆ filtration constant [s/cm] indicating the degree of permeability of the capillary wall to water. ∆π is the osmotic pressure dif ference (missing in Darcy’s law). 2.1 .4. .1. Fluid flow across such a membrane is described by a modified form of Starling’s law of filtration si milar to Darc y’s law (se e Se cts.11a) is limited to membranes with macroscopically uniform structures and ideal solutes (see Hu & Weinbaum. first mainly by convection an d th en alm ost exclu sively by di ffusion across microvessel walls. As is the surface area. 0 ≤ σ s ≤ 1 . 1.5): j v = L p As (∆p − σ s ∆π ) (2. an d b asement mem brane. Clearly.11a) can be re written in more practical terms with σ s ≈ 1 as: m& mˆ ≡ = k [( p c − π c ) − ( pi − π i )] (2. Furthe rmore.

because of the curved streamlines in the capillary due to radial fluid flux. and p v . Take µ = 1cP and k = 2. p a . ξ )] (E2. and assum ing steady laminar “fully-developed” flow.Chapter II: Biofluid Dynamics Concepts 181 Filtrated wat er which passes into the surroundi ng tissue is either reabsorbed into the capi llary blood or retu rned to th e b lood v ia th e lymphatic system (see Sect. π i . (E2. solve for u. we obtain (Oka.3).7-1a-d)) ∂r r =0 at the membrane surface u(r = R)= 0 but v (r = R) = k (p . k . v << u . As p c = p c (x) . pi .e . Solving the reduced N-S eq uations an d n eglecting ε − terms 2 . it is suggested to set pc: = p mean = 0.5(p arteriolar + pvenular) ≈ constant. and p. ∆α = p a − α . v ≠0 and the Poiseuille flow solut ion has to be modified. β = R / L . Sketch: Assumptions: Concepts: i • As stated • Reduced N- v S equation r z R u c a =ˆ arterial a v c =ˆ capillary • Starling’s z= 0 z= L i =ˆ interstitial law v =ˆ venous Solution: The appropriate boundary conditions are: ∂u at the centerline = 0 and v(r =0) = 0.5 x 10 −8 s / cm. ε = µk / R . As a simplification.7: Plasma Flow in Filtrating Capillary Introducing . v. and constant parameters µ . α = pi + π c − π i = c/ ∆p = p a − p v . so does m ˆ vary axially.a). bec ause of filtration v ≠ 0 in a st raight microt ube. ξ = z / L .7-2) 4µ L εR ∆p ∆α v= ( ) (2ς − ς 3 ) ( −ξ) (E2. but. ε = 1 x 10 −7 . π c . Specifically. 4. 1981): R 2 ∆p u= ( )(1 − ς 2 )[1 + ε f (ς . i. Example 2. so that with R = 5µm .7-3) µ L ∆p and . ζ = r / R ..

(E2. πR 4  ∆p    1  Q=   1 + ε f  .. ∆α ∆α when ξ < ⇒ outflow (filtration) and when ξ > ⇒ inflow ∆p ∆p (absorption).7-6) 8η  L    3  .7-2) over dA = 2πrdr yields the local volumetric flow rate. ξ ) =  2 ξ −  −( ) + − + β − β ς   β  ∆p  ∆p ∆p 3 2 4  (E2.7-5b) Note when k → 0 and hence ε = 0 .e. i. plasma/water outflow and η ∆p inflow depends on the sign of the radial velocity. i. ξ ) ∆p (E2. ξ ) = − 2  3β  3 2 ∆α  − β (1 − 2ς 2 )   4 ∆p  (E2.182 Biofluid Dynamics p = p a − ξ∆p + εg (ς .e. ξ ) = ( − ξ ) . ξ  (E2.. εR ∆α With v(r = R ) = v(1. the Poi seuille solution is recovered.7-4) where  8  ∆α  2 ∆α 2 ∆α 1 1 2 1 2 2  f (ς . Streamline Graph: z=0 v(z) v(z) z=L r=0 ξ=∆α/∆p Integrating Eq.7-5a) and  3 ∆α 2  ∆α 3 2   ξ − 3 ξ − 1 − 3 − β (1 − 2ς 2 )ξ  8 ∆p  ∆p 4   g (ς .

It consists of tiny. In its simplest form. i. Mid dleman. an d venu les for gas (O2 .1. Blood flow in thes e diste nsible mic rovessels is subject to arterial pressure pulsations and is controlled via contracting/expanding ring-muscle actions of arterioles and precapillary sphincters. Thus Q (x) can be ∆p plotted for three cases as shown in the flow-rate graphs. Clearly .. spleen. among others). we no w consider an entire capillary bed. A similar exchange of gases. both capillaries and tissue “co nnecting” arterioles. Figure 2.2a). Microcirculation and capillary bed mass transfer. when = . Flow-rate Graphs: Q Q Q 0<∆α/∆p<1 ∆α/∆p<0 1<∆α/∆p ξ ξ ξ 0 1 0 1 0 1 The net outflow across the membrane is: 2πR 4 ∆p ∆α 1 ∆α 1 mˆ = Q(ξ = 0) − Q(ξ = 1) = ε ( )( − ) = k As ( − ) µβ 2 L ∆p 2 ∆p 2 ∆α 1 Clearly. membrane-walled blood vessels (5~10µm) which are emb edded in h eterogeneous tissue. 19 89. Q = Qmin when ξ = ..e. Assuming transient convection-diffusion o f a species. nutrients.1. CO2 ) and material exchange (see Fig. mˆ = 0 . 197 2. i. microcirculation is another challenging research area. the capil lary bed can be modeled as an assembly of soft t issue cylinders with capillaries at thei r centers connecting the arterial to the ven ous end (cf. Y ang. In contrast to the single-membrane tra nsvascular-transport appr oach. 2. we have a balanced inflow and ∆p 2 outflow.7 depicts a single cylinder. proteins.Chapter II: Biofluid Dynamics Concepts 183 ∆α Clearly Q = Q ( f ) .e. i n the capillary . and wastes occurs thro ugh the walls of sin usoids (30~40µm) and venules which are contained in the liver. known as the Krogh model. c. and bone marrow .

12b-h) Solution of Eq.1.47)): ∂c ∂c D ∂ ∂c ∂ 2c (2. . and T =ˆ tissue) : ∂C B Capillary: = 0 for t > 0 and C B (t = 0) = C B 0 for all z ∂r r =0 ∂C B ∂C m DB = DM and ∂r r = R1 ∂r r = R1 Membrane/Tissue: ∂C M ∂CT DM = DT ∂r r = R2 ∂r r = R2 C B (r = R1 ) = CM (r = R1 ) and C M (r = R2 ) = CT Tissue: ∂CT =0 ∂r r =R3 (2. M =ˆ membrane. Equation (2.z) or jus t Q/A (t he averaged axial velocity in 0 ≤ r ≤ R1 ).1 . Dr and D z are the radial and axial diffusion coefficients in e ach re gion.1.7 Schematic of cylindrical tissue-capillary model.184 Biofluid Dynamics lumen and transient diffusion in the capillary membranes plus tissue.12a) +v = r (r ) + D z 2 + S c ∂t ∂z r ∂r ∂r ∂z where v is v(r. is left as an HWA (see Sect.12) subject to the appropriate bou ndary conditions.12) ha s to be form ulated for each of the three regions (see Fig.3).7).37a)) or Eq. 2. (2. the simplified governing equation reads (see Eq. (1. The associat ed boundary conditions are (where B =ˆ blood.1. respectively. plus its interpretation.4. and S c represents a species source or sink due to biochemical reactions. 1.4. R3 Tissue Membrane R2 R1 Blood flow r z Fig.1. 2. 2. (1.

8-3a-e) ∂CT ∂CT (r . and β is the Michaelis-Menten constant.1. = 0. ∂r (E2.0) = C B (r .1) = 0.2). z ) CT (r . ∂C B P Capillary: v z = −h [C B ( z ) − CT (r = R1 . Develop the governing equations plus bo undary conditions for the blood and tissue species concen trations when C o is the bl ood/ tissue inlet concentrat ion. z )] (E2.12) have to be written for both the capillary and the tissue.1.0) ∂C (r .3. A where α is the rate of met abolic species consumption. The velocity field is described by Poiseuille flow (see Sect.8-1) ∂z A Dr ∂ ∂CT ∂ 2 CT αCT Tissue: 0= (r ) + Dz − (E2. z )]. Sketch: Assumptions: Approach: r=R2 • Neglect • Analogous Tissue T membrane heat transfer r=R1 Membrane solution (see B r vz(r) z Capillary • Constant Cebeci. (2. T =0 ∂r R2 ∂z ∂z .0) = C 0 . while mass transfer at the tissue/capi llary P interface is S B ~ h ∆C and in the tissue S T = α CT /( β + CT ) .Chapter II: Biofluid Dynamics Concepts 185 Example 2. − Dr = h [C R ( z ) − CT ( R1 . r=0 parameter 2004) Solution: Reduced forms of Eq.8: Convective-diffusive Mass Transfer Assume steady fully-developed blood flow in a capillary and diffusion without any biochemical reaction in the surrounding membrane and tissue.8-2) r ∂r ∂r ∂z 2 β + CT subject to: ∂CT ( R1 .

the mode of lubrication can change from squeeze-film to hydrodynamic to boundary lub rication.5.8c was discussed in Sect.3. such as movable joints – natural or artificial. In general. potentially causing carti lage wear and surface damage. Specifically. potentially leading to bone tissue loss and hence joint failure.1.. e. 2003)..e. has to be overcome: F = fN (2. A major challenge is the right selec tion of cup-and-ball material.0..5. hydrodynamic and elasto- hydrodynamic lubrication. wear. As demonstrated with Example 1.1. Example 1.9).2. the synovial (i. where cartilage deformat ion is considered as well. Joint squeeze-film lubrication.g. Fu rthermore.1.01<f<1.186 Biofluid Dynamics 2. Such a case of hydrodynamic lubrication greatly reduces wear and hence surface damage. Fu rey in Schneck & Bronzino (2003). and replace ment of biological systems. Also important are the m inimum sustained .8a).2. damage. a ba ll-in-a- cup prosthesis. as po inted out by M. For ex ample. J. lubrication. running. the theory of friction. F . while squeeze-film solutions are d iscussed next. e. Wear debris may cause tissue reaction in an attempt to remove the particulate matter. a minimum fluid (or an gled p late) ve locity is required to ge nerate a loa d- carrying pressure field. during body motion. cartilage rubb ing a cartil age) when friction and wear are determined by the surface properties of the solids and the chemical nature of the lubricant rather than its viscosity. and f is the friction coefficient with a typical range of 0. The human synovi al joint is decep tively complicate d beca use of the biomechanical-biochemical interacti ons of articular cartil age with the synovial flu id whic h sep arates lo ad-bearing bo nes. e. friction and wear should b e consid ered as separate phe nomena.g. i.e. For the comm on hip replacement. for load-carrying lubrica tion relying on the wedge-film ef fect to work.g. a tangential resistance. Low friction coefficients can be achieved when the two solid body surfaces stay at all times separated via a suitable lubricant.13) where N is the normal load. as in the case of “boundary” lubrication (e. In order to slide two solids relative to each other . wear is the removal of one material by another due to sliding contact.3 Joint Lubrication Joint lubrication is a subset of bi otribology.1. and surface finish. 1. a frictional force. certa in lubricant components reduce friction while ot hers may enhance cartilage wear (see Furley. Other lubrication modes include squeeze -film. Not surprisi ng... The idealized lubrication model of Fig. theoretical work has been confi ned to joint lubrication. 2. 2. squeeze-film lubrication is the basic mechanism (Fig.1. egg-whitish) fluid in human joint s ( Fig. perfect sphericity. while the analysis of cartilage damage requires experimental contributions. i..

can provide shape design solutions which reduce wear of joint prostheses... Problems may arise when particles are entraine d due to wear. or b oundary lubrication occurs when the fluid film breaks. 2004). Typically. and cera mics. ε = 0. etc.8 Joint lubrication: (a) Schematic of the knee.005 mm for ceramics (Mabuchi et al. a lar ge head may necessitate a greater bone volume loss around the acetabular cup. Modern cup-and. e. where the latter are superior. . While larger sizes apparently reduce wear. A fluid-structure-interaction (FSI) analysis of the coupled squeez e-film and cup-a nd-ball deformation problem.g.ball materials include polyethylene. (b) Squeeze-film model. the ball diameter is 28mm. (a) Joint Schem atic Load Synovial Fluid Articular Χ Χ Χ Χ Χ Χ Χ Cartilage Χ Χ Χ Χ Χ Χ Χ Χ Χ Χ Χ Χ Membrane Χ Χ Χ Χ Χ Χ Χ Χ Χ Χ Χ Ligaments (b) Squeeze-film model (c) Hydrodynamic model Fnormal fixed ρ V0 ρ p p µ µ Ftang ××××××××××××××××× ××××××××××××××××× u0 Fig 2. metal alloys.Chapter II: Biofluid Dynamics Concepts 187 radial clearance and constant properties of the pseudo-synovial fluid in light of realistic load changes du ring walking. The cle arance betwee n he ad and cup surfa ces is in the range of 5 ≤ h ≤ 20 µm.1. labeled elasto-hydrodynamic lubrication. and (c) Hydrodynamic model. where t he low er lim it is related to the surface roughness.

9 Schematic of idealized joint prosthesis with squeeze-film lubrication. π L(t ) = 2πr 2 ∫ 2 p (Θ. B ecause of axisymmetry. Θ ∂h 2 Q = 2π ∫ ( ) r sin Θ dΘ (2. t ) which can be als o estimated from the vol umetric flow ra te.1.3. t ) only..5s we can write:  πt  Lmax sin ( ) for 0 ≤ t < Ts 2Ts L(t ) =   Lmax + Lmin + Lmax − Lmin cos  2π (t − Ts )  for t ≥ Ts  2 2    T  (2.16) 6µ ∂Θ Now. Assuming a sinusoidal load. 2. L(t). Q is also the flow rate during squeezing action. 1.9). with a period of T=1s and start-up time (or half- period) Ts = 0.2).15) 0 Equation (2.188 Biofluid Dynamics L(t) Ball r θ h Cup Fig.17) 0 ∂t .e.6kN.. a Poiseuille-type analysis of fluid flow in a rigid hemispherical ball-and-cup configuration is co nsidered (s ee Fig . A m ore realistic FSI description is given in Mabuchi (2004).15) ca n be employed to find p (Θ.1.1kN and Lmax = 1.e. 2. t ) sin Θ cos Θ dΘ (2. the film pressure is p = p (Θ.1.1.14a. Specifically (see Sect.1. where µ = 15mPa ⋅ s. In order to estimate the film pressure an d crit ical gap height.1.b) Typical measured values are Lmin = 0. πh3 sin Θ ∂p Q=− ( ) (2. resulting from walking. assumi ng Poiseuille fl ow i n the curved conduit. i. i.1.

If both the cup and the ball experience (el astic) deformations. e xpressed as 2 ⋅ ∆s ( Θ. t ) = ho [1 − ε (t ) cos Θ] (2.1. t ) + 2 ∆s (Θ. It can be estimated from the correlation hmin ≈ 2. i.19) Of special interest is the minimum film thickness. Newtonian.. t ) (2.9-1) r ∂r ∂z and the equation of motion to . t ) .” W ith this set of e quations.Chapter II: Biofluid Dynamics Concepts 189 where ∂h/∂t is the “sque eze ve locity. hmin . the elastohydrodynamic film thickness is (see Dowson & Higginson. Example 2. and viscoelastic.1. and ε is the eccentricity with ε (t = 0) = 0. ho < 30µm.1977): htotal = h(Θ.1. Model Schematic: Assumptions: Concepts: L • Transient 2-D • Reduced axisymmetric continuity and incompressible equation of { z p(r) Stokes flow motion h(t) v(z) r w = h& / 2 • Gap height • Insert h << R and appropriate R & rheology h << v L • Normal stresses models for τ rz are negligible Solution: Based on the assumptions.5 (2. respectively. the varying film th ickness u nder the ass umption o f rigid bo undaries ca n b e evaluated as: h(Θ.20) eb2 + ec2 where eb and e c are the surface roughness of ball and cup.9: Joint Lubrication Mode: Squeeze-film Consider squeeze-film lubrication under a constant load L for different fluids. we reduce the continuity equation to 1 ∂ (rv) ∂w + =0 (E2.e. power-law. to check full film lubrication.18) where ho is the initial radial clearance.

9-2) reads: ∂ 2 v dp (E2. (E2.9-4b) −h / 2 r ∂r r dr − h / 2 r dr and h/2 2 v= h ∫ v(r .9-2) across the gap.9-5a) ∂z Thus. Thus. (E2. t ) = τ rz (E2.9-7a) r dr  dr  h 3 dt subject to p(r=R)=0 and dp/dr=0.9-4c) Before integrating Eq. ∂v (i) Newtonian fluid: τ rz ≈ µ (E2.9-4a) h/2 h/2 1 ∂ ( rv ) 1 d ( rv ) 1 d − ∫ dz = − ∫ dz = − ( rv ) h (E2. t )dz 0 (E2.9-1) across the gap − ≤ z ≤ yields z 2 dh 1 d h& ≡ =− (rhv) (E2.9-3) dt r dr where h/2 h h h& h& ∫ −h / 2 dw = w( 2 ) − w( − 2 ) = 2 − ( − 2 ) = h& (E2. τ rz has to be defined. Eq. Integration yields: h 2  dp   z  1 2 v=     −  (E2.9-2) ∂r ∂z h h Integration of Eq.9-6) so that: 1 d  dp  12µ dh r  = (E2.9-5b) µ = ∂z 2 dr subject to v ( z = ± h / 2) = 0. (E2. we have: h/2 1  dp  h2 − h 3 dp hv =   µ  dr  ∫0 (z 2 − 4 ) dz : = 12µ dr (E2. double integration yields: .190 Biofluid Dynamics ∂ ∂ p(r .9-3).9-5c) 2µ  dr   h  4  Evaluating the average radial veloc ity as needed in Eq. (E2. z.

3m p (MPa) v/v 0.9-9) ∂z leading to (see HWA in Sect. .Chapter II: Biofluid Dynamics Concepts 191 3µR 2  r  p= h& ( ) 2 − 1 (E2.5m 0 0 0 0.3m 1 4 L=5e4 Nm-1. R=0.7): 2 π( 2 + 1 / n ) n n L= K sgn (h&) h& h − ( 2 n +1) R n + 3 (E2.4 0.3 0.6 0.9-10) n+3 ∂u (t ' ) t (iii) Linear viscoelastic fluid: τ rz = −∞ ∫ G(t − t ' ) ∂z dt ' (E2.5 0 0.5 -1 6 L=1e5 Nm . R=0.5 2 -1 L=5e4 Nm . 1. (E2.9-11b) Graphs: (a) Profile of radial velocity component (b) Radial pressure distribution 1.1 0.9-7b) h3  R  Now the load can be evaluated as: 3π 4 h& R R L = ∫ pdA = 2π ∫ prdr : = µR (E2.2 0. R=0.4 0.2 0. The pres sure distrib ution (se e Eq.9-11a) leads to: 3πR 4 t L= 2 0 ∫ Q(t − t ' )(h& / h 3 )dt ' (E2.9-8) 0 0 2 h3 ∂u n −1 (ii) Power-law fluid: τ rz = K ( ) .8 1 z/h r/R Note: Case (i) Newtonian fluid Comments: As expected. the radial sque eze-film velocity v(z) is parabolic due to · the approaching disks creating at time t with h disk a pressure gradient dp/dr d riving th e fluid ou twards.

1.4. a species mass transfer equation (see Eq. the wat er-like p lasma. e. As outlined in Sect. because the conduits are rigid and geometrically well defined. Other microvascular beds. deforming red blood cells.1.g. flow in a network of capillaries. (1. the local Reynolds and Womersley numbers drop well below unity and Stokes’ equation apparently holds (see Eq. i. Rosend ahl (2 000).. erythrocyte s or red blood ce lls (RBCs).36)) may be appropriate to desc ribe the cell concentration in a carrier fluid.. especially RBCs and most inhaled drugs a re nonspherical. form a sheet.e.1. multi-layer “rough” walls. complications of microcirculation.22) in Sect. 1. monoc ytes (a member of the leukocyte or white blood cell (WBC) fami ly). an d hi ghly . (1.2). As la rge b lood vessels o r airways bifurcate in a tree-like fashion into arterioles and venules or alveolar ducts with mean diameters in the micrometer range (10 ≤D≤100µm). III )..g. cou pled blood flow with viscoelastic. bio - MEMS (microele ctro-mechanical systems for biofluid flow and biochemical conversion). and platelets are most important in health (i.. known as the Euler -Lagrange approach. Table 2. among others.1 a ssume s pherical o r qu asi-spherical particles. Cell network. 1.3. and • passive as well as active wall mass transfer. 1. e. . While on the first glance. Of the numerous blood cells.1 summar izes characteristics of such particles.g.. and the disk radius.g. 1. when the particle mean diameter is below 1µm. is most suitable . The modeling equations provided in Sect. arise from: • fluid-particle dynamics.. the pulmonary capillary blood vessels. Blood particles.e.3. the conditions R e≤1 and W o<1 imply a major simplification. e. the Knudsen number is relatively high (see Fig.1. and Finley (2001). • fluid-structure interactions.192 Biofluid Dynamics (E2. e. we consider blood as a wh ole to be a m ixture o f a Newton ian fluid . such problems have been add ressed b y Chhabra et al. • abnormal flow-reduced cell functions. in both application areas. i. i. Less complicate d are medical devic es with m icrochannels. (1999).3.1). However.e.2. Alternatively.1) and molecular dynamics models may be appropria te whenever the continuum mechanics assumption is invalid.4 Cell Transport and Microvascular Beds Overview.3. homeostasis) as well as in the genesis of arterial diseases (see Ch. e. biological capillaries and microchannels. ca using the disk ap proach · · speed h disk = h ( L ) .. particle tracking (see Sect.4. For larger cells. 4. more than 109 particles per mm 3 of blood.4. 2.9-7b)) clea rly d epends o n the load .

3 5× 10-6♦ -.964 ♦ -. at shear ra tes below 200s -1.4.†† Volume 88 5.17 460 -. -- (µm3) Size (µm) 7 .5 -. Table 2. blo od exhi bits no n-Newtonian and for γ& ≥ 200s-1 Newtonian behavior (see Fig. (1.1 Properties of Key Blood Constituents. -.967 0. Material R ed Blood Platelets Leukocytes Plasma Whole Cells (all) Blood Density 1. and being highly deformable. with volume concentrations of 40 to 50%.03 1.g..09 1.45× 10-6 1.Chapter II: Biofluid Dynamics Concepts 193 deformable.07.7× 2.-- ρp /ρp † Representative mean value for unactivated platelet †† Platelet dimensions vary considerably * at 37°C (may vary from 0.8 3.014* 0. col lision-induced blo od cell d ispersion generates a “mixing motion” allowing cells and even large proteins to interact with the vessel surfaces. depending on the local shear rate and particle potential.09 1. however. in Eq. liquid-like cel ls either slipp ing past one anot her or aggregating. local concentration. 1. or Carreau-Yasuda (see Kleinstreuer. do not provide any information on cell transp ort.36).9 × 10-3 1. As a resu lt. -- Volume 0.012 to 0. Such a mixing effect requires at the very least a shear-rate dependent dispersion coefficient.2).12 × 10-6 3.46 1. such as the power law.-- α= 0. Fortunately . RBCs.1.0 × 106 -.-.054 (g/cm3) Viscosity -.42 – 0. e. of the form: .0†† 9. Thus.9 83 0.0309** (dyn•s/cm2) Blood cell 5 × 109 3 × 108 7. macroscopic blood rheology models.-- fraction (total) τ p (s) 1.4. Casson.018) ** Newtonian limit at H = 40% ♦ Value for monocytes Unfortunately. 0. 1. 2003).2 × 10-3 -. interact with the other cells and change their motion. and wall deposition. Quemada. -- count (total) (#/cm3) †. monocytes and platelets appear at low concentrations and S tokes nu mbers so that these cells do not af fect the blood flow field and hence their trajec tories ma y be compute d independently.069† 1.

10). 1. An alternative approach. major blood vessel. Well k nown for combating infectio n.3). Then. W BCs) are of great i nterest in cellular fluid mechanics (cf.10 Schematic of capillary wall layers. (1992).1). 200 2).194 Biofluid Dynamics Deff = (a 2γ& ) ⋅ fct ( Ht ) (2. 3. γ& is the shear rate . Realistic structural and biological cell responses to fluid dy namics forces an d the influence o f ce ll be havior. Because each organ. there are numerous geometric vari ations.21) ar e d iscussed by Zydney & Colton (1988). direct numerical simulation. p.4.2.1 . RBC WBC Flow in capillary r vwall Lumen surface v .4. a suspension of cells in plasma. Specific functi onal form s of Eq. Microvascular beds. Phillips et al. Microcirculatio n o f blood d eals with a non - Newtonian fluid of ch anging characteristics flowing in mi cro-vessels of complex structure due to a pre ssure gradient. . a phenomenon called the Fa hraeus-Linquist effect.1. or statistical simulation techniques... monocytes) in turn pl ay a key role in the onset of the arterial disea se “atherosclerosis” (see Sect. en dothelium types. and/or wall pumping action (Fig. η increases rapidly while RBCs.1.4) and decreases in ve ry sma ll ve ssels with diameters less than 300 µm to a minimum of 1cP at D = 8 µm. (2. future-oriented nano-sca le ana lyses. 1.5 cP for γ& < 200s-1) du e to red blood cell (RBC) aggregation (see Fig. . ca n be characterized by a n apparent viscosity. osmotic pressure difference.1. molecular dynamics simulation. in capillaries of D ≤ 8 µm. som e WBCs (e. an d m uscle g roup hav e u nique microv ascular b eds.1.2). 2. In addition to RBC s. Kamm. η. and others..g. and capillary wall characteristics.g.1. especially local cell aggregation. e. which increases at low shear rates (η > 3. is given by Buchanan & Kleins treuer (1 998) and discussed by Kleinstreuer (2003). have to get streamlined in order to squeeze through (see Sect. a nd Ht = 40-50% is the hematocrit. 2.e.21) where a is the ce ll radius.4. surrounded by plasma. on the blood rheology may require novel. c • • • ٛ• ٛ • • •ٛ •ٛ•ٛ•ٛ•ٛ•ٛ•ٛ •ٛ Glycocalyx •ٛ•ٛ•ٛ•ٛ•ٛ•ٛ••ٛ•ٛٛ•ٛ•ٛ••ٛٛ•ٛ•ٛ••ٛٛ••ٛٛ•••ٛٛٛ•ٛ•ٛ••ٛٛ••ٛ•ٛٛ•ٛ•ٛ•ٛ•ٛ•ٛ•ٛ•ٛ•ٛ•ٛ••ٛٛ•ٛ•ٛ•ٛ•ٛ••ٛ•ٛٛ•ٛ•ٛ•ٛ•ٛ••ٛ•ٛ•ٛ•ٛ•ٛ•ٛ•ٛٛ•ٛ••ٛ•ٛٛ••ٛٛ•ٛ •ٛ•ٛ•ٛ••ٛ•ٛٛ•ٛ•ٛ •ٛ•ٛ•ٛ•ٛ•ٛ • ٛ • ٛ ٛ ٛ ٛ ٛ 150nm ≈1 Endothelial ≈ 400nm cells Basement ≈ 5µm membrane Intercell cleft ≈2 20nm Tissue (20 ~ 100 µm) Fig. focusing on shear-induced drift of platelets in non-uniform blood flow (see Sect. Blood. white blood cells. circulating leukocytes (i.

Sketch: Assumptions: Plasma layer • As stated r=r0 • Presence of RBCs r=r0 -δ do not perturb vz(r) Ht (r) r V z (r)-profile z r=0 • Poiseuille flow Hcore Solution: The averaged tube and ou tlet hematocrits are defined as follows: 2 0 r r02 ∫0 Tube hematocrit H t = H t (r ) rdr (E2. p is the fluid sta tic pressure and π the osmotic pressure for the lumen ( l ) and tissue (t).10: Blood Flow in a Capillary: The Fahraeus Effect Considering steady laminar fully-d eveloped blood flow in a small tube. (1994. is smaller than the outflow hematocrit.e. who le b lood separates i nto a cell-free plasma- layer.e. i. Furthermore. H t < H o . the tube hematocrit. r0 ≤ 150 µm .4. nutrients. Ht. 2002) demonstrated that the flow resistance measured in vivo was ab out do uble compared to measurements obtained with glass tubes.10). 1997) (see review by Kamm. δ ≈ 5µm . 1.10-1) . under natural and pathological conditions.2. K is a permeability constant. Pries et al. Differences in terms of a nega tively char ged macrolayer .2 and 1..1.1. 2. the (glycoca lyx) over the endothelium and the surfa ce roughne ss cre ated by th e de forming e -cells may explain the observations. mass transfer of fluid. For example. producing tubular flow results different from those formed in lar ge blood vessels.22) where v w is the fluid volume fl ux (or wall velocity). Ho. and cel ls takes place between the lumen and tissue. As a result. Example 2.Chapter II: Biofluid Dynamics Concepts 195 The walls/surfaces of capillaries are complex as well (see Fig. along the tube wall an d an enriched central core. i..4): v w = K [ pl − pt + (π l − π t )] (2. called the Fahraeus effect. 0 ≤ r ≤ (r0 − δ ) . That could be simply expressed with S tarling’s equation (see Sects.

10-4)  r0  Eqs. 2) become: δ 2 H t = H core  1 – ---- (E2. ------t = ------------------------------2 (E2.e.b) 0 for r 0 −δ < r ≤ r0 and  r  v z (r ) = v max 1 − ( ) 2  (E2. H t < Ho when r 0 < 500 µm.10-1.196 Biofluid Dynamics r0 r0 ∫ Outlet hematocrit H o = H t ( r ) v z (r ) rdr 0 ∫ v (r ) rdr 0 z (E2.10-5)  r 0 and δ 2 δ 2 H o = H core  1 – ---- 2 –  1 – ---- (E2.10-6)  r 0  r 0 H 1 Thus.10-3a. the critical radius when the Fahraeus effect becomes apparent (see Graph).4 -3 10 10-2 10-1 100 δ/r0 .6 0.8 Ht/Ho 0. 10-7) Ho δ 2 –  1 – ----  r 0 Now assuming δ = 5µm.. i.10-2) With  H for 0 ≤ r ≤ r 0 −δ H t (r ) =  core (E2. (E2. Graph: 1 0.

The left part of the heart performs the major pumpi ng with significantly higher pressure levels (i. i. while the tricuspid and pulmonic valves regulate the blood flow to the lungs for ox ygenation (see Fig. For exam ple. Th e heart p umps b lood which transports oxygen-carrying he moglobin to the tissue as well as nutrients. until even deformable RBCs cannot pass vessels with d ≤ 2. 1. specie s ma ss transfer) be tween capillaries and cells a “constancy” of the body’s internal environment. and the four heart valve s.e.7µm. and curvatur es during the cardiac cycle. 8-15mm.. and waste p roducts to dif ferent o rgans. It sh ould be no ted th at these mechanical res istance parameters ch ange wit h time becau se of b lood flow ind uced remodeling. for d < 10µm the hematocrit and apparent viscosity increase again. Sp ecifically.2). Henc e. long-t erm changes in vessel material and geometric properties. The ventricular wall motions create the pumping fu nction of t he h eart. conce rning biomedical engineering. 1. 1. which com municate v ia the tricuspid valve (right) and mitral valve (left).1. However. the cardio vascular system consists of the heart.e.1 Cardiovascular Transport Dynamics The cardiovascular transport system assures via convection in vess els (i.Chapter II: Biofluid Dynamics Concepts 197 Comments: Associated with the Fahraeus effect is a decrease in blood viscosity in capillaries with 30µm < d < 300 µm (i. blo od flow) and di ffusion (i . present re search e fforts are focusing on the aortic and mitral valves as well as the left coronaries..2.2 The Cardiovascular System As mentioned in Sect. the mitral and aortic valves control the flow of oxygenated blood from the heart’s left side via the aorta to the body. thick-wall ed pressu re vessels with g reatly ch anging w all thickness.1). h ormones.e.. the hear t undergoes translation and rotation which significa ntly af fects the mechanics of the coronary arteries. t he Fahraeus-Lin dquist effect). arteries respond biodynamically to significant changes in blood flow conditions. 1. the ventricles are thre e- dimensional. say .e. b lood v essels. the left and right at ria and ventricles (see Fig. in contrast to pipe networks. 2.e. As a result of t he pumping action. 100 to 150 mmHg) than in the right ventricl e. .. hypertension may cause th icker and stiffer arteries and a chronic artery-wall de crease may lead to loc alized lumen constrictions (see Sect. 3.e.2).. Th e heart itself consists of four pumping chambe rs. in cluding p athological as pects. i... 2. which perfuse the muscle wall . Thus. an d th e circulatin g b lood. Specifically. i.e.1. the aortic arte ry (AV) and pulmonic artery (PV) valv es a s we ll as the mitral (MV) and tricuspid (TV) valves.

Ho wever. This loop.1.. similar to cardiac arrest. In summary . with out an y in terruption. the lungs supply oxyg en fro m th e ambient an d d ischarge carbon dioxide. for th eir m etabolism. the major pulmonary arteries have elliptical cross-sections rather . a h ost o f geometric and hemodynamic differences distinguish the two s ystems. across the diaphragm and into the abdominal cavities. 1.2 and 1. From there. ox ygen. 4.and macro-scale levels..e. that return the CO 2-rich blood to the heart’ s right atrium. On the meso. ion s. hea rt muscle. the brain. both the systemic c irculation and th e pulmonary circulation are powere d by the heart. species transport between or gans oc curs via co nvection. Now .3 and 4. pH . and p ermanent b rain damage can occur within four minutes without blood supply. For example. oxygen-rich blood enters the left ventricle and is pumped into the aorta. T issue metabolism generates CO 2 dissolved in blood which drains via the vessels into veins.) of th e interstitial fluid (or lymphs) takes place via diffusion which is here a ve ry effective and rapid process. which ascends. which is done by the or gans (see Chapter IV).g. receiving the same blood flow with equal frequ ency (cf. 1998).e. Pulmonary and systemic circulations. Thus. Nichols & O’Rou rke. 1. is called the pulmonary circulation. The pulmonary arte ries bifurcate and transport blood to the lungs where the CO 2 -O2 gas exchange occurs between the lung capi llaries (CO2) and the air sacs (O2) of the alveolar region. Th ese veins ultimately empty into two large veins. whe re the arterial blood flow through the tissue capillaries and al l species concentrations have t o be adequate. As indicated in Figs. Branches from the aorta supply blood to all the organ systems.4 and Sects. the returning blood from the lungs via the pulmonary veins to the left at rium is oxygen enriched. moving large species quantities because of the microscopic distances and huge surface areas involved (see Sect.198 Biofluid Dynamics known a s ho meostasis. blood with increased CO2-content from tissue metabolism returns to the heart’s right atrium.2. chest) cavity. the superior and inferior vena cavae. blood pat hway from the heart’ s right ve ntricle. uncons ciousness can occur within a few seconds after stoppage of cereb ral flow.1.. i. For example. The kidneys. makes a U-turn and then descends through th e thoracic (i. te mperature.3. os molarity.4). On a micro scale level. and via arterioles/capillaries supply oxygen and nutrients to the tissu e. and skeletal muscles do not recondition blood but need the blood’ s supply of oxy gen an d nu trients. blood enters the right ventricle which pumps it into the pulmonary trunk/ arteries. This implies that the nutr ients in th e b lood stream have to be continuously reconditioned. This loop is called the systemic circulation. t he regulation of species composition (e . In contras t. e tc. which receive over 20 percent of the cardiac output. adju st the electro lyte com position o f the blood an d hen ce co ntrol the electrolyte bal ance in the tissue. For example. through the lungs and back to the heart’s left atrium.

thrombosis. which prevents mixing of the blood. hypertensi on.. on the left side the fibrous . 80 Aorta Arteries Vena Cava Arterioles Veins 60 Capillary Bed 40 (b) Right Ventricle Output 20 •ٛ pmean (l) Arteries Veins 0 l (Vascular path length) Fig. While the mean pulmonary artery pressure is much lower . the fibrous skeleton.2. In contrast.Chapter II: Biofluid Dynamics Concepts 199 than circular ones. 2. hence. 2. CO 2-rich blood is received in the right atrium from the inferior and superior vena cavae while O 2-rich blood arrives in the left atrium from the pu lmonary vein s. 2. (see Chapter III). the right ventricle p umps the blood fo r re oxygenation to the lungs while the left ventricle pumps the O 2-rich blood via the ao rta throughout the bo dy. the tubular elasticity is higher and the wall thickness is thinner than in systemic arteries. and vena cava.e. is called tricuspid valve. etc.1 Mean pressure changes along va scular segments between (a) left ventricle. In general. the heart contains four chambers. ascending aorta. Specifically. In turn. pmean (l) 100 pdiast. in which one-way atrio-ventricular (AV) valves are embedded. t he pu lmonary circulatio n ha s at tracted somewhat less interest because of the ma ny cardiovasc ular diseases occurring in the systemic circulation. The right and left parts of the heart are separated by a muscular wall.2. i.1). two (upper) atria and two (lower) ventricles. i. 1998). 2. and pulmonary veins (after Nichols & O’Rourke. aneurysms.e. The atria an d vent ricles (or pu mps) are separated by a sheet of connective tissue..2 The Heart As mentioned. The AV-valve on the right side has three flaps and. an d (b) right v entricle. p ulmonary arteries. p [mm Hg] (a) Left Ventricle Output psyst. suc h as atherosclerosis. the pulmonary vascul ar resistance is equally lower . about one-sixth of the sys temic on e in bo th cas es (see Fig. called the septum.

75 to 0. The valve is closed due to an axial pressure difference which pushes the valve leaflets toward the center of the aorta. the blood pressure rises and closes the AV valves (see Fig. allowing blood to stream into the empty chambers.2). blood rapidly accelerates through the valve at peak velocity after about 200 msec and then swiftly decelerating (1. Aortic valve. The ca rdiac cycle is the repeating pattern of heart-muscle contraction (i. The pulmonic valve flow is similar to the aortic on e. at the end of systole there is some reverse flow. allowing blood to enter the pulmonary and the systemic circulations.2). generated during ventricular filling.e. bu t no t in the rev erse d irection.1) . the pressure increases and the AV valves open.2.200 Biofluid Dynamics skeleton has two flaps. As with the aortic valve.35-0. the diastolic phase). Clearly . The volumetric flow rate (or cardiac output) Q can be estimated as: Q = ∀ st ⋅ N (2.1. Then. Another set of on e-way valves is locat ed at the origins of the pulmonary artery and the aorta. Cardiac cycle. the pressure-regulated b lood flo ws fro m th e atria to the ve ntricles. t he systolic phase) and relaxation (i. Nevertheless. Du ring dia stole. which takes only 20-30 msec. C ontraction of th e ventricles in systole ejects about two-thi rds of the blood from these chambe rs..1. although with lo wer peak velocities (say. The fluid-structure interactions are further complicated by heart musc le translation and ro tation du ring pumping. filli ng the l eft ventricle completely . When th e ventricles are re laxed and venous blood fills up the atria. ca lled the bicu spid (or mitral) valve.35 m/s for healthy adults). 1. and the basic pressu re drop cause the closing of the mitral valve. While the ventricles contract. While the aortic valve connects the left ventricle with the aorta. These op en during ventri cular p umping action. vortices behind the leaflet. from 0. called the stroke volume ∀st. The tri cuspid valve flow exhibits a simi lar velocity profile. The fast closing action is assisted by vortices which develop behind the valve leaflets. When the a ortic valve opens. 1. t he atrium contracts causing a second flow acc eleration through the partia lly cl osed mitral valve. b lood flows from the atrium to th e v entricle because of t he positive pressure differential. the mitral valve separates the le ft ventricle from the left atrium as it receives oxygenated blood from the lungs via the pulmonary veins (see Fig.15 m/s).e. The pulsating heart motion moves t he valv e axially and reduces its ef fective opening which helps in the swift closing by the valve leaflets.. and they close during ventricular re laxation to prevent the backflow of blood into the chambers. which move the aortic valve and change its size. Mitral valve. although at lower speeds.

is measured in [ltrs/beat]..2). t he pressu res g enerated by t he righ t heart pu mp are considerably lower because the l ungs produce much less flow res istance than the sys temic organs (se e Fig. is called the Frank-Starling law of the heart.2a).e. while it is 24/8 mmHg for the pulmonary artery. while elevated arterial press ure te nds to dec rease ∀st by increasing the afterload on cardiac muscle fibers. the aortic valve suddenly closes. Figures 2. wit h a focus on the left heart ’s pumping action ( Fig. The contractility of the heart muscl e and high vent ricular filling pressure (i. in [ltr/min].04s. while paorta ≈ 80 mmHg. Q.. the following events occur: • As the ventricles start to contract.e.. a new cardiac cycle begins.2a..e.g. and the cardiac output. 2. and a “notch” in the aortic pres sure wave occurs du e t o the elast icity of the valve leaflets and because some blood flows backward to fill t he aortic valve leaflets as they close. 1.. where the peak systolic pressure is pmax ≈ 120 mmHg in both ventricle and aorta. and valve actions are all the same. • When pressure in th e ventricles dip s below that i n the atria. a typical syst olic/ diastolic pressure ratio for the aorta is 120/80 mmHg. ∆pLV = 40 mmHg. • When the pressure in the left ventricle exceeds that in the aorta . In numbers. the phenomenon that the heart contracts more forcefully when it is filled to a greater degree during diastole. 1. the heart rate N in [beats/min]. The difference between these maximum and minimum values is the pulse pressure. the left and right pumps) act synchronously. • Immediately after that. Specifically. it is the phase of isovolumetric contraction. • As th e ventricular con traction diminishes. . both sides of the heart (i. the stroke volumes. systolic/diastolic periods. Ultimately pventricle = 0 mmHg. However .2. the semilunar valves op en an d b lood ejectio n beg ins. e. the pressure falls below the aortic pressure.. The “preload” effect. Because the enti re heart is contro lled by a single el ectrical impulse. suppor ted by atrial contraction which pushes the rest of the blood into the ventricle. ∀st. and the left ventricle begins to fill up from the l eft atrium where the pressure initially drops. the atrium an d ve ntrical pre ssures slowly ris e a gain beca use the chambers are pass ively filled with b lood fro m the v eins. T hen. Later. W ith all valves closed for about 0.e. i. the blood volume stays the same.2.e. i. a phase of rapid filling takes place.Chapter II: Biofluid Dynamics Concepts 201 where the stroke volume. the rising chamber pressure causes the AV valves to close. preload) both increase ∀st.b depict the left an d righ t pressure wavefo rms as well as distributions of the left ventricular volume and aortic flow rate. du ring the diastolic ph ase the mitral valve opens when the ventricular pressure falls below the at rial pressure. i. The ventricular muscle then relaxes and all valv es are again shut du ring the shor t isovolumetric relaxation phase.

2.2b Cardiac cycle of the right heart.202 Biofluid Dynamics diastole systole constant-volume phases p left [ mm Hg] ∀left ventricle [m ltr] Qaorta [ltr/min] 1.2.2a Cardiac cycle of the left heart. .0 [s] Fig. pright [mm Hg] [s] Fig 2.2.

inviscid flow passes through the aortic valve. Note. Sketch: (a) Aortic Valve Aorta Pulmonary artery Pulmonary vein Bicuspid valve Left ventricle Aortic valve (b) Math Model θ C.V. C. (2004). 0 ≤ x ≤ l and r ≤ y ≤ a . a recent review of the fluid mechanics of heart valves has been provided by Yoganathan et al.Chapter II: Biofluid Dynamics Concepts 203 Example 2.S.11: Heart Valve Dynamics (Adapted from Bellhouse & Talbot. Develop an equation t o describe the pre ssure difference between the cusp root and tip as a function of geometry parameters as well as valve open/closure conditions. 1969) Assume th at th e ao rtic valve cusps bo und a co ne-shaped mo ving surface (see sketch) when pulsatile.

11-5b) where p A is the average pressure on the aortic-side of the cusps. (E2. thus the Bernoulli equation reads: . t ) = u1 − 1 +  (E2.11-4) y2 3y 2  a  Conservation of momentum implies ∂ r Fx = ∫ ∂t ∀ ρud∀ + ∫ ρu ( v ⋅ dS ) S (E2. dθ 1  dr  Given the angular velocity of the cusps Ω= =   and dt l  dt  β ≡ r / a .1 1-2) dt where ∀ is the volume contained by S. we have over C.1 1-1) i.11-3) is valid at any cross-section x with a radius of y. S : r r ∫S ρ (v ⋅ dS ) = 0 (E2. so that a2 aΩx 2  2 y  u ( x. Eq. ∀ = 13 πl (a 2 + ar + r 2 ) .11-5a) Here the axial force is Fx = πa 2 p1 − πr 2 pt − π (a 2 − r 2 ) p A (E2.. The conical tu be a xis is a lways a s treamline because o f s ymmetry. i.S..e. (E2.e.204 Biofluid Dynamics Assumptions: Concepts: • Inviscid flow • Conservations of mass and • The velocity variations across momentum planes perpendicular to the axis • Unsteady Bernoulli equation are negligible Solution: Considering the mass conservation for the conical control volume ∀. d∀ πa 2u1 − πr 2ut = (E2. Eq.11-2) can be rewritten as 2 Ωa  l  u1 ut = 2 −   (1 + 2β ) ( E2.11-3) β 3β 2  a  In fact.

The time-averaged a rterial press ure. (E2.11-6) yields p1 − pt du dΩ u2 = C1 1 + C2 a + C3 1 + C4 aΩ 2 + C5 Ωu1 ρa dt dt a (E2.11-3) and (E2. An e xample was giv en in Bellhouse & T albot (1969). (2. Arterial pressure.1)) as p A = QRtotal (2.11-4) in Eq. Their results showed that the velocity at the cusp tip was relatively constant during closure and the flow pattern established in the early stages of systole would persist until f orward flow ceased when the valve began to close.11-6) 0 ∂t Substituting Eq.11-7) if the time-depende nt velocity through the valve u 1 ( t ) and closure rate of the va lve a re kn own. (E2. (2.11-5) and Eq.2. p A . (E2. a ctually ∆p = p A − p v (where p v = 0 ).2a) .2.Chapter II: Biofluid Dynamics Concepts 205 l ∂u ( p1 + 12 ρu12 ) − ( pt + 12 ρut2 ) = ρ ∫ dx (E2.11-7) where 1l C1 =   β a 3 1 l C2 = −   3β  a  1 1  C3 =  4 − 1 2β  4 (1 + 4β + 10β 2 )  l  C4 =   18β 4 a 2 (1 + 2β + 3β 2 )  l  C5 = −   3β 4 a Comments: The velocity at the cusp tip u t and the pressure difference during the decele ration phas e ca n be c alculated with Eqs. It relates to the cardiac output (Eq. is the drivin g force for blood flowing through the systemic organs.

i . employing wall stress equations postulated by Laplace and Carlson (see “Concepts”). systolic minus diastolic pressure. Normal mean values for a person at rest are p A = 100 mmHg . . A reasonable estimate of p A is given by: 1 p A = pD + ( pS − pD ) (2.2.. arterial blood pre ssure is co ntinuously monitored by various sensors in the body.12: Pressure-Stress Relation in Left Ventricle Assuming that the left ventricle is a thin-walled. sequential muscle length L = 2πb..e. Example 2. multiple reflex responses are initiated whenever p A de viates from normal to adjust the cardiac output and/or total peripheral resistances as needed. develop a tran sient relationship between left ventricular pressure and ventricle volume (s ee Sketch). Sketch: Assumptions: Concepts: σR1 • Idealized • Laplace’s eq’n for thin- σR2 L=2πb geometry walled ellipsoids: •ٛ (muscle w/constant σ R σ R  y length) wall p =  1 + 2  h h thickness  R1 R2  a p(t) x and apical • Carlson’s wall-stress ∀(t ) lengths eq’n: (volume of semi-ellipsoid) σ w (t ) = σ iso − c ( dL dt ) b where σ w ~ σ R 2 Solution: The geometric parameters for an ellipsoid are: 2 • semi-ellipsoid volume ∀ = π a b2 .2b) 3 where p D is the diast olic pressure and p S-pD is the pulse pressure. semi-ellipsoidal shell. Q = 5L/min. and hence Rtotal = 20 mmHg/ L/min. Because proper operation of th e c ardiovascular sy stem is so important.206 Biofluid Dynamics where Rtotal is the s um of all pe ripheral res istances ca used by the vasculature. i. 3 • largest circumference.e. the heart can maint ain a p A of 100 mmHg by varying the b lood flow rate in light of changing flow resistance downstream. especia lly blood supply to the bra in and th e he art. Clearly. Specifical ly.

 dL  σ w (t ) = σ iso − c  (E2. σ R2 . yields an average wall stress. with a = constant and b = b(t). 0 ≤ x ≤ a . (E2. respectively. Laplace’s equation can be simplified to: σR p= 2 h (E2.12–3) an d u sing Eq.12-4b) in to (E2. .12-2b) we have L σw = p (E2. and L = L (t) is the muscle segm ent length. over the semi-e llipsoid. σ iso is the isometric stress.12-3)  dt  σw is the segmentally averaged wall stress.12-1) 6π With R1 >> R2.12-4a) and using Eq. (E2.12-5) 6π ∀(t ) ∀(t )  dt  where d∀ / dt =ˆ Q(t ) is the inflow rate and ∀(t) (or L( t) in Eq.12-1)) has to be given. 1 a π σw = a0∫ σ ( x) dx : = σ R2 4 (E2. 12-1) yields: a σ iso 4ch  d∀(t )  p (t ) = 8h −   (E2. 2 Thus.12-2b) 2 h 2πh which implies the assumption that σ R1 < σ R2 . Averaging the circum ferential st ress. (E2. In Carlson’s equation. a 2 ∀(t ) = L (t ) (E2. i. c is a muscle-tissue constant. (E 2..e.12-2a) R2 or. b L σR = p:= p (E2.Chapter II: Biofluid Dynamics Concepts 207 • ra dii of c urvatures at the s tress locations (s ee Sketch) of σ R1 and σ R are R1 = a 2 / b and R 2 = b .12-4b) 8h Inserting Eq.

79 t ) + 10..86 cos ( 7.0×102N⋅s/m3. i.395 t ) + 15. ∀( t ) = 101. recorded as a n electrocardiogram (ECG). The heart’s sounds and electrical activity provide non-intrusive m easures of its health status. The dual “lub-dub” sounds occur when the AV valves shut at the beginning of systole and then when the semilunar valves close. potential differences generated by the heart a re conducted to the body surface whe re electrodes (o r lead s) pick up the . at isovolumetric relaxation. For example.5 1 1. i. The distinct heart sounds are timewise correlated with the intensity of t he heart’ s elec trical signa ls. respectively. σiso=1.15 cos ( 14.15 sin ( 7. and are an example of fluid-struc ture-interaction.2 pro vides the br oader framework for aortic flow generation.e. Becau se tissue flui ds cont ain a high concentration of ions.63 sin ( 14.5 3 Time [s] Note: The parameters used are: h= 12mm.208 Biofluid Dynamics Graph: Left Ventricle Pressure p(t) [mmHg] 160 120 Left Ventricle Volume [ml] 120 80 80 40 40 0 0 0 0. an d can lead to pulm onary hypertension an d fluid in the lungs (i. the periodic heart/vessel expansion and contraction (see given ∀( t ) ) ca uses blo od inflow Q(t) an d sig nificant pre ssure bu ild-up.e.1.49 – 29. which generate ab normal blood flow patterns. c=3.395 t ) – 0. edema). The heart sounds are th e result of blood-induced vibrations of the cardiac structures.. a= 5cm.79 t ) Comments: Clearly.. while Fig.1.e. The Graph depicts the ∀( t ) and resulting p(t) interaction.40×104 N/m 2. The left ventricle volume ∀( t ) is approximately correlated from data in the literature.5 2 2. Heart disease detection. heart murmurs are caused by defective valves.

1 provides so me ge ometric chara cteristics of morphological characteri stics. which have major regulatory functions.2. which represent changes in potential between two regions on the heart’s surface that are produced by the combined effects of m yocardial-cell activities.5 mm 18 cm2 Arteries are viscoelastic multi-la yered conduits. Such signal s m odify t he heart’ s pum ping action as is appropriate to meet changing homeostatic needs of the body. deviation of a patient’s ECG from the age-related standard ECG spells trouble. the lumen surface plus la yers of smooth muscle cells.1. T he very im portant intima and me dia are comprised of the endot helium lining.000 cm2 Veins 200 0.5 cm 0.Chapter II: Biofluid Dynamics Concepts 209 conduction of electri cal impulses (i n millivolts) through the heart . All components of the circulato ry sy stem.500 cm2 8 Venules 10 70 µm 7 µm 4. Each cardiac cycle produces three distinct ECG waves. A dy sfunctional endothelium can lead to several arterial diseases as outlined in Chapter III. are line d with a layer of endothelial cells.5 cm2 Arteries 160 0. Table 2.2.3 The Blood Vessels As indicated in Fig. In summary. consisting of the intima.5 mm 40 cm2 Venae Cavae 2 3 cm 1.5 cm 2 mm 4.4 cm 1 mm 20 cm2 Arterioles 5 x 107 30 µm 20 µm 400 cm2 Capillaries 1 010 5 µm 1 µm 4.2. and • the ventricles must fill adequately during diastole. media. and advent itia from the inside-out.1 Geometric Blood Vessel Characteristics. healthy heart function requires that: • the contractions of all cardiac muscle cells occur at regular intervals in a synchronized fashion. Clearly.2. the cardiac function can be influ enced by neural inp uts from the auto nomic nervo us system. blood flow is propelled from the left heart chamber into the aorta and then pa sses co nsecutively through d ifferent types of vessels before it return s via the vena ca va to the right heart chamber. • the valves must open fully and close without leaks. including the blood vessels and heart chambers. and membranes as well as . 2. Table 2. Vessel Type Number Internal Wall Surface Area Diameter Thickness Aorta 1 2. • In addition to the electrical excitatory impulse. 1. • the muscle contractions must be strong.

1µm thick. are much smaller and more muscular. wh ich is a muscular sheet separating the thoracic and abdominal c avities. diameter redu ctions. After leaving the capi llaries. stenoses) or changes in tube material properties. th e d iaphragm. Thus. thereby creating a pressu re dro p b etween abdominal and thoracic vein segments which leads to blood flow back to the heart.. CO 2). the arterial system is viewed as a reservoir which cushions the pulsatile blood flow simil ar to an air -filled compression chamber (i. 1. where a capi llary bed constitutes an enormous surface area. b ends.210 Biofluid Dynamics collagen and elas tin fibers . Thus. Thus.. e lectrical signal. The arterioles then act as a high-resistance bed (or network). and about 0. Venous vessel s are thin-walled and quite distensible.5mm long. The Windkessel.g. venous capacity exerts considerable control in terms of cardiac pum ping and output. While skeletal muscles along veins can contract and squeez e venous b lood fro m the lower limbs to ward the heart. breathing can aid blood flow fro m the abd ominal to th e thoracic veins. . partial occlusions (e. Large veins feature one-way valves to prevent backflow. Capillaries are the narrowest of blood vessels and form the end points of the circu latory system . implying that they can work as reg ulators o f periph eral b lood flow through organs by changing the diameter and hence flow resistance. The la rge arteries expa nd when the blood pressure rises and contract during diastole. Arterioles. ex changing liquids. the normally dome-shaped diaphragm becomes flat or even protrudes into the abdomen. thereby supplying blood to the organs downstream. Specifically. i. causing near- uniform cree ping b lood flow in th e tis sue an d hence maintaining a high pressure throughout the c ardiac cycle. etc. veno us con tractions as well as skeletal muscle pum ping may strongly affect the venous blood volume and flow to the h eart. when a person inh ales. In the Windkessel model. gases (O 2. which is im portant wh ile st anding and duri ng exercise. A capi llary wall is just a layer of endothelial cells. and nutrients between the blood and the tissue. Peri pheral venu les and veins contain most of the total blood volume.1. the pres sure in the thoracic cavity decreases but increases in the abdome n. the blood pressure waveform generated by the hear t is assumed to be transm itted instantaneously throughout the arterial system and vanishes before the next heart beat so that each pressure waveform is the same in all major arteries. The W indkessel model ignores the facts that pressure waves require small travel times and that they are being reflected in somewhat short el astic tubes. blood is collected firs t in v enules an d then in veins (see Fig. a “Windkessel” in German) which smooths the intermittent pumping act ion of an old fire t ruck. nerve-triggered. known a s re sistance vessels.. During contraction. especially with downstream nonuniformities su ch as bifu rcations. cont racts. As a re sult.e.e.3 ).

2. (E2. Thus.13-2) Rt dt Separation of variables and integration yields for . from ventricular ejection) and the outflow Q exit.g. where E is Young’s modulus and ν is Poisson’s rati o. the storage rat e d ∀/dt is the dif ference between blood influ x Q in (e...4a-c) dt dt Rtotal Example 2. (E2.2. t s<t<T. determined by the pressure dr op or downstream resistance R total.4) we can write: p dp (i) Systolic phase: Q 0 – ---. Find p(t)/(Q0Rt) for the systolic and p(t)/p T for the diastolic phase. t s<t≤T.e. for DE=constant. the Windkessel) an d zero inflow du ring diastole.13: A Windkessel Application Consider constant blood flow Q0 during the systolic phase. into the arterial network (i. 0≤t≤ts. (2. Calculate the heart stroke volume ∀s.= D E -----. Sketch: Assumptions: Approach: DE/k • As stated • Lumped-parameter Windkessel model • p venous ≈ 0 Qin p(t) Qexit • Conditions: ∀(t) p (t = 0 ) = p 0 Rtotal • Qin = Q0 = c/ for 0 ≤ t ≤ t s p (t = T ) = p T Solution: Based on Eq. d∀ dp ∆p = DE = Qin − Qexit := Q(t ) − (2. Clea rly.13-1) Rt dt p dp (ii) Diastolic phase: 0 – ---. where T is the cycle duration and pT is the pressure at the end of diastole.Chapter II: Biofluid Dynamics Concepts 211 The differential change in storage capacity of such an elastic chamber with outflow resistance can be simply expressed as: d∀ = DEdp (2.2. Th e vascular b ed is described by ∆p=p-pvenous and the lumped resista nce R t.= D E -----.3) where DE is the chamber wall distensibility which is directly related to the bulk modulus k=∀/DE:=E/[3(1-2ν)].

8s. during outflow and vessel relaxation the pressu re drops to the prescribed pT=70mmHg. Pressure waves. Q0Rt=154 mmHg. D ERt = 0. The ef fect of pressure wave reflections generated by a single inlet pressure wave created b y liquid bolus injection into a straight el astic .= exp [ ( T – t ) ⁄ ( D E R t ) ] (E2. 2. (2. for example. sten oses) a t their ends. we have to d eal with pressure w ave re flections which lead to b lood pressure build-up causing peak pressure and peak blood flow to be out of sync (see Fig. the bloo d pre ssure pe aks at t= t systole=0. as shown in Fig.2.2. The W indkessel model is inappropriate for pulsating blood flow in (sh ort) elastic vesse ls wi th natural constrictions or pathological ob structions (e.13-3) Q0 Rt  Q 0 R t  D E R t p(t) and for Phase (ii): p diast ≡ --------.5s. Comments: As expe cted.13-5a) 0 For the present case ∀ s = Q0t s (E2. Then.3.5s when blood flow into the vessel system st ops.4 ts T t [s] Note: The parameters used are: p 0=pT= 70mmHg. The windkessel p(t) curve is a coarse approxim ation of an actual waveform.3) we have in general ts ∀ s = ∫ Qin (t )dt (E2.13-4) pT Related to Eqs. T= 0 .2. In such ca ses.2.1) and (2.g.13-5b) Graph: 120 psyst pdiast p [mm Hg] 80 0 0. t s= 0.557s.212 Biofluid Dynamics p(t) Phase (i): p syst ≡ ----------.3).2.= 1 –  1 – ----------- exp  – ----------- p0 t (E2..

the inlet (or incident) wave moves back and forth in the tube while it is r educed in strength due to viscous flow and wall ma terial ef fects. 2. p -p due to reflections at different tube ends and attenuations due to fluid viscosity and tube-wall damping.Chapter II: Biofluid Dynamics Concepts 213 liquid-filled tube with a partial opening located at different tube ends is illustrated in Fig.2.0 t/T Fig.2.5 1. 2.4. and resulting wave trains.3 Schematic pressure and flow waveforms during ventricular ejection in the ascending aorta showing delayed pressure peak after maximum flow rate due to pressure wave reflections. 2. pin(t). The reflected waves occur at longer intervals as the partially occluded tube ends are locat ed further away from the tube inlet. 120 p [mm Hg] p(t) 80 1000 Q [ml/s] ∆t 500 Q(t) 0 0 0. . Nevertheless. the mean pressure rises as a result of succe ssive reflections while the tube slightly expands by the volume of liquid injected. (a) Very short elastic tube (b) Medium to long elastic tubes pin (t) pin (t) Point of Measuremnts Partial tube-end •ٛ occlusion ٛ p (t) p (t) p (t) Fig. Thus.2.4 Schematic of single inlet pressure wave.

c) and 2π 1 Bn = π ∫ y(t )sin (n ωt ) dt 0 (2.. for exam ple). i. Q in = Asin ωt. + Bn sin (n ωt ) where: 2π 2π 1 1 A0 = 2π ∫ y(t )dt .g. Specifically . + An cos (n ωt ) + 2 (2. Example 2.5a) B1 sin ωt + B2 sin 2 ωt + . Find the blood pressures ps(t) and pd(t) during systole and diastole..14: The Aorta as an Elastic Reservoir Consider the sketched feedback control loop where the given flow rate during the sys tolic p hase. y(t) = y(t+T).d) are known as Euler’s formulae. a finite superposition of sine. and T is the period. after which y(t) re peats itself..2. can be readily analyzed with a Fourier series expansion. 1998. t is the time. e. ∀ Circulatory Network ..2.2. Also..5d) Equation (2. in radians per second. Sketch: Qin A ωt Qexit (t) Heart 0 π 2π Aorta pH(t) Qin (t) p.5c. which acts like an elastic reservoir.5b. 1971 or Greenber g..e.and cosine-functions (see Spiegel.2. respectively.214 Biofluid Dynamics For all practical purposes.. T-1 = f is the frequency and 2 πf = ω is the angular frequency. The postulate for any periodic function is the Fourier series: A0 y (t ) = + A1 cosωt + A2 cos 2 ωt + .5b) indicates that A0 = yaverage and Eqs. supplies blood to t he aorta. T ≈ 1 sec for the cardiac cycle.2. 0 An = π ∫ y(t ) cos (n ωt) dt 0 (2. being periodic functions. where y =ˆ p (pressure) or Q (flow rate).. (2. pressu re and flow waveform s..

14-2) dt Using the ∀(p)-relative for the aorta and Qexit = b p. a nd p 1. lumped-parameter approach.14-4) dt 0 during diastole subject to p (t = 0) = p 0 and p(t = π/ω) = p 1. p = p d ( t ) = p 1 exp [ – b ( t – T ) ⁄ a ] (E2.Chapter II: Biofluid Dynamics Concepts 215 Assumptions: • Systemic circulation consists of the heart.b)  0 for π < ωt < 2π (diastole) • ∀ = ∀ 0 + a p inside the aorta • Qexit(t) = b p Solution: Compartmental. p 0. and circulatory network • Heart rovides p Qin(t) to aorta during systolic phase • Aorta expands and accommodates the blood volume Postulates: A sin ωt for 0 ≤ ωt ≤ π (systole) • Qin (t) =  (E2. d∀ dp = Qin − b p = a (E2.14-3) dt dt or dp  A sinωt during systole a +b p =  (E2. aorta . 0 ≤ t ≤ π/ω: p = p s (t ) = p 0 e − b t / a + A[aωe − b t / a + b sin ωt - (E2. are also s ubject to the following constraints: ps (t = ts = π/ω) = p1 and pd(t =T=2 π/ω) = p0 ( E2. i.14-5b) The parameters a. where T is the cycle duration.14-6) . for the systolic phase.. d∀ = Qin − Qexit (E2.e. π/ω ≤ 2π/ω.14-1a.14-5a) aω cos ωt]/(b 2 + a 2ω 2 ) and du ring th e dia stolic p hase. Thus. b.

14-5) and (E2.. Find . as shown in Fig. plus the coef ficients a=2. which depends on the downstream re sistance of the circu latory ne twork.g.e. 1989)... the ao rtic pressure dec reases almo st linea rly du ring dias tole 0. E  R   2 p(t ) = Eε =   − 1 (E2. and R0 is the base radius.14-6).363 and b=1. fills the expanding vessel.216 Biofluid Dynamics Graphs: 160 ∀(t) 120 ps pd 120 p [mm Hg] ∀ [ml] 80 80 40 40 0 0.2. (E2.3). i.0. p 1=120 mmHg. Schmid-Schönbein et al.. i. p(t) drops marginally and then shoots up to p max (t ≈0. Th ese curves are essentially similar to ac tual waveforms (e. ∀(t=0)=50 ml. An empirical constitutive equation (cf. Qin delivered by the heart and Qexit leaving the aorta.e.916 according to Eqs.15: Transient Flow in an Elastic Capillary Consider Poiseuille-type flow in resp onse to a sudden change in inlet pressure for an elastic micro-tube.5. but they are more smooth. A=600 ml/s. Whil e blood is still being supplied the aortic pressure slightly decreases before the end of the syst olic phase 0≤t≤0.42s). Example 2. Comments: As net blood flow.5 1 t [s] Note: The parameters used are: p 0=80 mmHg. where E is the elast icity coefficient. R = R(t) is the tube radius. Based on the outflow condition Qexit(t)~p. 5<t≤1.2.15-1) 2  R0    relates l umen pressure to wall strain.

. i. (E2 . x) Solution: A 1-D mass balance yields: x ∂∀ ρuA x − ρuA x + ∆x = ρ ∂t ∂∀ ρu x +∆x ρu x ρ − ∂ (uA) ∆x = ∂A ∆x ∂t ∂x ∂t A(t) ∆x ∂A ∂Q ∴ =− (E2. as s hown in Sect. 7 (s ee Eq.15-1).7-6)) with ε=0: π [ R(t )]4 ∂p(t ) Q=− (E2. x=0)=p1 >p0 flow because ∆R << 1 • Linearly elastic tube wall and incompressible fluid Approach: • 1-D mass balance • Derive pressure equation for p(t. 3) to write: ∂p E ∂ (πR 2 ) E ∂A = = ∂t 2πR0 2 ∂t 2πR02 ∂t and hence .2 an d in Example 2.15-2) ∂t ∂x In light of Wo << 1.x) ω R0 2 Q x Wo ≡ << 1 ν x=0 p(t=0. where p = p0 at t = 0 for all x and then p = p1>p0 at x = 0 for all t. 1. we assume Poiseuille-type flow at any time t. we can use  2  R0  2 Eqs. x)=p0 x=L • Quasi-unidirectional p(t>0. Sketch: Assumptions: • Quasi-steady flow E because R0 << 1 and hence r R0 R(t) p(t.Chapter II: Biofluid Dynamics Concepts 217 the volumetric flow rates Q (x = 0) and Q (x = L) due to p(t.e. (E2.x). 3.15-2.15-3) 8µ ∂x 2 E R  2 E With A = πR and p =   − from (E2.

15-4) =− E ∂t ∂x Using Eq.15-4) ∂x can be approximately written as: ∂p E A0 ∂ 2 p − =0 (E2. the second term is negligible and Eq.6 in Sect.5 .τ ) 2 ∞ 1 = ξ + ∑ sin (πnξ ) ⋅ exp (− n 2π 2τ ) (E2.4 0.15-6b) ∂τ ∂ξ 2 As shown in Example 1.2 0.3 0. Graph: Q/Q0 6 P 1=1.15-7) P1 − P0 π n =1 n with whi ch t he volumetric flow rates at ξ = 0 and ξ = 1 can be estimated from Eq. and t = τ : E A0 ∂P ∂ 2 P − =0 (E2.15-3) in the form: 1 ∂p Q=− A2 µπ 8{ ∂x κ we can write ∂Q ∂2 p ∂p ∂A (E2.218 Biofluid Dynamics 2πR02 ∂p ∂Q (E2.2P 0 Q(ξ=0) 4 2 Q(ξ=1) 0 τ 0 0.2.15-3). (E3.15-5) = κ A2 2 + κ (2 A) ∂x ∂x ∂x ∂x ∂A With = O(± ε ) . using Eq. (1.1 0. x = ξL.2). P1 − P (ξ . (E2.15-1) (see Graph). (E2. Here.3. (E2.15-6a) ∂t 16πµ ∂x 2 16πµL2 In dimensionless form with p = EP. x). the separation of variables method leads to an infinite series representation for p(t.

2p0. as in the alimentary canal. the effect of p(t>0. 8µL Comments: With the simplification ∂A ⁄ ∂x ≈ 0 . F or m ost app lications we have small parameter values.04 and then after the time lag. λ its wavelength. Q(ξ=0) is high and then decreases exponentially. The key objectives are to compute the fluid axial pressure gradient resulting from the press ure difference at bot h co nduit end s and the sup erimposed traveling wall wave. Q=Q[p(x. peristalsis is the process of wavelike mus cular wall contractions that propel (axi ally) cont ained matter in tubular organs or other conduits. capill aries. i. glandular ducts. . While pressure-wa ve propagation in previous examples was c aused by upstrea m forces. e. Specifically. a redu ced form of the Navi er- Stokes equations is s et up with a sinusoidal displacement wave. Q(ξ=1)=0 for 0≤τ≤0. and the conditions of reflux. i.7) • the wave frequency ω = 2πc / λ (2.e.. κ << 1.6b) • a Reynolds number Re = Nca / v (2. we now can form: • a dimensionless wave number N = 2π a / λ (2.2. a s a boundary condition. and c. the maxi mum wave amplitude.6a) • an amplitude ratio κ = ho / a (2. and the alveolar region. Based on these parameters.3 . an d N ≤ O(1). the peristal tic pum ping parameters includ e: ho. .16 illustrates a peristaltic problem solution. trave ling in the conduit wall with const ant speed c.Chapter II: Biofluid Dynamics Concepts 219 4 π ER 2 Note: Q 0 = Q ( ξ = 1.e. Qexit approaches Q0. τ → ∞ ) = -------------0 ( 1 + 2P 0 ) ( P 1 – P 0 ) .. Exa mple 2.t)]..g. Re < 1. With the assumed outlet pressure pout=p0. With a sudden jump at the inlet from p0 to p1=1. the velocity field (or volumetric flow rate). heart-pumping action. 0. ξ=0)=p1 is not felt at the tube exit. For a basic elastic conduit of radius r0 or half-width a.. e.2. the wave sp eed.8) In order to solve peristalsis problems. and κ ≈ 0.2. the influence of R(t)-variations is eliminated and as a result the flow rate is determined by the pressure in space and time. N = 2. peristaltic fluid motion is caused by periodic contractions of the tube walls. the equilibrium flow rate.2. Peristaltic motion. i. An exception is the gastrointestinal tract where Re ≈ 10.

and u = uˆ + c .20b) reduces to: dp ∂ 2 uˆ 0= +µ 2 (E2. 1969) Consider S tokes flow in a 2-D channe l representing the ureter (se e Sect. (E2.16-3c) h ( x.t) a fluid motion moves with h(x.3. Sketch (Actual Frame): Assumptions: Concepts: c • Steady-state • Observer y u(x. and c ≈ 3cm / s .16-3b) λ η = y⁄a (E2. Find the velocit y profile.3. 2πa 2 dp 2 u= ( )(η − H 2 ) (E2. where capped values refer to the “wave frame.16-2a.16-3e) .t) in wave frame wave speed x • Poiseuille • Reduced N-S L type flow equations Solution: The stationa ry and movi ng frames of reference are connected as x = xˆ + ct . volumetric flow rate.16-3a) 2 µ λ dξ where.: = 1 + ----0 cos ( ξ – τ ) (E2. x ξ = 2 π --.16-1) dxˆ ∂yˆ subject to: ∂uˆ uˆ ( yˆ = h) = −c and =0 (E2. length L= 30cm. where Re < 1 an d λ >> a.b) ∂yˆ yˆ = 0 Double integration yields after tran sformation back to the stationary frame.y.220 Biofluid Dynamics Example 2. with a ≈ 2mm. λ = 12cm. and overall pressure drop. y = yˆ .” Thus.1).16: Peristaltic Pumping (after Shapiro et al. 4.16-3d) a a and τ = 2 πct ⁄ λ (E2. t ) h H = --------------.. the x-momentum equation (1.

and using Qˆ = Q − 1 from Eq. and when ∆ P λ = 0. H ) = (Q + H − 1) (E2.16-8) Qmax = 2 +κ 2 In the stationary frame.16-6) o H o H For one wavelength.Chapter II: Biofluid Dynamics Concepts 221 ∫ The volumetric flow rate. (E2. we obtain: 3κ 2 (1 − κ 2 ) 5 / 2 Q= − ∆P λ (E2.16-9) 2 H3 . Integration of Eq.16-4a) 3 µcλ dξ dp where 2πa ( ) /( µcλ ) : = dP / dξ is a dimens 2 ionless pre ssure dξ gradient. so that the dimensionl ess pre ssure gra dient ca n b e expressed as: dP 3[Qˆ (τ ) + H ] =− (E2. Q = u dA. we notice that Eq. (E 2.e.. (E2.16-4a) can be split into Qˆ (τ ) + H . λ = 2 πct/τ.16-4b).16-4a) over one wavelength yields the time-averaged flow rate: 2π 1 Q= 2π ∫ Q (ξ . integration of Eq. when Q = 0. (2. i.16-5) over the channel length L yields the pressure drop: dξ dξ L L ∆P(τ ) = −3∫ 2 − 3Qˆ (τ ) ∫ 3 (E2. 3κ 2 (E2.16-5) dξ H3 Now. the velocity can be expressed as: 3( H 2 − η 2 ) u (η . ∆ P λ is a maximum. can be calculated as: A 4 πa 2 dp 3 Q(ξ . Clearly.16-7) 2 + κ 2 3π (2 + κ 2 ) where ∆ P λ is the pressure c hange p er wavelen gth and 0 ≤ κ = h o ⁄ a ≤ 1 . Q .τ ) = − ( )H (E2.τ ) dτ = Qˆ + 1 0 (E2.16-4b) In order to eliminate Q .

τ) ∆Pλ=1 0 Q 0.4 0.e. and near/at the moving wall ( η= 1. Q = 0 .5 -0.4). κ∼ h0 (see Eq.1. As mentioned.6 u(η. c.6 0.e. x=12cm. fluid moves back and fo rth over 0 ≤ τ ≤ 2 π . is “infinite” and so is the wave speed. the velocity is the highest. blood is a shear -rate depend ent fluid flowin g in pulsatil e fa shion thro ugh non - rigid bifurcating nonplanar tubes with variable cross section s and “surface roughness. the local axial velocity oscillates.. i.. Assuming a peristaltic action whic h generates per wavelength/cycle zero net flow. i. wave reflections do not exist because Young’s arterial wall modulus . the time-averaged flow rate Q depends mainly on the degree of tube.4. Blood Flow in Bifurcating Arteries. When assum ing mediu m-to-large noncoronary arteries to be quasi-rigid due to age or disease. As ex pected.25 0.222 Biofluid Dynamics Graphs: (a) Time-averaged flow rate (b) Profiles of axial velocity 1 0.8 1 π/2 π 3π/2 2π κ τ Note: For figure (b). Comments: Clearly. pre ssure change s at the tube inle t f or thin-walled tubes ( h<<D) and negligible viscous ef fects are insta ntaneously propagated. E. in ac tual arterial tre es (or stenosed blood ve ssels) w ave refl ections are c aused by ge ometric viscoelastic non-uniformities. As a re sult. at the ce nterline. and κ=0. As discussed (see Fig. o ccurring simu ltaneously a t every po int a long the tu be.2 η=0 η=0.wall contraction. Th e application criterion is the magnitude of .2 0.4 ∆Pλ=0.2 0 0. η= 0.8 η=0.5 0. (E2.1 0.1 η=1 η=0.” While steady or pulsatile flow of an incompressible fluid in a straight rigid or sli ghtly el astic tu be are oc casionally re asonable starting points when m odeling the hemodynami cs.2 ∆Pλ=0 0 -0. and only weakly on the pressure drop per wavelength. ∆ p λ .0) it i s the lowest.1.75 0.16-3d) ). such traveling pressure waves may alter the (axial) pressure distributions and hence the velocity field and any particle tran sport.

we v r assume that ---. i.Chapter II: Biofluid Dynamics Concepts 223 c 2 = hE /( ρD) (2.17-1) r0 Sketch: Assumptions: Approach: E.« 1 . ρ t h • As stated • Reduced N-S r0 p(t) vr . Pulsatile fluid flow in elastic conduits. was illustrated in previous Exam ples and is further discussed with solutions to wave reflection problems by Zamir (2000) and Nichols & O’Rourke (1998). u • Frictionless flow z • Hooke’s law σθ }η(z • Displacement σθ . The flow is powered by a pressure pulse of wavelen gth λ and period τ where the wave speed c ≡ λ ⁄ τ » v z .t) and wall thickness h. thin-walled elastic tube of radius R(t)=r0+η(z. th e redu ced Navi er-Stokes equations read ∂v z 1 ∂ (continuity) + (rv r ) = 0 (E2.9) which is known as the Moens-Korteweg formula (see Mil nor. 19 89 and Example 2.t) we assume η σ θ = Eε 0 := E ---.∼ ---.t) acceleration η&& is • Newton’s 2nd law r p(t) dθ negligible Solution: Based on th e assumptions. Example 2.17-3) ∂t ∂r . (E2.17).17: Derivation of the Moens-Korteweg Formula Consider transient axisymm etric f low in a very lo ng. For the (dom inant) hoop stress in a thin- r 0 vz λ walled linearly elastic tube with radial di splacement η=η(z. the axial flow velocity. Furtherm ore.e.17-2) ∂z r ∂r ∂v ∂p (r-momentum) ρ r =− (E2. v equations r vz. blood vessels as well as l ung airways..2.

(E2. t ) + 2πvr 00 = 0 (E2.17-6) dη r ∂u =− 0 (E2. v ≡vr is the radi al velocity comp onent su bject to v (r= 0)= 0 and v(r=r0)=v0(z.e.17-9a) or (ρ t hr0 dθ ) d η2 2 = pr0 dθ − σ θ hdθ (E2.17-4).17-7) dt 2 ∂z Integration of Eq. (E2.t).17-10) and neglecting d 3η / dt 3 we have ∂p hE ∂η = (E2. (E2. yields ∂u 1 ∂p =− (E2. in order to relate the fluid mechanics and induced wall motion to the lumen pressure.17-1) and solving for p(z. i. 17-9b) dt 123 123 F pressure Fstress Inserting Eq. 17-10) r02 Differentiation of Eq. the radial wall velocity.17-5) 0 where u = u ( z . and hence with Eq. r 17-6) Here. (E2.17-7) yields . v 0(z. assuming that the pressure does not vary across the tube. we emp loy Newton’s Second Law of Motion for the tube-wall segment of area h(r0dθ) (see Sketch). t ) = η + hρ tη&& (E2. we obtain hE p( z.224 Biofluid Dynamics ∂v z ∂p (z-momentum) ρ =− (E2.t).17-1 1) ∂t r02 ∂t Combining Eq.. (E2.17-8) ∂t ρ ∂z Now.t)=dη/dt.17-11) with Eq. mtube a r = ∑ Fr (E2. (E2. Clearly . t ) . p (r ) = c/ . we obtain ∂ ∂z [ ] πr02 u ( z .17-4) ∂t ∂z Integrating the continuity equation over 0 ≤ r ≤ r0 with dA = 2πrdr and defining the volumetric flow rate as: r0 Q = u A = 2π ∫ v z rdr (E2.

1999). b) If also viscous effects for (blood) fl ow in thin tubular elastic arteries are important. This is very efficient because it eliminates the need for parallel blood vessels starting. (E2. z  z  p( z . thereby forming a tree structure (cf. hE ≡ c 2 (E2. the celebrated Womersley model as di scussed by Nichols & O’Rourke (1998) should be considered (see Sect.2. .17-13) = ∂z 2 hE ∂t 2 where the coefficient is the Moens-Korteweg formula. i.17-8) yields 2r0 ∂ 2 p ∂  ∂u  ∂  ∂u  1 ∂ 2 p = −  ≡ − = hE ∂t 2 ∂t  ∂z  ∂z  ∂z  ρ ∂z 2 so that ∂ 2 p 2 ρr0 ∂ 2 p (E2. The same holds for the bronchial trees in the lungs.e. In ord er to pro vide bl ood. main or parent vessels bifurcate into smaller daughter vessels.Chapter II: Biofluid Dynamics Concepts 225 ∂η r02 ∂p r ∂u (E2. t ) = A sin  − ωt  and η ( z . i. (E2. oxy gen and nutrients.5). t ) = B sin  − ωt  λ  λ  (E2.3).2. to different organs and tissues. (2.1. 1.. =− 17-12b) hE ∂t ∂z Differentiation with respect to time and using Eq.17-15a.17-12a) = =− 0 ∂t hE ∂t 2 ∂z or 2r0 ∂p ∂u (E2.17-10) may be ob tained with postulates such a s Eq. While a tree structure la cks re dundancy—for ex ample. Karch et al.2).. say.e.. In turn. from the left ventricle (see Fig. they bifurcate to even smaller ones. and it reduces the heart’s required pumping power. blo ckage of (the ) parent tub e cuts of f b lood to all su bsequent daughter tubes—the total power can be optimized for an arterial tree. for example. 17-14) 2 ρr0 Actual internal pressure and wa ll displacem ent solutions of Eq.

Specifically. in compressible.226 Biofluid Dynamics When estimating the total power. 1 d du 1 dp (r ) = = c/ (2.2. Hence.12) Equation (2 .11a) where: 8µL ∆p = Q (2. A ≡ 8µLQ /p and B is a positive constant independent of Q. R6 = 2A/B. i. (1.11c) πR 4 Clearly.e. 8 µL 2 Pdyn = Q (2.10) r dr dr µ dx du subject to u(r = R) = 0 and = 0 where 0 ≤ r ≤ R and dr r =0 dp − = ( pin − pout ) / L = c/ . an d fully de veloped in rigid circular tubes.11b) πR 4 from the solution of Eq.66) where g ψ ≡ 0).2.2.. a s the tube diam eters get sm aller in the arterial or bronchial tree .2..3. dx The pumping power Pdyn is requested to maintain a given flow rate Q.. filling and maintaining a vessel with fluid requires Pbio ~ V fluid ~ R 2 L which implies: Ptotal = Pdyn + Pbio = AR −4 + BR 2 (2. i. Thus. Thus. Q ~ R3 (2. Poiseuille flow (see Eq. s o are th e flow rates reduced according to the .e.2. i.2.2. (2. to find the optimal radius which minimizes Ptotal. lamina r. the pum ping power Pdyn ~ R −4 .11c). i.2. Pdyn = ∆pQ (2. we set dPtotal/dR: = 0.12) is kno wn as Murray’s law or the “c ube la w.” indicating th at m inimum energy exp enditure i s achieved under this correlation. lar ge blood or air conduits should be benefi cial. e. (2 . Indeed.. or. with Q known.2. However .10) for a given Q. to overcome the flow resistance employing a pressure difference ∆p = pin – pout. a major simplification is to assume that fluid flow is ste ady.e.11d) 2 From Eq. Ptotal = Pdyn + Pbio required for both fluid dynamics and biological purposes.

bifu rcations may b e asymmetric.2 =ˆ da ughter-tube radii. pro vided th at the underlying assumptions are met.12).2. Both daughter tubes have the radius R2 . It is of interest to note that the wall shear stress (in Poiseuille flow) relates to Q/R3 as well. 1+α 2 β= (2.e. Example 2.2.26 and R1/R0 = R2/R0 =0. wh ere R0 =ˆ parent-tube radius. Clearly.2. index α = R 2/R1.15a)  R0   R0  or in terms of the bifurcation index α as: β = (1 + α 3 ) −2 / 3 + (1 + α −3 ) −2 / 3 (2.14b) Then the area ratio b can be expressed as: 2 2 R   R2  β =  1  +   (2. and R1 at the inlet (x=0) and at the bifurcation ( x = l) by p x and 2Θ. 4µ Q τ wall = ( ) (2. out-of -plane.14a) and hence with the associated cube law (2.Chapter II: Biofluid Dynamics Concepts 227 proportionality (2.. i.18: Steady Blood Flo w in a Bifu rcating Tube and the Evaluation of a Potential Stent Migration Force Consider Poi seuille-type flow in an axisymmetric bifurcation.e.15c) (1 + α 3 ) 2 / 3 For example. and with different daughte r-vessel diameter s.2.2. The parent tube is characterized by p1.794. Apply the model to: Case (i) a bifurcating . and averaged veloc ity v 2 . Thus.2. v 1.0 and hence β = 1.. for a symmetric bifurc ation. length L D = ( L − l) / cos Θ. α ≡ 1.2. Recalling mass conservation: Q0 = Q1 + Q2 (2. and area ratio β = (R12 + R22)/R0.6): 3 3 3 R0 = R1 + R2 (2. These numerical values may serve as a guide in the analyses of a rterial a nd bro nchial t ree s tructures. outlet pressure p 2 .13) π R3 The cube-law can be employed when describing geometric aspects of bifurcations and ultim ately trees. three system-indicative parameters are a bifurcation’s angle θ. as demanded by local air or blood sup ply. and R1.15b) i.2.

p 2 Q2. leads to P ( x = l opt ) = Pmin . R1 .18-2) so that (see Eq.e.e.228 Biofluid Dynamics blood vessel and estimate the necessary pumping power P = ∆pQ as well as the best bifurcation point x = l opt when P = Pmin .. p 2 ). Then. i. Ptotal = Q1 ( p1 − p x ) + 2 Q 2 ( p x − p 2 ) (E2. p 1 • Steady laminar Case (i): R1 fully-developed • Ptotal = ∑(∆pQ) i x flow i • Constant fluid • Hagen-Poiseuille px properties solution Q ~ R 4 x=l • Axisymmetric bifurcations Case (ii): with smooth • Reynolds R2 rigid walls Transport Theorem θ • Pbio ≈ 0 and Bernoulli’s Equation x=L { Q2 . Sketch: Assumptions: Concepts: Q1. of interest is the pulling (or m igration) force exerted by the net m omentum change on t he b lood v essel o r equ ivalently o n a bifurcating stent-graft (see Sect. P = fct ( l. (2. dP / dl : = 0 .11a)).2.5.2. For Case (ii). Ptotal = Q1 ( p1 − p 2 ) (E2. an equation fo r the total pumping power as a function o f th e p arent-tube length has to be d erived.18-3) .11)): Ptotal = Q1 ∆p1 + 2 Q2 ( p x − p 2 ) i.. p1 . setti ng the first derivative to zero. h. e..18-1) where Q1 = 2Q2 (E2. Case (ii) a bifurcating stent-graft and evaluate the axial force potentially leading to stent-graft migration. R2 . i . Case (i): The pumping power required t o m aintain the flow rates through the bifurcation is (see Eq. p 2 h Solutions: For Case (i).4). (2.

18-7a. Inserting Eqs. (2..Ptotal ( x = l opt ) = Pmin .b) into Eq. b) Note that h 2 + ( L − l) 2 = LD = ( L − l) / cos Θ is the length of a daughter tube. (E2.18-7a. ∂r r =0 Double integration and invoking the boundary conditions yields: 1  ∆p  2 1 ∆p u (r ) =  4µ  l  ( 2  R − r .18-10) R18 − 4 R28 14243 h Recalling that . 0. b) so that in general r R π  ∆p  4 Q = ∫ v ⋅ nˆ dA = 2π ∫ u (r )rdr :=  R (E2.18-4) r dr  dr  ∂x ∂u subject to u (r = R ) = 0 and u (r = 0) is finite (or = 0 ).18-5a. Eq. (E2.18-6) A 0 8µ  l  In the present case. where 4µ l ) = u max (E2. i. with d 2 P / dl 2 > 0 we obtain l opt by setting dP / dl : = 0 .2. from which 4 2 R2 l opt = L − ( L − l) tan Θ (E2.18-3) can be written as: 4 4 π R1 R2 ( p1 − p 2 ) 2 Ptotal = (E2. (E2.18-2) yields: 4 4 p1 R1 h 2 + ( L − l) 2 + 2 p 2 R2 l p x (l ) = 4 4 (E2.Chapter II: Biofluid Dynamics Concepts 229 Considering Poi seuille fl ow with − ∂p / ∂x = ∆p / l = c/ and r v = [u (r ).0] .e.10) reduces to: µ d  du  ∂p r  = = c/ (E2.18-9) 4µ R1 4 h 2 + ( L − l) 2 + 2 R2 4 l Now.18-8) R1 h 2 + ( L − l) 2 + 2 R2 l so that Eq. . π p1 − p x 4 π p x − p2 4 Q1 = R1 and Q2 = R2 8µ l 8η h + ( L − l) 2 2 (E2.

9) .18-12a) c.3.. 2 p1 A1 − 2 p 2 A2 cos Θ − R x = ρ u1 A1 (−u1 ) + 2 ρ u 2 A2 cos Θ (E2. (E2.18-11)  R1  Case (ii): T he net axial u2 p2 force exerted by the fluid C.18-12) reads: 2 2 A1 2 Rx = p1 A1 + ρ A1 u1 − ρ u1 cos Θ 2 A2 (E2.∀.230 Biofluid Dynamics L−l π cos Θ = . Eq.e. i. can A1 be evaluated with a reduced C. (E2 .18-15)  ρ 2 A  2 − 2 A2  p1 + u1 1 − 1 2  cos Θ  2  4 A2  . where 0 < Θ ≤ . (1.18-14) Thus. u 1 A1 = 2u 2 A2 (E. eliminating p 2 and u 2 with Eqs.18-13) 2 2 From mass conservation. (E2.18-12b) From Bernoulli along a representative streamline in the parent- daughter tube for frictionless flow: ρ 2 ρ 2 p1 + u1 = p 2 + u2 (E.18-13 an d 14).s.18-10)) as: 4 R  cosΘ = 2  2  (E2. form of Eq. using u2 p2 averaged velocities: r r ∑ Fx = ∫ u ρ v ⋅ dA (E2.2.2. A2 flow due to net m omentum u1 x θ change.S. 2 2 h + ( L − l) 2 we can express the half-angle of bifurcation associated with Pmin (see Eq. neglecting frictional p1 Rx and gravitational effects.

Θ = 30° . e. i. . In contrast.19: Blood Flow in Noncircular Tubes Arteries. an d half-bifu rcation ang le 0 < Θ ≤ 80° . D 1 = 22 mm . the axial force increases with increasing parent tube diameter and bifu rcating an gle because sudden co nstrictions. find expressions for the volume tric flow rate a nd the neces sary pumping power. p1 = 120 mm Hg . aortic neck diameter 20 ≤ D1 ≤ 30mm . the parameter ranges are: blood flow rate 500 ≤ Q1 ≤ 2000 m L / min . As a base case. Graphs: (a) (b) (c) 5 10 6 4 8 4 3 6 Rx [N] Rx [N] Rx [N] 2 4 2 1 2 0 0 0 500 1000 1500 2000 20 25 30 0 20 40 60 80 Q1 [ml/min] D1 [mm] Θ [degree] Comments: The b lood flow ra te h as a negligible ef fect on the axial force. Assuming stead y ful ly-developed flo w of a Newto nian flui d..Chapter II: Biofluid Dynamics Concepts 231 For the human abdominal aorta. It shou ld be noted that frictional ef fects (neglected here ) a re indeed only minor contributors and that the computed stent-migration force range 2 ≤ R x ≤ 8 N matches in-depth analyses (see Sect.e. when noncircular .4. as well as a large bifurcation an gle cause hig her net m omentum chan ges. may be elliptic in cross-secti onal area. on e co uld take Q1 = 1 L / min . Example 2. from a relatively large parent tube to small daughter tubes.2). i. an elevated net pressure drop. and D2 = 12 mm (see Graphs).. 5.

K = R an d the Poiseuille fl ow results are recovered.19-5) πK 4 Clearly. z ) =  2 − 2 − 1 (E2..3.24 are directly culled from Chapter II an d th eir so lutions sho uld deep en th e b asic un derstanding o f biofluid dynamics concepts. y2 z2 (E2. White 2003): ∆p (bc) 2  y2 z2  u ( y.g .19-3) 2 µL b 2 + c 2 b c  The flow rate is: π∆p 4 Q= K = A⋅ v (E2. Mo ore & Zo uridakis. The solution then is (Zamir.3 HOMEWORK PROBLEMS Notes: Ho mework problems (HWPs) 2.19-4) 8µL where K ≡ [2(bc)3/(b2+c2)]1/4 and hen ce the pumping power can be expressed as: 8µL 2 Pdyn = Q (E2.65)) reads: ∂ 2 u ∂u 1 ∆p + 2 = = c/ (E2.19-2) ∂y 2 ∂z µ L subject to no-slip boundary condition and symmetry on y = z = 0. 2. when b = c = R.19-1) + =1 b2 c2 the governing equation (see Eq. "Reasonable system parameters" refers to the selection of physiologically correct values as g iven in BME hand books (e. . 2000. (1.1 -2.232 Biofluid Dynamics Sketch: Assumptions: Concepts: z • Steady laminar • Reduced N-S c unidirectional equations x flow • Pumping power y • Constant Pdyn = Fpressure ⋅ v b properties • Rigid walls where F p = ∆pA Solution: For axial flow in a duct with an elliptic cross section.

2.x). 2002. c T=cT(x.5. Truskey et al.3 to n com partments in parallel. 2. Graph cv(t) as well as cB(t) and cT(t) and comment on the influence of key system parameters. i.” Plot the species concentrations and comment. only P oiseuille flow u(r) and a very small membrane flux VM = c/ as a bo undary co ndition are assumed (see Sect. 2004. 2. where i=1.7 Revisit HWP 2.e. 2.5) ). (2.g.2-3) . 2.6 Consider t he embedded m embrane capillary depicted in Fig. select reasonable system parameters and inlet conditions and solve: (a) Equation (E2.2 List the underlying assumptions as well as the pros & cons of "well-mixed compartment" models. R2.1. 2.Chapter II: Biofluid Dynamics Concepts 233 2003) a nd su pplementary te xts (e. Provide exam ples of "equilibrium assum ptions/conditions" used to solve complex transport phenomena.t) (see Eq.2-2a.6 and lift the a ssumptions of co nstant tissu e concentration. .1.e. 2..8 Redo Example 2.t).5 Considering Example 2... Is t he Reynolds Transport Theorem a lumped-para meter blac k-box rese rvoir or compartment model? 2. i.).6b). u= u(r. dynam ic equilibrium ..4 Generalize the re sults of HWP2.2.. 1998. R1 and ∀2. 2.3 and Example 2. 2. Bird et al. however. N ichols & O’Ro urke.1 Contrast static vs . Fu ng. 19 97. Graph the results and comment.b). etc. S olve the PDE for c(x.2.9 Develop a " compartment-in-series" model with Tin ≡ TA = 32°C and possible thermal boundary conditions to simulate body heat transfer (see Fig.7).4 with a broader range of inlet conditions and system parameters. and (b) Equations (E2.2(b).1.3 In reference to Fig.1. set up the equations to find cv(t) for a given Q(t) and ca(t) as well as ∀1. n. assuming reasonable “c onstants.1. 2. 2.3 .. and uniform capillary flow.

red bl ood cell diameters may be s ignificant compared to vesse l diame ter s o that the fluid cannot be treate d as being homogeneous (Sigma effect).10) and provide a state-of-the-art review of a dvancements and applica tions of the biohe at transfer equation.1.e. 2. inflow. i.e. (2.. and outflow. pressure.1.11 Reconsidering Example 2. Find the velocity profile.16 Plot for a given L(t) (see Eq. set up a mathematical model and describe necessary laboratory experiments.7-2) to (E2. 2. In this c ase.20).19 Similar to Ex ample 2. which operational conditions and system parameters have to be reasonably changed to produce a nonlinear T(x).234 Biofluid Dynamics 2. 2. .. 2.t). Fin d th e app arent viscosity of blood considering the Sigma effect. now analyze the peri staltic pumping for two-dimensional flow in a circular tube of radius a. Graph the sim ulation results and comment. Which s ystem parameters are most influential? 2.12 Study the paper by Oka (1981) and related publications in order to derive Equations (E2. stressing theo ries of the three lubrication modes.1. velo city ∂h / ∂t .1). volum etric fl ow rate.14)) the squeeze-film pressure p(θ. i.1.14 Concerning “jo int lub rication an d wea r.t).7-2) to (E2.17 In very small vessels.13 Find the s pecies blood and tissue concentrations (see Example 2.5. 2.” pro vide an updated literature review .16. 2.18 Discuss th e relation ship among left ventricular vo lume. and spacing h( θ. E stablish lim iting parameter ranges for wh ich Equations (E2. an d ov erall pre ssure drop.15 Derive Equation (2. such as the capillaries. 2.7-5). parameter values and experim ental/computational results (see Table 5. 2.8) for reasonable sy stem pa rameters. show the influence of bioheat convection? Simulate a case study.7-4) are valid.10 Discuss Equation (2. the fluid m ay be ap proximated as a summation of finite cylindrical layers with thickness equal to the red b lood cell diameter .

(b) Multiply the equation through by exp(t/RC). and the cro ss sectional area A=A(x.20 Windkessel t heory (cf. (c) Select a couple of Q(t)-functio ns. and solve for p(t)..t). m easured. Sect.21 Pulsatile flow in elastic tu bes: An improvem ent over the Windkessel Theory is 1-D in viscid flow where the pressure is p= p(x. or left vent ricular ejection history. and comment.Chapter II: Biofluid Dynamics Concepts 235 2. and the diam eter-pressure relationship can be reduced to ∂u α ∂p + =0 ∂x R ∂t and ∂u 1 ∂p + =0 ∂t ρ ∂x (c) Show that the last two equa tions can be com bined to the wave equation: . Show that the continuity equation. d escribing a pres sure wave. (b) Assume u<<1 and hence u∂u / ∂x ≈ 0 as we ll as A = πR 2 .13): For example. the inflow rate. the simplified momentum equation.2. (a) Interpret the biofluid mechanics meaning of ea ch term and determine the dimensions of C and RC. 2 R = 2 R0 + αp with α being the com pliance constant of the tube.2. integrate from t=0 to t=t. the aorta is mode led as an elastic cha mber and the peripheral b lood ve ssels as a representative rigid tube with resistance R. plot p(t).3 an d Examp le 2. (d) How cou ld th e co mpliance b e d efined.t): ∂A ∂ + (uA) = 0 ∂t ∂x ∂u ∂u 1 ∂p +u =− ∂t ∂x ρ ∂x p − p ext = p ( A) where p ext is the pressure acting on the outside of the tube. calculate p(t). an d/or calculated? 2. i. can be expressed as: dp p Q (t ) = C + dt R where C is the wall compliance. Thus. Q(t). where . e. (a) Indicate the derivation of the first two equations.

236 Biofluid Dynamics ∂2 p 1 ∂2 p − =0 ∂x 2 c 2 ∂t 2 where . say.23 Flow through a concentric annulus. draw the secondary velocity pattern in the daughter tube during inspiration and that in the parent tube during expiration.h = h0 + α ( pb − p a ) (b) Assuming that h=h(x) on ly. 2.t) as well as the evolution of u(x. . 2. There is no slip at the inner (r=b) and outer radius (r=a). (b) Draw the laminar.” Section 6.t) for suitable constants and initial/inlet conditions.22 Fung (1990) outlines in “Biomechan ics. (e) Look up the analytic al solution to this problem and plot p(x. Considering steady fully- developed flow of a Ne wtonian fluid in the an nular s pace between two concentric cylinde rs. find expressions f or the volumetric flow rate and friction factor. (a) Derive the Laplace equation describing h=h(x. microcirculation as blood fl ow in a capillary sheet of.y) as  ∂2 ∂2   2 + 2 h 4 = 0  ∂x ∂y  where . where the sheet thickness h = h(∆p = pblood − pair ) .24 Bifurcating vessels: (a) Draw the streamline of flow in a model of the ab dominal ao rta containing th e left an d righ t renal arteries (see Graph A).c = R /(αρ ) (d) Postulate for the solutions of the equations in Section (b) p = p 0 f ( x − ct ) + pˆ 0 g ( x + ct ) and u = u 0 f ( x − ct ) + uˆ 0 g ( x + ct ) and show that “amplitudes” p0 = − ρ cu0 and pˆ 0 = ρ cuˆ0 Provide a physical meaning of these postulates/solutions. axial velocity profiles in a model of the carotid artery bifurcation (see Graph B). de rive an ex pression fo r th e blood flo w rate Qˆ = hU p er unit widt h an d p lot h(x) for reasonable parameters. the alveolar region. (c) Considering a symmetric bifu rcating airway (see Graph C).7. 2.

A. C.M. J.” Oxford University Press. Berger. New York. K.” Elsevier. Acad. 35:721-735. (197 6) “Biom edical En gineering Prin ciples. 335: 137. N ew York. (1 992) “ Mathematical Models in Bioheat T ransfer” In: Bioengineering Heat T ransfer (Y. O. Cebeci T .) (1996) “Introduction to Bioengineering. E. (1959) “Conduction of Heat in Solids. (1999) Powder Technology. and Kleinstreuer. (2004) “Analysis o f T urbulent Fl ows. ed. S. an d Sinha. 120: 446-454. (eds. ed .P. and Lightfoot.D. B.. and Holnes. a nd J aeger.) Advances in Heat Transfer. Carlson. (1998). DC. Cho. E. W.C... K.Chapter II: Biofluid Dynamics Concepts 237 Graph A: Graph B: Graph C: References Bellhouse. UK. Am. M.. H. F . . N. N.S. Agarwal.” John Wiley. Society. UK.I. Vol. K. Carslaw. and Talbot. D. Charney. J. J.. Stewart. NY. Fluid Mech. Biomech. Amsterdam . Remi ngton.R. (1969). Sci. (1980) Ann. Ph ysiol. Bird. B.. C. N. Buchanan. W . E. and Lewis..J. Goldsmith. 22: 19-156. London. R.Y.. L.R. L.R.” Dek ker.” Oxford University Press. R. 101: 288-295. Eng. Chen. Cooney.. J. Chhabra. (1957) “Kinematic S tudies of Mechanic P roperties of Muscles” In: Tissue E lasticity. W. ASME J. (2002) “T ransport Phenomena. Oxford. Washington.

Hoffman. Phillips. Oka.H.H. Academic Press. R. 29: 19-38. Philadelphia.” Wiley-Interscience. WA. M.” Cambridge University Press. Baltimore.” Williams & W ilkins. Pollack. K.. T. M.. an d O’Rourke.. J. Appl. (1 990) “Biom echanics: Motion.W. Hyun. J. New Y ork. and Bronzino. New York. J. and Kleinstreuer. NY. (Eds.” Prentice Hall. (2004) Clinical Biomechanics. 75:904-915. Fung. Gel s and the Engines of L ife: A New..” Ebner & Sons. New York. Circ. Res. (1948) J. NY. P.” Lea and Febiger." Marcel Dekker. D. Experimental and Clinica l Principles. T.” Taylor & Francis. Flow.. A. (2 002) “Bio logical and Bioenvironmental Heat an d M ass Transfer. S ecomb.. (1997) “Biomechanics: Circulation. M. NJ. 19: 362-369.. D. G. Nichols... S tress.” Wiley-Interscience. (1 998) “ McDonald’s Blo od Flow in Arteries: Theoretical. New York.J. (2001) “The Mechanics of Inhaled Pharmaceutical Aerosols: An Introduction”.F. W. W.” Springer. PA. (1993) “Heat Conduction. 1: 93-122. S. UK. New York.” Marcel Dekker Inc. Greenberg. Upper Saddle River. M. Kleinstreuer. S. (2001) “Cells.. Sperandio. Physiol. Gaehtgens. Karch. FL. (2001) "Num erical Metho ds for Eng ineers & Scientist.. R. and Brown. R. F. Gross. Özisik.. London. Neu mann.D. W. (1998) “Advanced Engineering Mathematics. Moore. New York. A.B. K. N.).” Springer-Verlag. Neum ann.. NY Fung. Middleman. and Ujihira. J.C. C. (1992) Physics of Fluids A. (2001) International Journal of Heat & Mass Transfer. (1 999) Computers in Biology and Medicine. Pries. . (2003) in: “Biomechanics : Principles and Applications. Gessner . (2003) “Biomedical Technology and Devices Handbook”.J. Unifying Approach to Cell Function. M. D. Inc. W.” Schneck. 44: 2247-2260 Kamm. CRC Press. Y. Finlay... NY. P. (1981) “C ardiovascular He morheology. R. and Weinbaum. (2003) “T wo-Phase Flow: Theory an d Applications. Furley. New York.R.. MD. (1989) “Hemodynam ics. New York. S. (1972) “T ransport Phenomena in the Cardiovascular System. R. M. Y.J. an d Gro wth.. Pennes.A. Ar mstrong.238 Biofluid Dynamics Datta. C. Seattle. 58:281-304. H. and Zouridakis. (2002) Annual Review of Fluid Mechanics. NY. an d Schrein er. S. FL. R. Hu. M. (1994). (1999) Microvascular Research. Sakai. 4:30-40. G. 34: 211-232.D. Boca Raton. CRC Press. R. Mabuchi. Milnor. Ota. J . C. C. W . X. Boca Raton.

. Truskey. and Colton. and Bronzino. Jacobs. BME-21: 494... (2003) “Fluid Mechanics. F. and Katz. New York.L.. (1974) IEEE Trans. (2000) Applied Mathematical Modelling.” M cGraw-Hill. Yuan.E. 24:11-25. Eng. S.H. Schmid-Schönbein. Secomb. 37: 799-825. (2004) “Transport Phenomena in Biological Systems. (1 988) PhysicoChemical Hydrodynamics. J. A. Jaffrin. .. Heart Ci rc. F. 10: 77-96. (2000) “The Physics of Pu lsatile Flow. L. . M.Principles and Applications”. F.J. Boston. NJ.. and Lemmous. (Eds.” AIP P ress.K. Zhu. Fluid Sciences.. P .W. (1971) “Schaum's Outline of Theory and P roblems of Advanced Math ematics for Engineers and Sci entists. (1988) J.Y. Physiol.. and Weinberg. (1997). Schneck..Y.. M. MA. Osterloh. G. D. MA. Am. A. Springer . Yang. (1996) “Cellular Bio physics (Vol. A. 110: 74. Spiegel. (1989) “Biothermal. H . Zydney.. W.. T. New York. J.” McGraw-Hill. Cambridge.J. M. C.. D. and Gaehtgens. 1 and 2). New York. Physiol .) (2003) “Biomechanics . Biomech. S.. (1989) Biorheology. 273: H2272-2279 Rosendahl. Biomed. Boca Raton. Eng.L. M. White. M. NY.W. Principles and Applications. W.” Pearson Prentice Hall.R. M.D. and Sutton. D.. Wulff. Lee.W. A. 26: 215-227. FL. CRC Press. Shapiro.. D.. Zamir. Weinbamn. ” T he MIT Press.” Hemisphere.Chapter II: Biofluid Dynamics Concepts 239 Pries. (1969) J Flu id Mech. S. K.R.F. Sperandio. Upper Saddle River. Weiss T. G.

1. Specifically. key factors in aneurysm formation.1). after decades of detrimental biochemical processes either “plaque” formation (i.g.1.1 VESSEL OCCLUSIONS Vessel occlusion often occurs due to rapid thrombus formation in the vicinity of significant intimal thickening. Biofluid dynamics applications at different stages of an arterial disease process may include: (i) the simulation of transport and deposition of cells and biochemicals in blood vessels as well as abnormal stress-strain fields in the vessel walls. This chapter describes causes of partial or complete vessel occlusion. due to atherosclerosis.. as reviewed by Longest (2004). causing the disease. called stenosis. 3.2). 3. 3. end-to-side bypass grafts (Sects. hyperplasia and/or thrombosis (Sect.2. (ii) an analysis of disease evidence in terms of atherosclerotic plaque formation or vessel dilation. Chapter III ANALYSES OF ARTERIAL DISEASES Of special interest in biofluid dynamics are arterial diseases which originate inside the arterial wall and weaken the vessel wall. as well as treatment options. possibly leading to rupture. b). 3.2.3).2 and 3. or the locally weakened arterial wall dilates and forms a balloon-like vessel section (aneurysm).1a. which progressed over decades (see Figs. and (iii) the evaluation of remedies. generating locally geometric changes. focusing on the abdominal aorta (Sect. changing the local geometric configuration.. atheroma protruding into the lumen) greatly reduces blood flow.e. optimal design and placement of implants such as bypass grafts or stents.1. Davies (1994) describes two distinct patterns by which occlusive 241 .g.grafts (Sects. For example. 3..5) and stent.4 and 3. e. 3. e.

it is important to know the outcome of a ruptured plaque cap. only stenosis severity is used. Clearly. ranging from occlusive thrombosis to a healed fissure with no increase in stenosis. Nevertheless. A nice multicomponent analysis of human carotid plaque is given by Tang et al. 1997). continue to be major reasons for mortality and morbidity in the Western World. It has also been speculated that a factor contributing to the vulnerability of the plaque is infiltration of the cap tissue with macrophages (Salunke & Topoleski.1 Atherosclerotic Plaque Formation As is well known. 1997). This form of occlusive thrombosis typically occurs near a high-grade stenosis and is likely the result of endothelial denudation. hypertension.3). Clinically. ulceration.e. In contrast. If a model of occlusion risk were available (see Sect. Such a structure will lead to a concentration of circumferential wall stress on the plaque cap in systole because of the inability of the pool to carry a load that is redistributed to the vessel wall. i. definitive conditions for plaque-cap rupture have not been established. three quarters of occlusive thrombus are related to plaques that have undergone a deeper injury. plaque tearing or disruption. accelerated by environmental factors. for coronary arteries.1. aneurysms. Approximately one quarter of observed thrombi are superimposed on stenotic plaques. the basic disease mechanisms relate to abnormal or excessive cellular behavior: . a variety of outcomes are possible once the plaque has been disrupted. and stroke. a large amount of collagen is exposed to the blood providing a major stimulus for thrombi formation. to determine the necessity for surgical intervention (Wootton & Ku. cardiovascular diseases. et al. possibly advances to a degree that fills the remaining lumen.242 Biofluid Dynamics thrombosis may occur. or rupture. very often inappropriately. leading to tissue ischemia and other major complications. The thrombus itself forms initially within the interior of the plaque. 3.. Currently. Once the plaque cap has ruptured. where the plaque itself does not undergo dramatic changes. it could be combined with a model for plaque-rupture risk to decide which patients are good candidates for surgical treatment and which patients can be managed medically. it was observed that a majority of plaques that rupture and result in occlusive thrombosis are eccentrically situated and have a lipid pool that does not have an internal lattice of collagen supporting the cap (Richardson. Its mechanical properties can be classified as nonlinear and inelastic (see Salunke & Topoleski. Although the causes are of genetic origin. (2004). Atherosclerotic plaque exhibits a wide range of material properties that are associated with its multiple components and variable composition. called fissuring. ischemia. 3.. 1989). or is sheared off and suddenly fully occludes a smaller blood vessel downstream. thrombosis. Due to a wide range of plaque-cap structures and the long-term cyclic loading of the hemodynamic forces. typically extending into the core of the plaque. such as atherosclerosis. 1999).

aggregation.1 Two examples of vessel occlusion. inflammatory.e. Truskey et al. 2002.. stenotic development Artery wall Endothelium Lumen •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ opening •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ Thrombus •ٛ •ٛ •ٛ •ٛ •ٛ Plaque •ٛ b) Partially occluded artery: atherosclerotic plaque plus thrombus Figure 3. degratory molecules. • the production of vaso-active. Plaque •ٛ•ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ •ٛ Ulceration •ٛ low •ٛ •ٛ Endothelium df •ٛ •ٛ oo Lumen •ٛ Bl •ٛ •ٛ or vessel •ٛ •ٛ •ٛ•ٛ a) Schematic of “intimal thickening.Chapter III: Analyses of Arterial Diseases 243 • cell migration. 2004. among others). replication.” i. .. growth-regulatory. and rupture. and • the synthesis and organization of constituents of the extra cellular matrix (see Humphrey.1.

the mechanics of the tissue (e. the slow process of vessel occlusion has two phases. these stenotic developments may be the result of late- stage arteriosclerosis lesions as discussed. or a thrombus (i. endothelial cell dysfunction and elevated permeability of the endothelium. over long periods of time intimal thickening creates a progressive reduction of the cross sectional area at critical locations in medium to large arteries. or due to a more rapid response to injury of the inner arterial wall. coronary arteries located on the heart surface develop atherosclerosis.e. 3. What follows is that monocytes gobble up oxidized LDL.” i.g. Note. a process called atherosclerosis. complex aggravating blood flow patterns in curved or bifurcating.244 Biofluid Dynamics 3. In general.. medium-to-large arteries. Alternately. Initiation of atherosclerosis involves monocyte migration through a dysfunctional endothelium (Ross. while the intramural ones.. high blood pressure. the hardening of arteries.. Sub-endothelial migration of monocytes may lead .1. i. both of which involve smooth muscle cell proliferation and lipid accumulation within the intima. i. which is referred to as hyperplasia. muscle.. Furthermore. This second phase leads to the acute ischemic syndromes of myocardial infraction. The same observation holds for vessels surrounded by the vertebrae.e. which may result from atherosclerotic plaque or intimal hyperplasia. Specifically.1. and much more frequently.e. atherosclerosis is a common form of the more encompassing arteriosclerosis. unstable angina.) surrounding a blood vessel may influence atherosclerotic plaque formation. For example. Advanced intimal proliferation lesions of atherosclerosis may occur by at least two pathways as described by Ross (1986): Monocytes attach to the endothelium which may continue to secrete growth factors. bone. a working model of atherogenesis has been developed. and cerebral infarction. sudden ischemic death. thrombus formation superimposed on a stenotic development.2 A Pathway Model for Atherogenesis Based on cell culture and in vivo studies over the past decade. etc. accounts for the ultimate mode of gradual or sudden vessel occlusion. 1999). This primary process of moderate lesion development may occasionally cumulate in vessel occlusion. and critical blood particle accumulation are the precursors of focal cell proliferation inside the arterial wall..e.1. turning into macrophages and leading to smooth muscle cell proliferation inside the arterial wall – labeled “intimal thickening” or stenotic development (see Fig. those penetrating the heart muscle. In any case.1).1. do not. As mentioned. or tissue overgrowth.1 A Particle-Hemodynamics Model Many of these abnormal cellular functions may be triggered by locally “disturbed flow.1. very low or very high wall shear stresses. allowing for migration of low-density lipoprotein (LDL) and monocytes (WBC family) into the arterial wall. The first stage is stenotic development. 3. blood clot) arriving from an upstream vessel location.

and stented vessels. For example. in similar sized plaques those developing due to hyperplasia tend to have a higher concentration of smooth muscle cells and a lower concentration of lipid accumulation than do atherosclerotic lesions which usually appear in low-shear regions (see Caro et al. Hyperplasia is often viewed as an accelerated form of atherosclerosis due to the similarities in the lesions. which occurs at a rate proportional to the degree of the injury (Chervu and Moore.. providing three possible sources of growth factors. However. Ross (1986) advanced the “response-to-injury” hypothesis to propose that intimal hyperplasia may be an early lesion on the pathway to atherosclerotic plaque. hemodynamic factors may be a primary or contributing source of endothelial dysfunction.e. platelets. Endothelial injury or dysfunction often occurs at sites of bypass graft anastomoses. Other sources concurred with and expanded on the observations of Fry (1968) by stating that endothelial cell modification can be induced by severe hemodynamic conditions or by surgical processes such as the creation of an anastomosis. At such locations. 1990). 1990). Some of the smooth muscle cells in the proliferative lesion itself may form and secrete growth factors as well. macrophages. Fatty streaks may become directly converted to fibrous plaques through the release of growth factors from macrophages or endothelial cells or both. 1971). and endothelial cells. As the atherosclerotic lesion progresses. Subsequent studies have shown that such injury produces smooth muscle cell proliferation. typically associated with local turbulence. (1983) were first to clearly demonstrate that acute injury to the intima and media can produce hyperplasia. The attached platelets play a major role in plaque growth by stimulating smooth muscle cell migration and proliferation even though they may not significantly influence initiation. macrophages may lose their endothelial cover and platelet attachment may occur.. and this is the stage in which a true thrombogenic dimension is involved in atherosclerosis. locations of balloon angioplasty.2 Intimal Hyperplasia Development Intimal hyperplasia is the rapid abnormal continued proliferation and overgrowth of smooth muscle cells in response to endothelial injury or dysfunction (Chervu and Moore. Platelets then adhere to the exposed connective-tissue matrix underlying the endothelial defects. there are gaps in the endothelial surface within which lipid-filled monocytes are impacted. then.1. The injury model. 3. i. Macrophages may also stimulate or injure the overlying endothelium. In some cases. does not require that . The work of Fry (1968) was the first to clearly demonstrate that acute endothelial changes could occur from extremely high wall shear stress. The growth of intimal hyperplasia may or may not subside when the endothelial layer is re-established.Chapter III: Analyses of Arterial Diseases 245 to fatty-streak formation and release of growth factors. and the presence of foam cells in peripheral blood suggests that some are emigrating from the intima. Clowes et al.

resulting in smooth muscle cell proliferation.. there is a limit to this relationship. i. especially in the presence of a tissue factor. called a thrombus. Some of these colliding platelets may adhere to the vessel wall. As mentioned. Platelet deposition at dysfunctional sites increases with increasing shear rate. the clotting system which prevents blood loss in the event of vessel injury.2. Activation also permits the aggregation of platelets. and flow reattachment sites are all areas of high platelet-wall interactions. Platelets do not normally adhere to a healthy endothelium. by other proteins in the blood or inside the platelets. flow stagnation points.1. Arterial thrombosis may also develop on artificial surfaces such as vascular grafts. an arterial thrombus is usually found superimposed on an atherosclerotic plaque which has fissured (or ruptured) to expose subendothelium plaque components to the blood. A general view of the currently understood mechanisms of arterial thrombosis is provided in Fig. Thrombi formed in the low shear environments. called membrane glycoproteins (GPs). The observable difference in thrombus composition due to environment is an indication that shear stress plays a primary role in thrombus formation. 1994). with some fibrin and trapped red blood cells. 3.. i. however. as found in high shear regions. or may . Platelets and other critical blood elements that adhere to the dysfunctional endothelium may be a major factor in stimulating this smooth muscle cell proliferation. As discussed the adhesion of a small number of platelet layers may significantly contribute to smooth muscle cell proliferation and intimal thickening. In normal flow. a chemical mediator present when the endothelium is injured. which may form a platelet plug that occludes the vessel. is composed primarily of platelets. typical of the venous system. The converging flow field around a stenosis.e.246 Biofluid Dynamics endothelial cells be denuded. heart valves. platelets collide with the vessel wall due to both convective transport and the diffusive motion imposed by collisions with red blood cells. are composed mostly of red blood cells which are held together by a fibrin mesh.1. 3. as well as in aneurysms or injured arteries. which are typically found distal to the platelet- rich part of the thrombus. such as a fibrin tail. and by a network of polymerized plasma protein called fibrin (Colman et al. but that they be greatly disturbed by adverse hemodynamic conditions or a surgical procedure. and stents. inside an artery or vein. regulated by membrane glycoproteins and proteins in the vessel wall. The platelet may be activated which makes adhesion. An arterial thrombus.3 Thrombogenesis Thrombosis is the formation of a blood clot.e. more permanent.. A thrombus is primarily composed of platelets and red blood cells bound together by molecules in the cell membrane of the platelets. Thrombosis develops by the same mechanisms that control hemostasis. Low shear regions in which these thrombi are found include regions of flow stasis and recirculation which commonly occurs within branching blood vessels.

thus. such as reduction by embolization. 2004).Chapter III: Analyses of Arterial Diseases 247 embolize (break free) totally or in part.1.2 Simplified schematic of arterial thrombosis (Longest. a positive feedback mechanism is established.4 Particle-Hemodynamics In this section.3 and 1. The computational methods used to solve these equations are discussed by . are poorly understood. Simultaneously. Adhesion • Platelets adhere to proteins on the surface via GPs • Triggered when a thrombogenic surface is exposed to blood Coagulation Cascade • Predominately driven by thrombin Activation Aggregation • Platelet functions triggered by • Leads to formation of platelet plug chemical or physical stimuli • Incoming platelets may be • Chemical stimuli: ADP and chemically activated by passing thrombin through an agonist cloud produced • Physical stimuli: shear stress by the adherent thrombus • Other stimuli: biomaterials • Un-activated platelets may be • Activation produces positive bound and then activated feedback for activation of other platelets Figure 3. 1. 3. thrombin is produced at the injury site due to tissue-factor (TF) exposure. Thrombin causes coagulation of fibrin into a polymer network that may reinforce a platelet plug and incorporate red blood cells into the clot.1. the equations governing Eulerian-Lagrangian particle- hemodynamics analysis are re-presented (see Sects. Healthy endothelial cells on the other hand trigger a negative feedback mechanism that limits the propagation of the clot to the region of injury.4). accelerating the aggregation process. Other mechanisms which control thrombus growth rates. During early hemostasis more thrombin is released as platelets are activated. reducing the sustained rate of clotting and breaking down the clot.

A mathematical point-force model for blood particle motion that is valid in both far.100 s-1.and near-wall regions is employed. (3.1) and momentum: r ∂v ∂t r r 1 + (v ⋅ ∇ )v = ρ rr − ∇p + ∇ ⋅ τ ( ) (3.3) 2D rr where D is the strain rate tensor such that rr r r T 2 D = ∇ v + (∇v ) (3.. η.047723 s-1.2) is the additional shear-rate modifier. used in Eq.2c) is computed from the second scalar invariant of the rate of deformation tensor. and the shear stress tensor is given as: rr τ = η (γ& ) [∇v + (∇v) T ] r r (3.248 Biofluid Dynamics Longest et al. Normal blood density has been assumed to be ρ = 1. and λ = 0.1.2c) using the methodology outlined in Buchanan et al.1. Using the above constants. The local shear rate. (2004).2a) r where v is the velocity vector. τ 0 = 0.1.. is calculated with the transformed Quemada model of Buchanan and Kleinstreuer (1998): 2  τ 0  η (γ& ) =  η ∞ + (3. i.054 g/cm3. γ& ~ 10 −2 s −1 ). that allows for a better blood data fit in low ranges of the shear rate (i.4. (2000). γ& 2 = II rr (3.1. η ∞ = 0.1. 2002): rr II rr 2D = 1 2 ( )  I rr  2 D 2 − trace  2 D 2    (3.1. at a shear rate of 4.. the popular Casson model (see Sect. The flow field equations describing transient laminar incompressible hemodynamics are the conservation of mass: r ∇⋅v=0 (3. Fluid flow equations and boundary conditions.2c)   λ + γ&  The advantage of the Quemada model over.1.1.. γ& . the Quemada model predicts that the absolute viscosity is within 5% of the Newtonian limit. i.2b) The shear-rate-dependent absolute viscosity. ρ is the fluid density. Coefficients used in the original form of the Quemada model for a normal hematocrit (H) of 40% were taken from Cokelet (1987).07687 dyn·cm-2. (3. The coefficients were transformed for Eq. for example.0309 dyn·s·cm-2. λ.5a) .1.e. p is the pressure.4) The second scalar invariant can be expressed as (Bird et al. resulting in the empirical constants η ∞ = 0.e.0309 dyn·s·cm-2.e.1.

inlet and boundary conditions. In general. is the primary application site for most researchers. 1991). An extended inlet of approximately three diameters was added to allow non-Newtonian aspects of the velocity profile to develop fully.1.. and (b) facilitate numerical computation. Following the work of Longest and Kleinstreuer (2003). 1998) and assumed a second “exercise” pulse for patients under active conditions. However.6) A femoral bypass. fully developed flow profiles with no significant radial velocity component. waveforms that vary significantly from the standard femoral pulse are typically associated with both early and late graft failure (Inokuchi et al.g. the inclusion of a bypass graft significantly alters waveform characteristics compared to measurements made in natural arteries. in addition to the “no-slip” wall condition.1. 1998). For multiple outlets. (1993) made Doppler ultrasonography measurements of a mid-femoral flow waveform which are summarized in Table 3. The transient velocity profiles were then computed using the Womersley solution for Newtonian flow. constant mass flow rate boundaries and flow division ratios were specified.5b)   Expanding the shear rate. Furthermore. a pressure boundary patch was typically used.e. Nichols and O’Rourke. 1986. a transient Womersley solution was used (see Womersley. (1996) employed a “resting pulse” derived from femoral flow velocity measurements made in a dog (cf. Flow field outlets were extended far downstream such that the velocity was normal to the outlet plane. resting or exercising.1. and conditions under which the measurements are made. The various combinations of femoral diameter and flow rate can induce a wide range of Reynolds number waveforms. location of the flow measurement. Steinman et al. were selected in a manner to: (a) match physiological conditions as closely as possible with available data. 1955 or Nichols and O’Rourke. it can be expressed as γ& 2 = | 4 D11 D22 + 4 D11 D33 + 4 D22 D33 + 4 D12 D21 + 4 D13 D31 + 4 D23 D32 | (3. Kleinstreuer et al. For example. Flow rates through the femoral artery vary largely. 1982. For single outlet conditions..Chapter III: Analyses of Arterial Diseases 249 where the first scalar invariant is given by the trace of the rate-of- deformation tensor rr I 2 Drr = trace  2 D  (3.. To generate the time varying inlet velocity profiles. depending on the quality of an individual’s circulation system. .1. clinical evidence indicates that pulse severity alone does not necessarily indicate a predisposition for graft failure. 2000). Flow rate information over variable time-steps was used to compute a complex Fourier series approximation of the pressure gradient pulse (Buchanan. Significant stenosis developments may lead to occlusions and a general loss of wall compliance.. i. as considered here. e. Okadome et al.

164.1 Reynolds Number Waveforms for the Femoral Artery.8 75.6 -0.054 g/cm3.2) has been selected as a representative basis for comparison and is illustrated in Fig.7 33.1. Table 3.2). Consistent with Type I waveforms.8 113.031 dyn*s/cm2 and ρ = 1.1.4 ml/hr Figure 3. To formulate a representative baseline pulse for numerical investigation of the femoro-popliteal bypass.6 -0.6 702.3 Selected femoral artery waveform (Type I) for numerical analysis.2 conditions ml/hr cm/s cm/s cm/s (Ø = 6 mm) Graft 164.45 18.2 Characteristics of the Selected Type I Waveform.5 60 Exercise pulse 480 295 356.1.95 527.4 303. This pulse is well within the range of measured mean flow-rates and falls within peak velocity and Reynolds numbers guidelines for non- stenosed grafts (cf. Therefore.4 151.1. -5.9 5.1 conditions ml/hr cm/s cm/s cm/s (Ø = 8 mm) Note: All variables based on the Newtonian limiting viscosity of blood η = 0.1.250 Biofluid Dynamics Table 3.3.0 60 (Steinman et al..5 100 Standard pulse 980 -375 125. a waveform should be selected due to frequency of clinical occurrence and the absence of extraneous complications. only a .9 9.1.9 21. Remax Remin Remean f (beats/min) Resting pluse 600 -67.7 -3. 3. a Type I pulse (see Table 3.9 1054. 1993) Table 3.4 conditions ml/hr cm/s cm/s cm/s (Ø = 4 mm) Graft 164. -2. Qmean Remean Remax Remin Vmean Vmax Vmin Arterial 164.4 202.9 -0.

1987): dv i Du i m f d mp = mf − ( v i − u i ) − 6πa p µ ( v i − u i ) + (m p − m f ) g dt Dt 2 dt term 3 term 4 term 1 term 2 6πa 2p µ  d dσ v i (0) − u i (0)  d  a 2 ∇ 2 ui   t − πv  ∫ 0 dσ ( v i − ui) t −σ + t  + mf dt   20      term 5 term 6 π  d ∇ 2 ui ( ) dσ ∇ 2 u i (0)  ∫ t + πa 3p µ∇ 2 u i + a 4p µ term 7 v  0 dσ t −σ t   term 8 (3. i. which is referred to as a point-force approximation.1. In expanded form then.1 Equations of Particle Motion Maxey and Riley (1983) have presented in detail the equation of motion for a small rigid sphere in a nonuniform flow field with creeping relative motion.7) dt S where v i is a component of the particle velocity. which occurs around t/T = 0. the surface integral is over the sphere.8) ∂x ∂x   j i  This derivation makes clear the fact that contributions to the particle motion equation include forces from both (a) the undisturbed flow field (although it may be nonuniform). n is the unit outward normal. 3.9) where v i is the particle velocity.1. spherical particle subjected to a nonuniform transient three-dimensional flow field of a Newtonian fluid reads (Maxey. and a p is the particle radius. and (b) the local disturbance flow which is created by the presence of the sphere. u i is the fluid velocity.1.5s.4. the equation of motion for a small. This equation requires that the particle Reynolds number approaches zero .e. and the fluid stress tensor is given by ∂u  ∂u  σ ij = − p δ ij + µ  i +  j (3.Chapter III: Analyses of Arterial Diseases 251 small amount of net retrograde flow is observed. They began with the equation of motion for a spherical particle as dv i mp = m p g i + ∫ σ ij nˆ j dS (3.1. The resulting velocity profiles display a significant amount of retrograde flow in the near-wall region throughout diastole. These forces are applied to the centroid of a discrete particle. Stokes flow. rigid..

e. where T is the pulse period and τ p is the particle’s relaxation time (or momentum response time). Michaelides and co- workers have applied Laplace transforms to arrive at an analytic expression of particle motion that is explicit (cf. A small particle Reynolds number allows for the application of Stokes drag. and implies a small difference between the particle and material derivatives d/dt and D/Dt. Terms six through eight are the Faxen corrections to the added mass.1.e.983 for platelets and α = 0.10b) L where L is a characteristic length of the flow field.1. Analytic solutions to the equation of particle motion (including the history term) are available by several techniques (Clift et al. the drag.1.252 Biofluid Dynamics 2 | ui − v i | a p Re p = << 1 (3. however. Therefore. following Coimbra and Rangel (1998). The second term in the parentheses of the fifth component accounts for the effect of an initial relative velocity. the history term is only important for values of D ≡ T / τ p < 100. and the history integral. which accounts for both shear and pressure contributions in a uniform linear non-accelerating flow field. The second term is the added mass. which is a transient term resulting from the linear acceleration of the fluid surrounding a particle. i.964 for monocytes. they do not account for or rely on particle rotation. i. Lift forces are assumed negligible due to the absence of particle rotation and the assumption of a small shear Reynolds number (2a p ) 2 dU Re g = << 1 (3.. term 3..9) is the result of pressure gradients that exists in the undisturbed non-uniform flow. in plasma α = 0.e. Buchanan.1. D ≈ 105 to 106. The analysis of Buchanan (2000). Michaelides.12) 9µ c For blood flow in humans. 2 ρ p a 2p τp = (3. the .10a) v and the particle radius is sufficiently small ap << 1 (3. The fifth term is the history integral.1.11) vc dy The first term in Eq. 1978. which is a transient consequence resulting from the delay in near-particle viscous effects as the relative velocity changes in a linear uniform flow field. respectively.. 1997). i.. the pulse period is T≈ 1 s and density ratios are very near unity. such as swirl and curved particle paths. Blood particle relaxation times are on the order of τ p = 10-6 s resulting in a very large value of D. (3. These addenda account for the effects of nonuniform flow. For example. 2000). or Basset term. indicated that for a density ratio α = ρ f / ρ p < 8/5.

1. To provide an accurate model for lateral motion.9) is the Basset- Boussinesq-Oseen (BBO) equation which features the first five terms of Eq.1. Furthermore.1.13) dt Dt τ p where v i and u i are the components of the particle and local fluid element velocity. the movement induced by this mixing results in a persistent dispersion. and deformations of. the above equation can be written dv i 3 Du i β = βα + (u i − v i ) (3. (1998) and Kleinstreuer (2003).1. the resulting equation for blood particle motion. Finally. For platelets and white blood cells. If virtual mass effects are to be included.693 (3. the red blood cells in flowing blood induces a form of mixing. among others.Chapter III: Analyses of Arterial Diseases 253 history term may be neglected for the transport of platelets. (3. can be modeled as a fluid element (cf. and is often modeled with effective diffusion coefficients which are two to three orders of magnitude greater than those due to Brownian motion. local red blood cell concentration. both the drag and pressure terms make a significant contribution to the equation of particle motion due to variable flow fields. assuming only pressure and drag effects. models for a realistic red blood cell concentration field are rather complex. A reduced form of the Maxey equation (3. some of the larger proteins found in the blood to interact with the wall.1. . Longest & Kleinstreuer. respectively. and cell deformation. therefore. local variations in hematocrit will be ignored in defining particle dispersion coefficients. similar on a macroscopic scale to the intermolecular collisions that result in Brownian motion. However. the continued collisions and hydrodynamic interaction between. possibly. as in the annular expansion of Karino and Goldsmith.9). particle dispersion coefficients should be a function of the local velocity gradient. Furthermore. is written as: dv i Du i 1 =α + (u i − v i ) (3. This motion becomes a major mechanism for permitting cells and. and red blood cells in plasma. Buchanan (2000) pointed out that the added mass term makes no significant contribution for liquid-solid multiphase flows or colloidal suspensions (such as blood) where the densities of the phases are very similar.14a) dt 2 Dt τ p where 1 β= := 0. monocytes. Unfortunately.14b) 1+α / 2 It has been shown that the motion of a hardened red blood cell. 2003). as discussed by Crowe et al.

making it impossible for a particle to reach the . δ = hp − a p (3. the first term is the traditional Stokes drag while the second accounts for the drag modification due to the presence of the wall.16) δ where d is the distance between the particle surface and the wall. It is this second term that is often referred to as the lubrication force and can be viewed as an interaction force as derived by Crowe et al. In the near-wall environment. (3. the lubrication coefficient may vary from zero to infinity in the limit.. at which a no-slip condition holds.13) cannot accurately predict the wall-normal motion. The Stokes drag term in Eq. the enhanced shear field and the particle rotation may influence the particle’s trajectory by generating lift and modifying the drag term. When a particle is close to a solid boundary. Consider a particle moving near a wall with both a normal and tangential velocity component with respect to the boundary.1. Eq.1.17) For truly Stokesian flow.1. as well as lift and biochemical interactions.15) τp τp where n is the wall-normal direction and h p is the distance between the center of the particle and the wall. the shear forces and the rotation rate of the particle increase. but to a lesser extent.. The near-wall forces to be considered for bio-particle transport include near-wall drag modifications (i. However. i.1. In the above representation.1. the boundary influences the viscous drag on a particle moving tangential to a wall.e. as a particle approaches a fluid boundary. Similarly. These assumptions are most valid for fluid-particle flow in a boundless media.2 Near-Wall Forces Assumptions required by the equation of particle motion.4. Cox and Brenner (1967) showed that for Stokes flow. Hence.1. the lubrication forces). Specifically. the Stokes drag approximation must be modified. the form drag component in the Stokes drag approximation assumes that there is not a boundary in the direction of particle motion acting to increase the pressure in front of the particle. Dahneke (1974) suggested the following fit to the results of Cox and Brenner (1967) for the lubrication coefficient at all particle wall separations: ap λn = (3. the drag on a spherical particle of radius a p moving normal to a wall is 1 1 Fn = − m p (v n − u n ) − m p (v n − u n ) λn (3. (1998) for the case of two approaching spheres. include a relatively small shear Reynolds number and negligible particle rotation. (3.254 Biofluid Dynamics 3.13) should be modified to include the effects of the wall on the trajectory.e. as presented above.

it only diverges logarithmically. (1967a & b) derived approximations for λt and illustrated that it was a function of shear rate. In a strongly sheared flow.18) τp τp Goldman et al. 3. angular velocity. In the context of Eqs. (1967a & b) showed that the drag force on a sphere near a wall in linear (uniform shear) flow is given by 1 1 Ft = − m p (v t − u t ) − m p ( v t − u t ) λt (3. There is a lack of rigorous mathematical justification for using the available results for lift forces (McLaughlin.1..18. Furthermore.g.7 (3. inertial lift forces are likely to be important. However.Chapter III: Analyses of Arterial Diseases 255 wall in a finite amount of time.19b) hp Loth (2000) reports that these approximations provide a solution that is within 2% of simulated results. however. There are two causes of a composite lateral lift force on a spherical particle. Recently. which were derived as asymptotic solutions in the limit of true Stokes flow.1 λn = (3.9). 1991). As the particle approaches a wall.1. the drag coefficient in the tangential direction is also affected by the presence of the wall. Similar to wall-normal motion. all the existing theories of lift forces at small but finite particle Reynolds numbers are derived for steady laminar flows. unlike λt . for particles moving through a gas. Hence. (3. λt diverges. and the dimensionless ratio h p / a p .1. the continuum model overpredicts the magnitude of the stress on the particle surface in the gap. Young and Hanratty. have proven to be reasonable estimates in comparison with experimental data (e. These enter the BBO-equation at a higher order in the particle Reynolds number than the terms in Eq.1. the approximations are expressed as 1. The near-wall drag modifications given above.1.15 & 3.1. 1994). the continuum approximation breaks down when the gap between the surface of the particle and the wall becomes comparable with the molecular mean free path of the gas. Goldman et al.19a)  hp   − 1  ap    and ap λt = 0. Loth (2000) has reported near-wall drag modifications based on resolved-volume simulations of spherical particles for Rep << 1. The first is known as the Magnus lift force and develops due to .

8081 + 0.1. the velocity scale γ& a p should replace us in the above . The lift is created by a pressure differential between opposite lateral sides of the particle resulting from a velocity differential due to rotation.20) a   p us  where us is taken to be the wall-tangent slip velocity u s = (v t − ut ) (3. and has been corrected via an empirical correlation to allow for higher relative velocities (Mei. the Cherukat and McLaughlin (1994) lift approximation can be written as  h p γ& a p  Flift = ρ f a 2p u s2 ⋅  . Kleinstreuer. inter-particle collisions. 2003)..6760 − 0.7631 + 0.24139  − + 2.1837κ 2 + 0. Assuming that the wall lies within the inner region of the sphere’s disturbance flow. Shear induced torque may be calculated directly if detailed knowledge of the particle’s influence on the flow field is known. which induces the Saffman lateral lift force. For cases in which the slip velocity approaches zero.9009κ 2 + 0.8248κ − 0.3561κ − 1.21) Assuming the particle to be a freely rotating sphere.1. 1997). Cherukat and McLaughlin (1994) numerically integrated their asymptotic result and found it could be approximated as [ I = 1.256 Biofluid Dynamics rotation of the particle. or due to the enhanced near-wall shear field.1. The second source of lift is the shear field of the fluid. which may cause lateral lift on the sphere even in the absence of rotation.845163κ 3 ]  3.4616κ 2  Λ (3. which must be different from the local fluid element rotation rate.  (3. the Magnus force is usually neglected (Michaelides. particularly in the presence of a moderate number of collisions. The sphere may rotate because of collisions with the wall.98149κ 3 Λ2 ] where κ = h p / a p and Λ = γ& a / u s . Cherukat and McLaughlin (1994) derived an expression for Saffman- style lift which is applicable when the distance between the particle and wall is on the order of the particle radius. Calculation of the Magnus effect requires direct knowledge of the particle rotation rate.22)  κ  [ + 1. The basic Saffman lift force is relatively easy to compute (Crowe et al. Crowe et al. 1998. 1994). Collision- induced torques further complicate the calculation. Due to the difficulties associated with the calculation of particle spin. A velocity gradient may give rise to sufficiently high and low pressures on opposite sides of a particle. otherwise no lift is produced (cf.8796κ − 1.. 1998). Calculation of the rotation rate requires solving the angular momentum equation.

1. based on hemodynamic wall interactions.Chapter III: Analyses of Arterial Diseases 257 equations. 3. as well as the size of k. (3. the near-wall lift expression is valid on the order of ten monocyte diameters. Assuming that Eq. the near-wall lift expression of Cherukat and McLaughlin (1994) is no longer valid.1.23a) where the Stokes length scale is ν LS = (3. . for the case of blood particle deposition in bifurcating geometries. Hemodynamic wall parameters are intended to identify sites where intimal thickening and thrombosis formation are likely. each valid in a band of wall separation distance.1. Lift expressions valid outside of the narrow near-wall region require significant axial lengths before a contribution to radial particle concentration is realized. the near-wall lift may be directed toward or away from the wall. Wang et al. LG } (3. Hence.3 Hemodynamic Wall Parameters As discussed in Sects. LG is typically the limiting condition due to elevated shear fields and small particle response times.23a) implies that hp should be one order of magnitude less than LG . Depending on the signs of γ& and us. as a particle approaches the wall in the limit. the lift force inevitably acts to separate the particle from the wall. insignificant particle lift will be assumed. However. In the context of bifurcating geometries. this is a necessary assumption owing to the fact that outside of the near-wall region collisional forces dominate lateral particle motion.23c) γ& For blood particle simulation. Beyond this region. For particle wall distances significantly greater than the particle radius. Application of the Cherukat & McLaughlin (1994) lift expression is based on the condition a p < h p << min {Ls . (1997) suggested using a compilation of the available lift expressions.1.1. transverse motion on the order of a particle diameter becomes important with respect to particle deposition. However. near-wall lift is the most critical.2.1.1. endothelial cell injury or dysfunction and the interaction of critical blood particles with the vascular surface both play significant roles in the biophysical evolution of intimal thickening and thrombosis formation. To account for lift in wall bounded flow at all particle wall separations. the region of interest is of insufficient length for significant radial motion away from the wall to occur as the result of lift.4. 3. in the near-wall region.23b) us and the Saffman length scale is ν LG = (3. Moreover.1 and 3.

. when subjected to as much as a tenfold increase in blood flow. • Enhanced wall permeability gradient. Aggravating Flow Events • Flow separation/reattachment • Low oscillatory wall shear • Vortical flows • Stagnation point flows • High shear stress regions • High pressure flows • Long particle residence times Abnormal Biological Processes Indicator Functions • Endothelial cell dysfunction • Wall shear stress (WSS) • Injury of endothelium • Variations of WSS. angle deviation. the response .4 Flow events.e. • Wall influx of LDL and monocytes • Normal pressure gradient • Aggregation/deposition of platelets.1. Zarins et al. Furthermore. shear index. gradually increased in diameter until a mean WSS of 15 dyn/cm2 was restored. (1987) demonstrated in vivo that the arterial wall.. 3. SMC proliferation • Wall particle density Blood Vessel Diseases Minimization of suitable Indicator Functions • Atherosclerosis • Hyperplasia Virtual Prototyping and Design • Thrombosis Recommendations Figure 3. 2001). Wall shear stress. etc. • Particle deposition pattern fibrin. i. For example. biological processes. 2001). Hemodynamic wall parameters that are to be considered for (femoral) graft bypass and graft-end optimization are summarized in Fig.1.258 Biofluid Dynamics Traditionally. identifying sites of endothelial cell dysfunction has been a primary purpose of the wall shear stress (WSS) based hemodynamic parameters as well as sites of excessive particle-wall interactions (Kleinstreuer et al.4. and methodology governing improved graft-end designs (Kleinstreuer. It is widely held that the arterial wall is capable of changing its diameter in response to changes in WSS.

1. 1985). 1996)  τ w dt  T r 1  ∫  OSI = 1 − 0 r (3.g.Chapter III: Analyses of Arterial Diseases 259 of endothelial cells to the time-averaged direction and magnitude of WSS in vitro has been well documented (e. The shear stress is a tensor within the flow field and reduces to a vector on a surface r rr τ w = nˆ ⋅ τ (3.temporal mean wall shear stress direction • n -tangential to the surface and normal to m .. respectively.24c) The oscillatory shear index monitors differences in these two time- averaged values. The numerator of the shear stress fraction represents the magnitude of the time-averaged WSS while the denominator represents the time-average of the WSS magnitude...1. The shear stress fraction can vary from 1 to 0 which indicates no cyclic variation to 180-degree cyclic variation of the wall shear stress direction with time. quantifies disturbed flow interaction with the wall and is formulated as (He and Ku. Cyclic departure of the wall shear stress vector from its predominant axial alignment indicates flow disruption over time and is known as the oscillatory shear index. respectively. From a biological standpoint. the time-average of the wall shear stress magnitude can be written r1 T r τw = T 0 ∫ τ w dt (3. endothelial cells have been shown to align themselves with the mean flow direction which corresponds to the local direction of the time-averaged wall shear stress.1.24b) Alternatively.1. indicating the least and most severe temporal shear rate conditions. The magnitude of the time- averaged wall shear stress vector can be written as 1 T WSS = τ w = T 0 τ w dt ∫ (3. 1991). The OSI. The OSI can vary from 0 to 1/2. Surface coordinates that elucidate the interaction of instantaneous wall shear stress vectors and endothelial cells are defined as: • m . therefore.24a) where nˆ is the local surface normal vector. Wall shear stress gradient.25) 2 ∫ τ w dt  T   0  r where τ w is the instantaneous wall shear stress vector. Helmlinger et al. Oscillatory shear index. or OSI (Ku et al.

surface normal direction The wall shear stress gradient can be obtained by calculating spatial r derivatives of the wall shear stress vector ( τ w. x iˆ + τ w. which results in a nine component tensor ∂ ∇τ w =  iˆ + ∂ ˆ ∂ ˆ j+ ( k  τ w. i. y  ∂x ∂y ∂z   ∂τ   w.i ) .e. tangential.27b)  ∂m ∂n ∂l   ∂τ w. i. m ∂τ w. m ∂τ w.. z   ∂x ∂y ∂z  r The ∇τ w tensor with respect to the xyz-coordinates (x1. k = aik a jl (3.1. are of no interest because the aggravating effects on the endothelium are caused by changes in surface. l     ∂m ∂n ∂l  Due to the coordinate system chosen.i ∂τ w. n  ∇τ w =  (3. l ∂τ w.e. Specifically.s2. the components of the tensor affect the endothelial cell in normal and tangential directions. z kˆ ∂z  )  ∂x ∂y  ∂τ w. n ∂τ w.s3) by a standard component-wise tensor transformation ∂τ w′ . forces. However. all normal components and variations.260 Biofluid Dynamics • l .. z ∂τ w.1. the components related to the l-directional wall shear stress and l- coordinate.26) ∂τ ∂τ ∂τ =  w . x   ∂x ∂y ∂z   (3. m     ∂m ∂n ∂l  r ∂τ w..x2. mnl.k ∂τ w. l ∂τ w. y ˆj + τ w. n ∂τ w. .1.27a) ∂s j ∂xl where the primes denote the mnl-coordinate system. i.x3) can be transformed to the mnl-coordinate system (s1.e. The resultant tensor is  ∂τ w. y w . y w . x ∂τ w. y ∂τ w. The terms aik and ajl represent the directional cosines of the xyz-coordinates rotated to the mnl- coordinates.

29) in which do is a reference diameter and τ o = 8η u mean / d o is the reference Poiseuille-type wall shear stress corresponding to the mean flow rate. and.28)  ∂m   ∂n   In order to employ the WSSG-concept to assess the impact of non- uniform hemodynamics. Specifically. was devised for the publication of Longest and Kleinstreuer (2000). sites of intimal thickening. Lei (1995) suggested that the normal r components of the ∇τ w tensor. Wall shear stress angle deviation or gradient.Chapter III: Analyses of Arterial Diseases 261  ∂τ wm ∂τ w.1.1. The wall shear stress angle deviation (WSSAD) for a control volume of area dAi is formulated 1 r  1 T  τ o ⋅τ n  WSSAD = T ∫  ∫ arccos r A  o n    | τ | ⋅ | τr |  dAi  dt (3.30)  S  0 where the surface stress vector at the location of interest is represented by r r τ o . n ∂τ w. which has not appeared elsewhere. A scalar combination of the normal components can be written as 1/ 2  ∂τ m  2  ∂τ n  2  WSSG(t ) =   +   (3. (2004) proposed that a measure of the angle between adjacent wall shear stress vectors would indicate regions of dysfunctional endothelial cells. those creating tension.27c)  ∂τ w. n   ∂m ∂n  The diagonal components ∂τ w. 1 do T LSWSSGnd = T τo ∫ 0 WSSG (t )dt (3. are the most important ones with respect to intimal thickening due to atherosclerosis or hyperplasia. n / ∂n generate intracellular tension which causes widening and shrinking of cellular gaps. i. is provided in the following discussion. mesh independent parameter.. The rationale for this parameter.e. and τ n represents the four surrounding surface stress vectors. Hyun et al. The off-diagonal components ∂τ w.1. . A similar. m / ∂m and ∂τ w. n / ∂m cause relative movement of adjacent cells.1. m  r   ∇τ w :=  ∂m ∂n  (3. the wall shear stress angle gradient. the absolute value of the local instantaneous wall shear stress gradient is either used directly or in its time-averaged dimensionless form. m / ∂n and ∂τ w. hence.

m and ∂τ w. the distance between the locations at which τ w.31b) or . respectively. It is usually difficult to meet this requirement with the use of structured meshes for complex geometries. In this formulation.31a)  |τo | ⋅ |τo |  where. a scalar field of angular differences can be computed as with the WSSAD. to formulate a mesh independent measure of spatial variation in the WSS direction. vectors over each surface control-volume edge. in the region of interest.1. A mesh independent measure of change in τ w.n values are sampled would be ignored resulting in a mesh dependent formulation.e. In this formulation r r  τ o ⋅τ n  φ = arccos  r r  (3.. i.1. the gradient operation can be used. the inter-nodal distances at the wall are nearly equivalent. 1 ∂φ 1 ∂φ 1 ∂φ WSSAG = Ai ∫ S ∂x dAi iˆ + Ai ∫ S ∂y dAi ˆj + Ai ∫ S ∂z dAi kˆ (3. Similarly.n in the m and n directions is formed using the gradient operation (or directional derivatives). Using the central mean WSS vector as a reference.e. i.m and τ w.n values across one control volume in the m and n directions. i. For example. the WSSAD reports spatial variation of the mean shear stress direction.e. The region of interest is a surface area segment defined by a central control- volume face and its four surrounding nodes. Grids should therefore be constructed to ensure that.m and τ w. The gradient of the scalar field of angular variations is then taken on an area-average basis assuming that each angular difference is constant for its respective control volume edge. the stress vector at the location of interest is represented by r r τ o and τ n represents the surrounding stress vectors. (3. If the WSSG components ∂τ w. again. Similarly. m / ∂m and ∂τ w. they would simply be the differences in τ w.30). respectively.262 Biofluid Dynamics Considering Eq. A mesh independent wall shear stress directional parameter may be formulated using a gradient operation. the WSSAD can be used to evaluate single geometries and make comparisons among designs if and only if the inter- nodal distances between adjacent surface nodes are approximately equal. the WSSG scalar indicates how much τ w. n / ∂n were calculated as the WSSAD is. The resulting wall shear stress angular gradient (WSSAG) is composed of three components.. variation of the m and n directions.m and τ w.1.n vary in the m and n directions.

1. it is often assumed that the initial point of particle wall contact represents a critical location of interest (cf. this parameter is mesh independent even though the central vector was used to compute the scalar field of angles. Blood particle stasis may then serve as a qualitative hemodynamic wall parameter. Buchanan et al. Kunov et al. In the absence of near-wall forces. As discussed. This research hypothesizes that blood particle deposition is most likely in regions of near-wall particle stasis and/or elevated concentrations.1. 1 T 1 WSSAG = T ∫ 0 Ai ∫S ∇ φ dAi dt (3. For a similar stenotic configuration. the magnitude of the WSSAG vector is a coordinate independent scalar that can be computed on a time-average basis ∫ (Θ ) 1 T + Θ y2 + Θ z2 1/ 2 WSSAG = 2 x dt (3. Alternately. However. To derive a general model for the near-wall residence time (NWRT) and hence for the . Regions of enhanced particle-wall interaction may be quantified by the proposed non- dimensional near-wall residence time (NWRT) parameter. Hyun. blood particle depositions and stasis have been widely implicated with intimal thickening and thrombosis formation.31c) The gradient vector can then be transformed to the m and n coordinate system (such that the normal component l vanishes) to quantify specific directional changes in the surface coordinates. For pulsatile axisymmetric stenotic flow. i.31d) T 0 or.. more succinctly. coincident with regions of dysfunctional endothelial cells.Chapter III: Analyses of Arterial Diseases 263 WSSAG = Θ xiˆ + Θ y ˆj + Θ z kˆ (3. Lagrangian-based wall parameters. Ehrlich and Friedman (1977) examined the distribution of blood particle stasis in a Y-branch two-dimensional (2-D) model by tracking Lagrangian fluid elements assumed to be blood particles. 1998.1. Buchanan. (1996) tracked Lagrangian fluid elements which were assumed to be groups of activated platelets in a symmetric (2-D) stenosis and applied the quantitative concept of a local volumetric residence time which takes into account where particles accumulate and how long they remain there.32) which represents the magnitude of the shear stress angle deviation over a distance. the inclusion of near-wall forces coupled with the assumption of spherical particles and Stokes flow prevents particle wall contact. 2000). Due to the use of the gradient operation. any other surface vector could be used for this purpose. This is because the mean shear stress vector is simply a reference.e. (2000) generated maps of localized fluid element residence time for various input pulse frequencies.

Eqs.33) v  r For a local near-wall region of volume V NW .264 Biofluid Dynamics probability of blood particle deposition.3. The distance between a particle center and the wall. rolling. s is set to unity for monocytes and a factor of two for platelets. the higher the probability of particle deposition. and nano-scale bond formations responsible for possible blood cell attachment. A near-wall layer of biophysical interaction is defined by h=50µm. Independence of mesh size and particle loading In conjunction with the correct particle dynamics equations (cf.  apr  S Qav n v r NWRT = noV NW ∑ ∫   p =1 path . at least 50. Distance between the particle and the vascular surface 4. h p . Asurf .1. For the calculation of NWRT. p  h p  r 2 ⋅ dr  (3. is a key scaling (or probability) factor in the term (a p / hp ) s . results in convergent profiles once a sufficient quantity of particles has been simulated. to match experimental results. Specifically. where a p is the particle radius and. .1.1. i. V NW = Asurf × h . a nondimensional near-wall residence time parameter (NWRT) is proposed. 3.13-3. actual blood particle shape. Inclusion of the total number of particles. or re-suspension cannot feasibly be included in simulations involving relatively large-scale 3-D geometries. Availability of particles 3. The near-wall volume.000 blood particles have been released randomly in both space and time over one pulse period at an upstream slice prior to the junction area.e. Bio-particle activation and surface reactivity (which may be proportional to flow induced surface dysfunction) 5. the larger the NWRT value.5). It indicates the probability of particle deposition. The average flow rate term Qav is a physically relevant constant.22). accounts for variations in near-wall control-volume size. n o . v is the magnitude of the particle velocity. n is the number of activated blood particles that pass r through the local volume. such that the solution is practically mesh independent for a limiting value of local surface area. Simulations are continued until all particles have exited the flow field (see Sect. the following factors have been considered: 1. which is included to form a dimensionless parameter. and dr is the local directional unit vector for integration along each trajectory. Time that a particle spends in close proximity to an attached site 2. The NWRT concept is an approximate method which is particularly useful and necessary given that biophysical factors such as vessel surface roughness..1.

4 (Ansys.. minimizing such wall parameters (see Fig. Figure 3. the particle trajectories can be evaluated after the velocity and pressure fields are known.1.4 are now applied to a femoral bypass (see Fig. If a patient’s sapheneous veins are not available as bypass tubes. Inc.. . PA). does not influence the blood flow. Example of occlusive diseases in branching blood vessels: (a) Femoral-to-popliteral bypass graft with (b) Restenosis in the region of the distal anastomosis (Kleinstreuer & Longest. or blocks (see Figs.1 -3. 3. at the distal end where the graft meets the artery (see Fig.1. the vessel wall deteriorates to such a degree that surgical removal of the atheromas is impossible and the diseased artery segment has to be bypassed.1. Unfortunately. This validated finite-volume program employs structured meshes where the complex flow domain is divided into somewhat uniform subregions. ePTFE or Dacron) are used. aggravating particle-hemodynamics factors (see Sects. 3.1.4) play a significant role in restenosis.4) and thereby generating amiable graft-to-artery geometries produces smooth laminar flow fields in the critical junction regions and.1. a dilute suspension of micro-size monocytes and platelets.1.5) restenosis may develop.e. as a result..b).5.5). Thus. high patency rates. i. synthetic grafts (i. Hence.1. Canonberg. Clearly. 3.Chapter III: Analyses of Arterial Diseases 265 3. the flow field subject to suitable initial and boundary conditions was solved using CFX 4.1. as a first step.e. Typically with blood vessel occlusion and age. 3.1. 3.5 Treatment Option: Femoral End-to-Side Graft Bypass The treatment option and supportive equations discussed in Sect. Because the presence of the critical particles. 3. 2003). especially in the toe region.6a.

266 Biofluid Dynamics (a) Mid-plane computational mesh with block boundaries (b) Sample surface mesh Figure 3. the input pulse. 3.1. Parameters necessary for both temporal and mesh convergence were evaluated for a sample femoral- bypass configuration.1.5. . multiblock.6 Computational mid-plane and whole domain mesh for a 1. body-fitted coordinate discretization scheme was employed. A structured.1 Computational Fluid-Particle Dynamics Solution The solution of the time-varying velocity field has been carried out with a validated finite-volume based algorithm (CFX 4.4) with user-Fortran programs added to account for the blood rheology. These conditions provided a starting point for convergence validation in all subsequent configurations studied. and the hemodynamic parameters.5:1 graft- to-artery diameter ratio configuration.

For each outer iteration. using the most recently calculated values of the variables. That is. With the exception of advection. Specifically.1. The fully implicit backward Euler algorithm was chosen to rapidly generate a sufficiently large number of temporal velocity field solutions such that linear interpolation in time could be used for the integration of particle trajectories. (3. the first-order scheme selected with sufficiently small time steps is capable of the accuracy achieved by higher-order schemes with larger time steps.1.35) are discretized in space using second-order central differences. The outer iteration is a Picard scheme used to update the nonlinearities of the problem. the set of linearized difference equations is passed to a simultaneous linear equation solver which uses an iterative solution method.. The advection terms are discretized using the higher-order upwind (HUW) differencing scheme of Thompson and Wilkes (1982). The computational cost incurred by implementing this first-order method was reduced by a variable time-step routine for the solution of the velocity field. i. regarding all other variables as fixed. variable time-step size and not the order of a scheme determine the accuracy of the solution. The discretized transport equations are solved iteratively. before each inner iteration..e. the SIMPLEC algorithm of Van Doormal and Raithby (1984) was employed.35) where the pressure gradient and stress divergence have been included in the source term. The coefficients of the convective terms were obtained using the Rhie-Chow (Rhie & Chow. 1992). There are two levels of iteration: (1) a linear inner iteration which solves for spatial coupling for a particular variable. A fully implicit backward Euler scheme was used for temporal discretization. However. S. r ∂ρv ∂t rr r ( r + ∇ ⋅ ( ρvv − η∇vu ) = − ∇p + ∇ ⋅ η (∇v ) T ) (3. 1983) interpolation formula which has been extended to include non-uniform grid corrections.1.Chapter III: Analyses of Arterial Diseases 267 Solution of governing equations. Each variable is taken in sequence. such as the Crank-Nicholson method. This upwind scheme was derived by integrating fluxes over the control-volume and is completely conservative and consistent with the control-volume formulation (Shyy et al. The finite-volume solution relies on a modified form of the governing equations. and (2) a nonlinear outer iteration which solves for the coupling among variables. Continuity is enforced by solving a pressure-correction equation to update the pressure and velocity fields. all terms in Eq. The coefficients of the discretized equations are always reformed. Higher-order schemes which allow for larger time steps. An exact solution is not required because this is just one step in . are available.34) Integration of this equation for individual control or finite volumes results in r ∂ρv rr r r r ∫ ∂t dV + ∫ ρvv ⋅ ndA − ∫η∇v ⋅ ndA = ∫ SdA (3.

268 Biofluid Dynamics
the nonlinear outer iteration. With the exception of the pressure correction
equation, all variables were solved for using an algebraic multigrid (AMG)
method. This method solves the discretized equations on a series of
coarsening meshes (cf. Lonsdale, 1993) chosen algebraically. The AMG
method takes advantage of the properties of multigrid schemes which can
converge in an ideal number of iterations for certain problems (Ferziger and
Peric, 1999). However, rather than using nested grids which vary by a
factor of two in mesh size to obtain approximate solutions on finer grids,
the AMG method performs restriction and prolongation operations on the
linearized matrix itself using nonsquare matrices to reduce the problem size
and then performs smoothing operations. This “algebraic” method of
prolongation and restriction is what gives the algorithm its name as well as
its performance boost. The advantage of the AMG scheme is that it reaches
a convergent solution monotonically and in fewer total (outer) iterations
than the other schemes. The cost is that each iteration is more expensive in
both CPU time and memory required, but the overall computational time is
reduced. In general, for the transient three-dimensional problems solved,
the AMG method has been used.
The typical convergence criterion for hemodynamic simulations is to
halt the outer iteration procedure and advance the time-step when the global
mass residual has been reduced from its original value by three orders of
magnitude, in cgs units. However, Longest (1999) demonstrated that this
criterion is often insufficient for variable viscosity conditions.
Furthermore, the above convergence criterion often halts a solution process
that is significantly reducing the mass residual. Therefore, monitoring the
rate of both momentum and mass residual reduction, in addition to mass
residual size, is recommended. To monitor convergence rate, a residual
reduction factor (rrf) can be defined as
avg. residual reduction over the last 5 outer iterations
rrf = (3.1.36)
current residual magnitude
The inverse of the rrf indicates the number of outer iterations that
would be required to drive the residual to zero, at the current rate of
reduction. Appropriate values that resulted in a sufficiently convergent
solution without excessive outer iterations were found to be rrf mass = 0.01
and rrf mom = 0.06 . The composite convergence criterion is then:
(i) mass residual mass flow rate × 10-3
(ii) rrf mass ≤ 0.01
(iii) rrf mom = 0.06
The mean mass flow rate divided by a factor of two was used in
condition I when the time varying mass flow rate dropped below this value.
To ensure that a converged solution had been reached, the above factors
were reduced by an order of magnitude and results were compared. The

Chapter III: Analyses of Arterial Diseases 269
stricter convergence criteria produced a negligible effect on both velocity
and wall shear stress fields.
An adaptive time-stepping routine was implemented to maintain a
highly efficient routine throughout the widely varying input pulse. Time-
step size selection was based on the convergence rate of the previous time-
step. Time-step size was either increased or decreased to maintain the total
number of outer iterations around 75. This is a reasonable number of
iterations considering typically under-relaxation factors for the AMG
solver were 0.6 for momentum and 0.25 for viscosity. The resulting time-
step size varied from 0.001 to 0.01 s and was found to sufficiently resolve
Lagrangian particle tracks.

Blood particle properties and trajectory simulations. Human blood is a
concentrated suspension of various types and shapes of deformable
interacting cells in a complex aqueous solution. The platelets are very
small oblate spheroids, or discoids (White, 1994) and constitute less than 1/
800th of the cellular volume. Monocytes are a subclass of leukocytes, or
white blood cells, that are also a dilute cellular constituent and are generally
spherical in shape. More important to blood rheology are red cells which
occupy about 45% of the blood volume and are capable of deforming their
shape, in 0.006 seconds, under the influence of shear. The cellular contents
of blood (including red cells, white cells, and platelets) are immersed in
plasma, which displays Newtonian characteristics. Specific details of blood
components are given in Table 3.1.3, repeated here for the reader’s
convenience. It is the red cells which, at the macroscopic level, determine
the flow properties of blood, and which, at the microscopic level, determine
the motions of the platelets and white cells.
Based on a dimensional analysis of blood particles, it has been shown
that the interparticle hydrodynamic effects are significant and the effect of
the red blood cells on the flow field cannot be ignored. However, due to a
low Stokes number, the two-phase flow can be considered a multi-
component mixture (homogeneous flow) with modified properties (see
Kleinstreuer, 2003). Specifically, blood in sufficiently large vessels is
typically modeled as a multi-component flow with a constant density
influenced by components of the mixture and a variable viscosity
dependent on shear rate and the mean hematocrit. The mixture density of
whole blood is simply the sum of the bulk densities of the constituents. The
viscosity of blood will be modeled with the semi-empirical Quemada
formulation (Cokelet, 1987) which captures the red blood cell based shear
thinning nature (cf. Eq. (3.1.2c)).
The low concentration and low Stokes number of monocytes and
platelets allows for the motion of these cells to be computed in a one-way
coupled manner, i.e., these particles do not affect flow field momentum.
However, the continued collisions and hydrodynamic interactions with red
blood cells results in a persistent lateral motion of platelets and monocytes.

270 Biofluid Dynamics
Table 3.1.3 Properties of Blood Constituents.
Material Red Blood Platelet Leukoytes Plasma Whole
Cell (all) Blood
Density 1.09 1.069† 1.07-1.09 1.03 1.054
Viscosity -- -- -- 0.014* 0.0309
(dyn•s/cm2) **

Blood cell 5 × 109 3 × 108 7.0 × 106 -- --
count (total)
Volume 88 5.17†,†† 460 -- --
Size (µm) 7.7× 2.8 3.0†† 9.5 -- --

Volume 0.42 – 1.9 × 10-3 1.2 × 10-3 -- --
fraction 0.46 (total)
τ p (s) 1.45× 10-6 1.12× 10-6 3.35× 10-6♦ -- --
A= 0.967 0.983 0.964♦ -- --
ρp /ρp

Representative mean value for unactivated platelet (Corash et al., 1977)
Platelet dimensions vary considerably
* at 37°C (may vary from 0.012 to 0.018)
** Newtonian limit at H = 40%

Value for monocytes
In general, blood particle transport is characterized by highly pulsatile
flow and large discrepancies in luminal and near-wall time scales. Hence,
it was found necessary to adopt a highly flexible step-size adaptive
integration routine. Particle tracking algorithms that are coupled to
commercial flow field solvers often limit user access to step-size control
parameters, if active step-size control is even implemented. Furthermore,
commercial routines solve for the transient flow field as the particle
trajectory is progressed; i.e., transient flow field solutions are usually not
stored. Particles that remain in the flow domain for multiple pulse cycles
and/or successive simulations within a single domain result in hundreds or
even thousands of unnecessary flow field solution steps. To overcome the
limitations of available commercial software, blood particle simulations
have been computed in an effective ‘off-line’ manner using a separate
Fortran 90 (f90) routine. This routine stores all velocity and geometry data
in array format for all time-steps of one complete pulse cycle. It then
repeatedly accesses and interpolates velocity field data for the calculation
of particle trajectories over multiple pulses, in this one-way coupled

Chapter III: Analyses of Arterial Diseases 271
formulation. Furthermore, the off-line f90 code provides full access to all
adaptive step-size control parameters, allows for the specification of near-
wall approximations, and has been effectively parallelized. The resulting
algorithm is capable of computing 50,000 particles within a geometry of
90,000 control-volumes for approximately 10 pulse cycles within 22.5
hours, compared to approximately 200 hours required by typical
commercial CFD packages. Furthermore, once the flow field solution has
been generated and stored in array format, each additional group of 50,000
particles may be computed in 2.5 hours. The drawback of the f90 algorithm
is the massive amount of storage required to accommodate the transient
flow field solution. Model Validation
Validation of the particle tracking algorithm applied in the vessel
lumen away from wall boundaries has been established by comparison to
the annular expansion results of Karino and Goldsmith (1977). The motion
of an elliptical hardened red blood cell in an annular expansion under
sinusoidal flow for a specific initialization position was reported as
illustrated in the upper panels of Fig. 3.1.7. The lower panels of this figure
illustrate the comparable numeric result, including drag and pressure
gradient effects in the equation for particle motion. Indeed, the agreement is
quantitatively very good (see Example 1.17 for details).

Figure 3.1.7 Comparison of: (a) experimental observation of a red blood cell
trajectory (Karino and Goldsmith, 1977) (viewed in two stages); and (b)
computational simulation of an idealized spherical particle trajectory.

The near-wall terms in the equation of particle motion (see Eq.
(3.1.13)) are grounded in analytic derivations. Hence, their inclusion is
fundamentally valid. Specifically, the interaction of blood particles with a
responsive vascular surface is dependent on a variety of complex physico-
biological mechanisms including particle pseudo-pod extension, receptor-
ligand molecular binding, and active surface response. Hence, experimental
studies reporting near-wall particle interaction and particle deposition

272 Biofluid Dynamics
cannot be used to evaluate the validity of the near-wall forces without the
inclusion of a particle-wall deposition or wall-interaction model. For these
reasons, the probabilistic NWRT model indicating possible particle
deposition has been proposed (see Eq.(3.1.33)). Validation and applications
of this model, which is inseparably coupled to the validity of the
appropriate near-wall expression, are given in Longest et al. (2003, 2004). Results for the Distal End-to-Side Femoral Bypass
Atherosclerosis may significantly occlude the femoral and popliteal
arteries, more distal arteries, e.g., tibial and peroneal, or a combination of
these vessels. In such cases, vascular bypass grafting is often the favored
surgical option to restore blood flow to the lower extremities. Peripheral
bypass grafts typically originate at the femoral artery (Fig. 3.1.8) with a
proximal end-to-side anastomosis. In cases where the primary occlusive
development resides in the lower femoral artery, the bypass typically
terminates at the popliteal artery with an end-to-side distal anastomosis
either above or below the knee, i.e., the femoropopliteal bypass. In cases
where occlusive developments are more extensive, the distal bypass
anastomosis may be positioned at the tibial or peroneal arteries, or at
multiple sites. The use of an end-to-side anastomotic configuration is
generally preferred for peripheral bypass grafting as it provides blood flow
to terminal branches both proximal and distal to the anastomosis.

Figure 3.1.8 (a) Illustration of a conventional end-to-side distal anastomosis.
Arrows indicate direction of blood flow and shading depicts typical sites of
hyperplasia development (Sottiurai, 1999); (b) a vein-cuff (or Miller cuff) has been
constructed between the synthetic graft and artery to potentially reduce DAIH
(Miller et al., 1984); (c) A vein-patch (Taylor patch) has been constructed (Taylor
et al., 1987).

Chapter III: Analyses of Arterial Diseases 273
Subsequent late-stage graft failures are predominately due to
restenosis resulting from intimal hyperplasia and/or thrombosis at the distal
anastomosis, i.e., distal anastomotic intimal hyperplasia (DAIH).
Hyperplasia formation (see Sect. 3.1.2) has been reported to localize along
the anastomotic suture-line, particularly at the graft toe and heel, and along
the arterial floor opposite the graft as illustrated in Fig. 3.1.9 (Sottiurai et
al., 1983; Sottiurai, 1999; Keynton et al., 2001; Loth et al., 2002). A
common bypass failure scenario consists of a moderately occluded graft-to-
artery anastomosis. Thrombosis (platelet adhesion, cell aggregation, and
coagulation) may then occur in the region of constricted flow, resulting in
sudden vessel occlusion.

Figure 3.1.9 Realistic in vitro flow models from casts. (a) Model from canine
iliofermoral grafts implemented by a variety of researchers, e.g., White et al.
(1993) and Loth et al. (1997); (b) Models used by How et al. (2000) including a
raised arterial floor in the region of the anastomosis.

While the pathological mechanisms responsible for the development
of DAIH have not been fully elucidated, certain hemodynamic parameters
have been qualitatively linked to stenotic development in arterial bypasses
and other branching blood vessel configurations, both macroscopically and
at the cellular level. These parameters include low mean and oscillatory
wall shear stress, large spatial and temporal gradients in wall shear stress
magnitude, and gradients in wall shear stress vector direction (see Eqs.
(3.1.24)-(3.1.33) and Kleinstreuer et al., 2001). Other factors that have been
implicated with DAIH formation include excessive intramural wall stress,
graft-to-vessel compliance mismatch, blood particle stasis, and blood
particle deposition (Kleinstreuer et al., 2001; Sottiurai, 1999; Ku, 1997).
To minimize DAIH occurrence, native saphenous vein is typically the
conduit of choice for arterial bypass graft construction. Resulting long-
term success (i.e., patency) rates have been reported as high as 80% at two
years (Dalman and Taylor, 1993) and 49% at 4 years (Veith et al., 1986).
However, the use of autologous saphenous vein is not an option in as many
as 60% of patients requiring a lower-limb bypass (Pappas, 1998). In such
cases, expanded polytetrafluoroethylene (ePTFE) conduits are typically
implemented, which result in success rates similar to saphenous vein for
applications where the distal anastomosis resides above the knee (Veith et

274 Biofluid Dynamics
al., 1986). For below-knee distal positioning, PTFE grafts typically perform
poorly, with four-year patency rates of approximately 12%. To improve the
success rates of below-knee applications, inter-positioned vein cuffs (Miller
et al., 1984) and vein patches (Taylor et al., 1992) have been suggested and
continue to show promise in clinical studies. For instance, Stonebridge et
al. (1997) found a 52% success rate at 24 months for below-knee popliteal
bypass grafts with inter-positioned vein cuffs (see Fig. 3.1.8). As surveyed
by Longest (2004), the geometric, biofluids, and biomechanics parameters
of interest in bypass graft design include:
• Graft-to-artery diameter ratio and graft angle, where, typically,
d graft > 1.5d artery and 5° ≤ Θ ≤ 45°
• Inlet curvature, especially in the toe region
• Anastomosis shape and length
• Input waveform and maximum, mean, and minimum Reynolds
• Outlet flow division, i.e., percentage of retrograde flow
• Graft-artery wall compliance and distensibilities.
The widely accepted view that hemodynamic factors initiate a
biophysical cascade that is ultimately responsible for distal anastomotic
intimal hyperplasia (DAIH) suggests that geometric modifications may
significantly improve distal anastomotic performance. Considering the
problematic below-the-knee distal femoral anastomosis clinical evidence
indicates that inter-positioned vein cuffs and vein patches significantly
extend graft function. A number of potential surgical, hemodynamic, and
biological mechanisms have been associated with the improved patency of
vein-supplemented designs (see Noori et al., 1999; Leuprecht et al., 2002);
however, no single factor appears responsible. While clinical results appear
promising, wide-spread implementation of vein-supplemented designs has
been limited by the surgical rigor associated with constructing a cuff or
patch as well as the presumed increased risk of surgical complications and
technical errors. In order to expedite the surgical procedure and further
mitigate inciting hemodynamic factors, several researchers have suggested
the implementation of expanded grafts (Lei et al., 1996 & 1997; Harris and
How, 1999; Longest and Kleinstreuer, 2000; Longest et al., 2000; cf. Fig.
3.1.4). As a result, expanded graft-end configurations are now
commercially available (Fisher et al., 2002).
In an effort to improve distal anastomotic performance, Lei et al.
(1996 & 1997) showed that large anastomotic flow areas, small
continuously changing bifurcations angles, and smooth junction wall
curvatures reduced hemodynamic wall parameters including the WSSG and
OSI. They suggested an ‘optimized’ S-shaped configuration intended to
minimize adverse ‘disturbed flow’ characteristics associated with DAIH
(Fig. 3.10a). Following the lead of Lei et al. (1996), Harris and How
(1999) proposed a pre-shaped PTFE cuff intended to eliminate the need for
vein collar construction. While the design objective of Lei et al. (1996 &

Chapter III: Analyses of Arterial Diseases 275
1997) was to reduce disturbed flow patterns in the junction region, the pre-
shaped cuff design of Harris and How (1999) was intended to increase
vortical flow and wall shear stress (Figure 3.1.10c). Harris and How (1999)
claimed the pre-shaped cuff eliminates regions of low WSS; however, no
quantitative details in support of this argument have been provided.
Longest and Kleinstreuer (2000) linearly combined several WSS-based
hemodynamic wall parameters into one ‘severity parameter’ for analysis of
an arteriovenous (AV) bypass configuration. Based on severity parameter
mitigation, an ‘optimized’ AV distal anastomotic configuration was
proposed (Fig. 3.1.10b). Indeed, clinical performance of anastomotic
configurations designed to hemodynamically mitigate restenosis depends
largely on the appropriateness of the wall parameters analyzed. While the
pre-expanded designs of Lei et al. (1996) and Longest and Kleinstreuer
(2000) successfully mitigated the hemodynamic parameters considered,
e.g., the WSSG, recent findings strongly indicate multiple pathways for
DAIH formation. For instance, expanded designs significantly reduce
changes in WSS magnitude and direction by providing larger anastomotic
areas to accommodate flow redirection. A downside of this strategy is
lower WSS-values, which has been widely associated with DAIH
development. Results of the current study indicate that significant
interactions between critical blood particles and the vascular surface
provide an extremely aggressive pathway for DAIH development.
Furthermore, regions of low WSS and lumenal vortical patterns cannot
effectively determine regions of micro-scale particle-wall interactions
including adhesion. Therefore, analysis of WSS conditions and ‘persistent
vortex’ formations may result in an insufficient model for DAIH formation
as well as for graft design. Significant hood curvatures, which characterize
the pre-expanded configurations (Fig. 3.1.9), are expected to result in
elevated particle-wall interactions along the graft surface in the region of
the anastomosis. For example, a low-angle configuration can be
characterized by a continuous and relatively smooth hood curvature.
Nevertheless, significant particle-wall interactions were observed along
such a graft surface, which were consistent with the DAIH observations of
Loth et al. (2002). Considering particle-wall interactions as a potential
mechanism for DAIH formation, alternative anastomotic designs may be
necessary. Thus, Longest & Kleinstreuer (2003) showed that the
hemodynamic factor, expected to most aggressively elicit a localized
hyperplasic response within the distal femoral anastomosis, is the
composite NWRT model for platelets including surface reactivity and
platelet activation conditions. Monocyte interactions with the vascular
surface in regions of low-WSS also have the potential to incite DAIH
formations. As supported by a number of studies at the cellular level,
endothelial response to regions of significantly low WSS are considered a
second pathway for DAIH development. Highly focal regions of
significantly elevated WSSG and WSSAG values provide a third inciting

276 Biofluid Dynamics
mechanism for localized IH formation. Factors such as compliance
mismatch, intramural stress and strain, surgical injury, and technical errors
were not directly assessed in this study.

Figure 3.1.10 Expanded graft-end configurations intended to reduce DAIH
formation: (a) S-shaped connector suggested by Lei et al. (1996, 1997); (b) ‘new
graft-end’ design suggested by Longest and Kleinstreuer (2000) for arteriovenous
access; (c) Distaflo graft proposed by Harris and How (1999) and marketed by

Figure 3.1.11 Geometric surface models including the currently implemented
Miller cuff as well as virtually prototyped anastomotic configurations. Novel System Designs and Discussion
Of interest are realistic and virtually prototyped distal anastomoses
intended to reduce DAIH formations. Models include the clinically viable
Miller cuff, as well as smooth unexpanded virtual prototypes with graft-to-
artery diameter ratios of 1.5:1 (Fig. 3.1.11a). The Miller configuration has

Chapter III: Analyses of Arterial Diseases 277

been constructed using a straight 45o graft-end cut and a 4 mm high venous
cuff. In contrast, virtually prototyped models (Fig. 3.1.11b) implement a
smoothly curved inlet-graft intended to gradually redirect the flow and
potentially reduce disturbed hemodynamic conditions, particularly in the
immediate junction area. Similarly, curvature of the recipient artery allows
for a smooth transition, which is intended to redistribute IH away from
regions critical to anastomotic flow delivery. Models A and B are
characterized by high- and low-angle ‘concave-up’ grafts, whereas Model
C represents a low-angle ‘concave-down’ configuration. In contrast to the
previously expanded anastomotic configurations, utilizing both proximal
and distal graft and arterial curvatures allows for smoothly connected
unexpanded junctions intended to mitigate disturbed flow occurrence
without reducing critical WSS-values and elevating particle residence
A transient Type I input pulse, consistent with post-surgical
observations of the femoral bypass, has been implemented for all grafts (cf.
Table 3.1.2).
While recent studies have suggested a variety of mechanisms
potentially responsible for the improved clinical performance of the Miller
cuff, current results indicate that hemodynamic conditions, including
particle-wall interactions, may partially account for the redistribution of
significant DAIH formations away from the critical arterial region.
Nevertheless, the use of an autologous vein cuff is potentially the factor
most responsible for the reduced hyperphlasic response in the region of the
arterial suture-line (Kissin et al., 2000). Furthermore, the suturing of a vein
segment to the artery facilitates the surgical procedure, resulting in a higher
likelihood of a technically sound anastomosis.
Considering the virtually prototyped models, anatomic features
consistent with venous anastomoses reduced the particle-hemodynamic
potential for DAIH formations in locations critical to flow delivery. For
instance, a concave-up graft inlet configuration (Fig. 3.1.11b) resulted in
elevated particle velocities and WSS-values in the region of the
anastomosis, which were observed to significantly reduce occurrences of
the composite NWRT models for platelets and monocytes. Significant
graft and arterial curvatures, as well as the application of an unexpanded
design, reduced and redistributed WSS-based hemodynamic parameters
and NWRT occurrence; however, the particle-hemodynamic potential for
DAIH formation was not eliminated. Considering the proliferative nature
of IH formations, it appears that eventual occlusion of the virtually
prototyped configurations is expected.
In conclusion, the application of a multiple-pathway particle-
hemodynamic model for IH in distal anastomotic design indicates that
occlusive formations are an inevitable consequence of the un-physiological
distal end-to-side anastomosis, particularly for the case of proximal
outflow. Nevertheless, surgical benefits of the end-to-side distal

The male-to-female ratio of ruptured aneurysms leading to death is 4:1 for AAA-patients (over 55) and about the same for BA-patients.. A key biomedical engineering example of fluid-structure interaction (FSI) dynamics is given in the next section. nausea. a catheter which is a small plastic tube containing an implant is inserted into the patient’s femoral leg artery and navigated . However. such as vascular staples and clips (Segdi et al. atherosclerosis. BA-patients may experience the worst headache in their lives. often augmented with angiography. loss of sensation. Relatively new graft fasteners.g. light sensitivity. most frequently. neck pain. etc. stroke symptoms. life style (e. epicardial (i. Aneurysms can be detected via angiography (i.2 ANEURYSMS An aneurysm is an abnormal irreversible outward bulging of an artery.. results of this study suggest the implementation of concave-up graft inlets. blurred vision. In EVAR. speech complications. in order to prevent possible rupture at a later user stage. coronary) artery or. As such. expected in the immediate junction and lateral wall regions. and/or vomiting.. fainting. loss of balance. The latter. can detect the aneurysm location and shape with precision. 2001).. Graft and arterial curvatures can be clinically accommodated by the use of prefabricated external supports. The two treatment options are either major open surgery or minimally-invasive endovascular aneurysm repair (EVAR). stress. may potentially reduce suture-line IH. smoking. Clinical testing will be necessary to determine if the unexpanded anastomotic design suggested can accommodate moderate IH formations. Possible causes include inherited genetic disorder.e.1). Specifically in about 40% of BA cases. people with unruptured aneurysms will experience peripheral vision deficits. as found in a brain capillary. short-term memory difficulty. such as ease of construction and the ability to deliver proximal outflow. the abdominal aorta below the renal arteries (see Fig. ultrasonography. without significantly altering graft function. eye pain.e. or computerized tomography (CT) scans. ensure its continued implementation until a better alternative is proven. Another important application of computational particle- hemodynamics is the prediction of geometrically optimal. diet. no regular exercise. 3. hypertension.2. Only 10% of AAA-patients have symptoms such as abdominal pain.278 Biofluid Dynamics anastomosis. and/or sudden changes in behavior. Both brain aneurysms (BAs) and abdominal aortic aneurysms (AAAs) develop often over decades unnoticed. biodegradable scaffolds on which complex arteries can be grown in vitro. when rupture suddenly occurs. 3. vomiting. injected contrast dye plus X-rays). and/or death. Coupled to such an analysis is the evaluation of the stress field in those arteries under typical pulsatile loads. back pain. relatively unexpanded anastomotic configurations.) as well as hemodynamic and biomechanical factors. and arterial curvatures which moderate flow redirection.

.1 Schematics of aneurysms: (a) brain and (b) abdominal aorta.2. making it the 13th leading cause of death in the US. typically twice the aortic neck diameter.000 deaths every year.2.1 Aortic Aneurysms The overall mortality caused by sudden AAA-rupture is 90%. that implant is typically a platinum coil to be deposited into the brain aneurysm to fill out the cavity (Fig.e.Chapter III: Analyses of Arterial Diseases 279 through the vascular system into the afflicted artery segment where the implant is deployed.. about 15.1 compares the pros and cons of these two surgical interventions where the total cost for the modern procedure is much higher than for conventional aneurysm repair. As mentioned.2. 3. Figure 3.1a). while for AAA-patients it is a bifurcating stent-graft which forms a new blood vessel and hence shields the weakened AAA-wall from the pulsatile flow (Fig. 3. For BA-patients.2. to prevent such a terminal event either open surgery or endovascular aneurysm repair (EVAR) is recommended when an AAA reaches a critical size.2. i. Table 3.1b).3.

preventing over- dilatation and rupture of the vessel.3 % 3. In any case. 2004). SMCs seem to be of minor importance in the abdominal aorta. Collagen is 1000 times stiffer and is load-bearing at high pressures.and gender- related differences are most pronounced in the aorta which may explain the 4:1 male-to-female ratio of AAA-mortality among the elderly. failure in the collagen matrix might not be detectable during normal pressure conditions and therefore the risk of vessel fragility cannot be predicted.8% No long-term surveillance necessary Long-term surveillance necessary Average total cost: $12. stiffness increases with age as a result of an increase in the collagen-to-elastin ratio in the wall. and smooth-muscle cells. In healthy arteries. Aneurysm rupture occurs when the imposed stress exceeds the tensile strength of the wall. Open surgery EVAR Suitable for almost any patient Only for AAAs with “right anatomy” Large abdominal incision Small incision in groin Average 6-day stay in hospital Average 2-day stay in hospital Full recovery after 6 weeks Full recovery after 2 weeks Morbidity: 29% Morbidity: 18% Blood transfusion: high Blood transfusion: low Mortality rate of selective surgery is Mortality rate 1. However. Thus.. the collagen-to-elastin ratio is the principal determinant of wall mechanics in the aorta. .2. these age.1 Mechanisms of AAA Development The arterial wall mechanics and integrity are mainly determined by the matrix components of the wall.500 Average total cost: $20. Furthermore.2.280 Biofluid Dynamics Table 3.2 these are predominantly elastin. collagen.3 and indicated in Fig.1. 2. As alluded to in Sect. with men having stiffer arteries than women. Smooth muscle cells (SMCs) have the potential for contraction and relaxation with modulation of the wall mechanics.2.1 Comparison between open surgery and EVAR (Greenhalgh et al. Thus. changes in composition and structure of the arterial wall will alter the wall mechanics.2. 3. the increasing collagen-to-elastin ratio results in increased wall stiffness and decreased tensile strength.000 3. The distensible elastin is load-bearing at low pressures and responsible for the elastic recoil of the artery.

where the σ .2. 3. while its breaking stress decreases significantly.2.2 Comparison of elastin and collagen fibers in the arterial wall (after Raghavan 2002). .2. The elastic modulus is much higher than that in the normal aorta.3 Comparison between AAA wall and normal aorta mechanics (after Raghavan 2002). The stress-strain effect of the much higher collagen-to-elastin-ratios in AAAs is shown in Fig.Chapter III: Analyses of Arterial Diseases 281 Figure 3.ε curve of the AAA moves to the left considerably.3. σ Breaking stress of normal abdominal aorta Breaking stress of AAA Elastic modulus ε Figure 3.

the yield stress of the wall will decrease significantly with respect to age. the yield stress is possibly lower than the mechanical stress in the AAA wall. 3. are not yet widely used (cf. Generally.2. because stress measurements are not available in vivo. or almost three times that of the normal arterial wall whereas its yield stress is only 50% of the normal artery.282 Biofluid Dynamics Yield stress Stress Wall stress Rupture point Time Figure 3. Computational stress analyses with CAT or . the wall yield stress will accordingly decrease. rupture may happen suddenly. Li & Kleinstreuer. although the maximum AAA diameter and growth rate as well as aortic neck asymmetry are very important as well and easier to measure. how to define the critical values of a threshold stress and yield stress for different patients is rather complex. In summary. along with the increase of wall stiffness. 3. because of the complex mechanism of rupture and often complex AAA geometries.e. and Von Mises stresses. Young’s modulus in an AAA wall may reach over four MPa. Also.e. 2005b).. once the (local) mechanical stress exceeds the yield stress of the AAA wall. which stress actually causes AAA rupture may differ from case to case. such as Ansys-CFX Multiphysics.. longitudinal. even though stiffness may become large with age.2.4 Effect of yield stress decrease on AAA rupture. because. Enhanced wall stiffness is not necessarily advantageous for preventing AAA rupture. the best and most accurate tool to determine AAA-rupture risk is the maximum wall stress (see Li & Kleinstreuer. Clearly. are considered in AAA rupture analyses.1. circumferential. 2005a).2. while a higher Young’s modulus over time may reduce AAA- wall stress. Furthermore. For example.4). However. and software packages for fluid- structure analyses. i. AAA rupture still may occur as the wall becomes stiffer (see Fig.2 AAA-Wall Stress and Rupture The general consensus is that the peak wall stress is the best indicator of AAA rupture. i. three stresses. Hence.

simple tubular ones which are also used (without the graft sheath) for scaffolding coronary arteries. material fracture or deterioration. Thus. and hence rupture of the artery may be prevented. However. which consists typically of an NiTi- wire mesh embedded in synthetic graft material such as ePTFE or Dracon (see Fig. given realistic flow input/output boundary conditions.2 Treatment Option: Stent-graft Implants In general. called coil embolization or “coiling” (see Fig. on a representative EVG. i. i.5. and friction forces. For an abdominal aortic aneurysm (AAA) a stent-graft (or endovascular graft. it restores normal blood flow and shields the weakened aneurysm wall from the pulsating pressure. i.2. stents (or endografts) are expandable metal/plastic perforated tubes.e.. etc. potentially most devastating is EVG migration.3. major complications may occur because of stent-graft failure which implies improper implantation.e. As a result. For a brain aneurysm. .4.. the aortic aneurysm or brain aneurysm. called fluoroscopic imaging.6). inertia. i.. 3.” a thin platinum wire forming “coils” is deployed to fill out the bubble.Chapter III: Analyses of Arterial Diseases 283 CT-scan based AAA models have been performed by several investigators. a significant radial pressure drop occurs.. potentially leading to aneurysm rupture. endoleaks. axial EVG displacement over time. Once an EVG inside the catheter reaches its destination. and/or migration. to visualize the patient’s vascular system and to deploy the stent accurately. e. EVG) is employed. a small plastic tube. geometries. and material properties..1a). bifurcated ones making up the vast majority.2.e. i. through the patient’s vascular system starting with the femoral artery. There are three basic EVG configurations. which is typically a “blister. coils. the aneurysm cavity may be again exposed to high blood pressure. Thus. These events are often coupled and each has a multitude of underlying hemodynamics and biomechanics causes.e. pressure..3. They are delivered in a catheter. and aorto-uni-iliac implants which have to be used when one iliac branch is highly stenosed. the blood flow into the aneurysm is blocked.e. wire-meshes. For example. Clearly. all the way to the afflicted site.2. that implant functions like a new blood vessel.1b and Fig. This is a coupled fluid-structure-interaction (FSI) problem which can be solved numerically. The endovascular surgeon uses real-time X-ray technology. which occurs when the downward pulling force exceeds the EVG anchoring (or fixation) force. it is deployed either via a self-expanding or balloon-expanding mechanism.g. if an EVG is correctly placed and no future complications occur. as discussed by Li (2005). it is important to know the maximum migration force exerted by the net momentum change.

For a cylindrical NiTi-stent interwoven with graft material. ∆ Re m / Re m < 5% .1) D AAA. 3. thus. Table 3.2 s. stagnant blood.2.2 to 2.49.7 mmHg. 3.0-1.284 Biofluid Dynamics 3. as: (2. It should be noted that the time-dependent Reynolds number will be somewhat different for the nonstented simulations as the internal diameter of the lumen changes slightly.2 + (3..2. Considering the ballooning effect the wall thickness is assumed to be only 1. and AAA wall. a maximum migration force F.5c). The cavity is filled with stagnant fluid experiencing a time- dependent pressure as a result of the dynamic fluid-structure interactions between EVG. 2000). i. The inlet velocity profile is uniform and fully developing in a tube of length ten times the neck diameter.7cm) as the inner arterial wall.e. To study incipient EVG migration.2.e. we applied a pre-stress in the neck of the EVG before the simulation began. A typical inflow velocity waveform is shown in Fig.2.2. is calculated for different practical situations. max − Dneck The healthy artery section (neck) is incompressible with a Poisson ratio of 0. of which Young’s modulus is a key factor influencing the wall stress. 3. no direct experimental data was available. Young’s modulus is assumed to be linearly changing with AAA diameter. the peak and average pressures are 122 mmHg and 98. The EVG is assumed to be a uniform shell and self- expandable with the same diameter (say. i. Here. Experimental data indicate that Young’s modulus of an AAA is much higher than for a normal artery. respectively (Meter. 2001).5b with a maximum Reynolds number (Remax) of 1950 and average Reynolds number (Reaverage) of 330. The maximum EVG fixation force is the key factor affecting migration. To simulate the effect of self-expansion and maintain the contact between the EVG and arterial wall.2)( D AAA − Dneck ) E = 1. an equivalent parameter value supported by experiments was employed (Suzuki et al. changing from the neck to the maximum diameter site based on volume conservation.7 − 1. and the AAA wall is nearly incompressible with a Poisson ratio of 0. For the measured outlet pressure (see Fig. The aneurysm wall thickness also plays an important role in wall deformation and stress. thus.3 Stented AAA-model Analysis In order to demonstrate the benefits of a stent-graft implant. the EVG neck expands by 15% due to over-sizing together with the arterial wall to assume permanent contact.5a). from 1.2. required to dislodge the EVG from the AAA neck..7 MPa. we consider laminar axisymmetric blood flow through a tubular stent-graft (or EVG) interacting with the stagnant blood in the cavity and the aneurysm wall (Fig..5 mm. . 1.2 lists the structure parameter values used in the present simulations.45. The pulse period is chosen as T=1.

4 0.2 -400 Time (s) (c) Exit pressure waveform 125 120 115 Pressure (mm Hg) 110 105 100 95 90 85 80 0 0.2 Time (s) Figure 3. and (c) Output pressure waveform.2 0.6 0.8 1 1.2 0.2.8 1 1.4 0. .5 Abdominal aortic aneurysm model: (a) Schematic of stented AAA.Chapter III: Analyses of Arterial Diseases 285 (a) AAA wall Normal Proximal Cavity Distal artery neck neck EVG F Displacement force required to dislodge EVG (b) Input velocity waveform T2 2100 1600 1100 Re T3 600 100 0 T1 0.6 0. (b) Input velocity waveform.

j + ρf i in F Ω (t ) (3. k 0 =4.2.i = 0 (3.5 mm 0.2 mm thickness thickness mm thickness Maximum Inner Outer 17 mm inner 60 mm 17 mm diameter diameter diameter Inlet: 30 mm Length Length 80 mm Length 120 mm outlet: 30 mm Equivalent Young’s Young’s 1.8MPa modulus modulus MPa modulus Equivalent Poisson Poisson 0.12 Equivalent 6.2 Parameters required for the stented AAA simulation. . j + τ ij . are: (Continuity) u i .3) ∂t (Stress tensor) τ ij = ηγ&ij (3.2-2. γ& r = γ&/ γ& c is a relative shear rate with γ& c being defined by a “phenomenological kinetic model” (Buchanan et al.014 dyn⋅sec/ cm2 is the plasma viscosity.286 Biofluid Dynamics Table 3.7 1.0-1.45 Poisson 0. ρ is the fluid density. Ω (t ) is the moving spatial domain upon which the fluid is described.27 ratio ratio ratio 1. j = − p i .0MPa Young’s 19..12 g/cm3 Density g/cm3 density /cm3 The basic equations in tensor (or comma) notation.0 g Density 1.5 Equivalent 1.49 0. γ& ij is the shear rate tensor.2.2.5) where u i is the velocity vector. η p =0. uˆ i is the wall F displacement velocity at time t.2. f i is the body force at time t per unit mass.2. 2000).4) ηp (Non-Newtonian fluid model) η = 2  1  k 0 + k ∞ γ& r1 / 2   1 −   Ht    2  1 + γ& r 1/ 2   (3. p i is the pressure scalar. Normal Artery AAA Wall EVG Wall Wall 1. following Einstein’s repeated index convention.2) (Momentum) ∂u i ρ + ρ (u j − uˆ j )u i .5862 is the lower limit Quemada viscosity constant.

3. In order to analyze the stress distributions in both the endovascular graft and the arterial wall. 3. Specifically.2917 is the upper limit Quemada viscosity constant.3. σ ij is the mechanical stress tensor. The force field due to the blood flow was then connected to ANSYS 7.) to obtain wall movement and wall stress distributions.) to obtain the blood pressure distributions which.5. wall motion changes the cross sectional area and hence influences the blood flow.8) and (Constitutive) σ ij = Dijkl ε kl in S Ω (t ) (3. multinomial Lagrange interpolations for load and deformation between CFX and ANSYS were required.e. Ti is the surface traction vector at time t. Li & Kleinstreuer (2005a) employed the finite-volume solver CFX 4. Dijkl is the Lagrangian elasticity tensor. in order to satisfy mesh and solver compatibility. Inc.2. 2 σ Von Mises = (σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 (3. used as a material fracture criterion in complicated geometries. i.2). the coupled set of PDEs has to be solved numerically. For this particular problem.2. is employed. represent the loads on the walls.1.1 (Ansys. and σ 3 are the three principal stresses (see Sect.2.Chapter III: Analyses of Arterial Diseases 287 k ∞ =1.2.7) dt (Equilibrium condition) σ ij n i = Ti on S Γ (t ) (3.6) where a i connects fluid flow dynamics with structure mechanics.2.2. ni is the outward pointing normal on the wall surface S Γ(t) .2. and ε kl is the strain tensor. Specially. after integration... 3. 2000). duˆ i ai = (3.. Lagrange interpolation was used to obtain the new domain boundaries and then the CFX-subroutine “Moving . i.1 Basic Structure Equations The general governing equations for structure dynamics are: (Momentum) ρa i = σ ij . S Ω(t ) is the structure domain at time t. velocity and pressure. Inc. in order to find the maximum von Mises stresses in the AAA without and with a stent-graft.9) Here. S Γ(t) is the boundary of the structure domain. σ2 . the Von Mises stress. and Ht =40% is the hematocrit (see Buchanan et al. In turn.e. j + ρf i in S Ω (t ) (3. Concerning the flow solver.10) 2 where σ1 .4 (Ansys.2 Numerical Method Now.

In order to fulfill the FSI convergence between CFX and ANSYS. Concerning the structure solver. To connect CFX and ANSYS successfully.6. specific coupling routines were written in FORTRAN for CFX.2.6 Flow chart for coupled CFX-ANSYS procedure. meshes for both fluid and solid domains were refined until grid independence of the results was achieved. A flow chart of the coupled fluid-structure interaction procedure is given in Fig.2.288 Biofluid Dynamics Grid” was used for re-meshing. it needs several loops for each time step. as well as UNIX scripts. For fluid-structure interactions in the AAA cavity. To reach mesh convergence. . can perform load and deformation transfer automatically. Start Use CFX to simulate fluid flow Calculate loads on boundaries Loads interpolation for ANSYS mesh Use ANSYS to simulate structures dynamics Obtain moving boundary information from ANSYS Deformation interpolation for CFX mesh Update geometry and re-mesh CFX domain No Complete time step in this pulse ? Yes No Converged Result ? Next pulse Yes End Figure 3. a special finite element. 3. APDL is a scripting language for ANSYS data and parameter operations. so that the simulations with the two software packages could be synchronized. APDL (ANSYS Parametric Design Language). Fuild79. the flow pressure field was automatically interpolated to a pressure load-applying function in ANSYS.

231 displacement stented AAA (Malina mm in stented AAA.2.3b.115 MPa 3. et al. 1998) .2. pressure. 1990) Radial deformation 0.3... CFX-ANSYS Theoretical simulation results (Nichols et Error results al. were considered. and pulsatile wall motion for stented AAAs.3% (internal wall) Radial strain 0.8% stress equation) Table 3.2. 80% drop 2003) 10% (Blankensteiju Diameter decrease 10.16 MPa AAA circumferential 0.0155 2. The small differences between our present simulations and theoretical analyses as well as clinical observations are listed in Tables 3.0209 0.3 Model Validations In order to test the accuracy of the coupled CFX-ANSYS solver with user-supplied programs to simulate interacting flow and wall variables.3a Comparison of results between simulations and theoretical analyses.2 mm in stented AAA. For example. theoretical results for elastic cylindrical and spherical walls as well as clinical data. several computer model validation studies were performed.0203cm 4. CFX-ANSYS Clinical Simulation results observations Von Mises stress 75% (Sonesson et al.3a and 3. Table 3.3b Comparison of results between simulations and clinical observations. 0.8% (external wall) Maximum 0.2. 2002) 1 mm in nonstented 1.0203 4.2.0161 0.57 mm in non- Maximum wall AAA and 0.4% Straight Artery (internal wall) Circumferential stress (internal 0.17 MPa (Laplace’s 5.0212cm 0. including diameter..119 MPa 0.2% & Prinssen.Chapter III: Analyses of Arterial Diseases 289 3.4% wall) Radial strain 0.

77E+05 0.7 Comparison between stented and nonstented AAA. when compared to the (cyclic) cardiac pressure.4 Comparison of EVG wall mechanical behavior between transient and steady state flow (Re = 1200). so that Re = 1200 was selected as a typical case.2225 Steady 6. Maximum Von Mises Maximum deformation stress (Pa) (mm) Transient 6. blood pressure is the main force to cause wall deformation and the exit pressure reaches its maximum value of 120 mmHg when Re = 1200.2.2.3. for reasons of computational efficiency.2. Table 3. under the same cyclic exit pressure.2.2.10).8 and 3. equivalent steady flow was assumed to simulate aspects of the EVG effect and EVG-neck interface force. It should be noted that Figs.5). As mentioned.2. while for the others equivalent steady flow was assumed with Pout=120 mmHg and the associated Re=1200 because at this Reynolds number the peak systolic pressure is achieved. 3. the absolute pressure difference between the inlet and outlet for steady and pulsatile flows is very small.2. The influence of flow patterns and hence wall stress effects on the structure are not as important as the cyclic pressure changes. maximum deformation and wall stress values are very similar for steady and transient flows (see Table 3.9) and aspects of the EVG-neck interface force in light of possible EVG migration and hence the specter of aneurysm rupture (Fig.4 Results and Discussion The results deal with the advantages of an endovascular graft (EVG) protecting the weakened arterial wall (Figs. .7-3. 3.80E+05 0. Although the wall-pressure distributions differ between steady and transient flows.4). Hence.2216 (a) No EVG (b) With EVG Figure 3.2. 3.2.9 were generated under pulsatile flow conditions (see Fig.290 Biofluid Dynamics 3. 3. It indicates that.2.

5 mm and the maximum Von Mises stress reaches 0. 0 1. i.7b. It can be seen that an EVG .0. 4 AAA wall deformation (mm) 1. 6 With EVG 0.0. and the location of the maximum Von Mises stress moves from the AAA wall to the EVG.e. 0 0. it can be deduced that the EVG is carrying most of the blood pressure load. the aneurysm expansion is now minimal.7 the highest values in dmax -expansion and wall stress are displayed.. 6 0. Fig.2. only 24. 8 1. max. 4 . 3. It was also found that the sac pressure was reduced from 122 mmHg to 29.8 (a) Comparison of AAA wall deformation changes with time. the change in maximum diameter is only 0. 8 .0.7a and 3. 3. the maximum diameter change is up to 1.e. 2 time (s) (b) 6 Axial WSS (Pa) 5 AAA-wall without EVG 4 AAA-wall with EVG 3 EVG-wall 2 1 0 -1 . 0 0. 3.19 mm. 8 0. 4 0.1% of the average lumen pressure. Specially.153 MPa. 4 0. 2 0 0.2. 6 0. 4 Without EVG 0. 8 1 -1 x/L AAA/2 x/LAAA -2 -3 Figure 3. Comparing Figs. 6 1.2. for the nonstented AAA.4 mmHg. while the maximum Von Mises stress in the AAA wall has decreased to 0.. 0 0.2. While in Fig. After placement of the EVG.Chapter III: Analyses of Arterial Diseases 291 (a) 1. and (b) WSS comparisons between stented and nonstented AAA models. 6 .589 MPa in the AAA wall.2. 2 0.0. i. 2 0. 2 0. 2 1.8a depicts temporal changes in dAAA.

Such areas may be prone to thrombi formation. there is no published clinical evidence regarding the role of WSS in the expansion and/or rupture of AAAs. which means that the influence of the blood pressure profile on potential AAA rupture is remarkably reduced. Our simulation shows that the cavity pressure changes between 26.9a. Furthermore. At present.2.2 0.8 1 1. 3. because the stagnant blood is incompressible. The simulation results are in agreement with clinical observations and experimental results (cf. (a) (b) 30 0. In AAAs.9 (a) Sac pressure changes with time. the magnitude of wall deformation at the maximum diameter has decreased to only about 14% of that in the nonstented AAA. i. generating a near- uniform WSS distribution on the internal wall of the EVG.26 0.7%) during one cycle in the absence of endoleak in this study (Fig.e.3 28 psac/plumen 0. and (b) Ratio of sac pressure to lumen pressure changes with time.292 Biofluid Dynamics may reduce the impact of pulsatile motion on the AAA wall significantly. In stented AAAs. Indeed. if only the EVG/AAA wall is flexible.4 0. the wall motion is not sensitive to the pulsatile pressure wave anymore.9 to 29.6 0.34 0. even without endoleaks. strong sustained WSS changes may damage the endothelium (Kleinstreuer et al. When comparing Fig. 2001).2.2 time (s) time (s) Figure 3. Thus. 2003 and Gawenda et al. Figure 3.4 0..2. under the conditions of fluid-structure interaction. it is evident that the sac pressure and the exit pressure waveform vary almost synchronously.. the sac pressure can be generated by the moving boundary resulting from hydraulic effect and transfer to the surrounding wall.24 26 0. the sac-pressure is nonzero in the absence of endoleaks and the stagnant blood can transmit pressure. high mechanical stresses and low wall shear stress (WSS) regions almost coincide.4 mmHg (psac/plumen varying 24%-32. . 2003).2 0..28 27 0.32 29 Sac pressure (mm Hg) 0. the blood pressure in an AAA cavity is clinically not zero (Sonesson et al... 3.2 0 0.8 1 1.6 0.22 25 0. thereby influencing the deformation of EVG and AAA walls.2. 3. 2004).2 0 0. Sonesson et al.9a). furthermore.8b indicates that an endovascular graft may reduce WSS-values in an AAA to zero.2.5c with Fig.

This test force was increased gradually. The simulation results show that high blood pressure may significantly decrease the maximum fixation force (Fig. this correlation between the maximum fixation force and the friction coefficient is nonlinear for the entire EVG-AAA contact system.e. Due to the lack of clinical test data for EVG-neck friction coefficients. inappropriate aortic neck configurations (or tissue damage). reaching a maximum when f >0. under the same inlet velocity and exit pressure conditions. and for them trans-renal fixation may be a promising choice. the value of the maximum pulling force F needed for EVG displacement was recorded.5a).. In general.10a that a higher friction coefficient increases the maximum EVG fixation force.Chapter III: Analyses of Arterial Diseases 293 Stent-graft movement may occur in AAAs if the EVG-neck fixation is insufficient to overcome the maximum force induced by the blood flow. i. twisting and kinking. and ultimately EVG failure. in order to calculate the maximum fixation force that the EVG can provide.2. pressure. the proximal neck length should be greater than 13 mm. a pulling force F was applied to the EVG (see Fig.2. 3. if the patient has a long infra-renal neck.2. limb thrombus and occlusion. It should be noted that momentum change and wall shear stress may be ignored in this straight tubular EVG when compared to the transient pressure load on a bifurcating EVG. migration is triggered by momentum change. this pulling force is then equal to the maximum EVG fixation force. if possible. Thus. for the following migration analysis a steady luminal outlet pressure (pout=120 mmHg) was assumed as the representative load for the coupled CFX-ANSYS simulations.1 ≤ f ≤ 1. Hypertension is very common in AAA patients. once the EVG began to move. Many patients have short infra-renal necks. partly because of “neck dilation” discussed below. Figure 3. which results in cyclic axial force changes.10b shows that the maximum fixation force is proportional to neck length. Migration may cause endoleaks. and wall shear stresses caused by the local blood flow. the aortic neck length and hence the contact area is important.8. Therefore. Clinically.2. and/or biomechanical degradation of the prosthetic material. the risk of migration is very high. In addition to a sufficient friction coefficient. the friction coefficient and contact area determine the fixation force. the best method to increase the fixation and to avoid EVG-migration is to utilize the extra length. tortuosity. It should be noted that the increased blood . Now. The reason is that the structure behavior of the EVG-neck contact system is nonlinear in light of the large displacements.3. the flexible wall deformations are very similar for both steady and transient flows. If the blood pressure p is greater than 160 mmHg. Thus. Based on Coulomb’s linear friction law. EVG-migration is still a prevailing problem after endovascular aneurysm repair (EVAR). (1999).10c). 3.0 was selected based on the experiments of Lambert et al. It can be seen from Fig. although Coulomb’s linear friction law is employed for each contact element. a friction factor range of 0. As mentioned.

neck dilation is presently seen as the main reason for EVG migration.1 0. and (d) neck dilation. contact still remains between the AAA-neck and EVG. which may lead to EVG migration. As mentioned.5 0.2. the AAA neck diameter increases with time after EVG placement. As a result. Neck dilation is a very common problem in EVG applications. “Neck dilation” implies that. 3.294 Biofluid Dynamics pressure can cause the contact to loosen but not to separate because the EVG is self expanding. high blood pressure plays a very important role in EVG migration for both tubular and bifurcated EVGs. for a bifurcated EVG.2. In order to test this event. if the arterial wall deformation caused by blood pressure is less than 15% of its diameter. the contact should remain. (a) (b) 8 10 EVG displacement force (N) EVG displacement force (N) 8 6 6 4 4 2 2 0 0 0. the contact between EVG and AAA neck diminishes and the maximum fixation force that the EVG-neck can provide will be reduced. a pre-stress on the EVG is applied which can produce 15% of over-sizing between EVG neck and arterial wall.7 0. But the contact force will decrease accordingly which leads to the small force required to dislodge the EVG.3 0.5 4 5 3 2 4.. i. the pulsatile blood pressure may also cause a large longitudinal force on the EVG.10 Relationship between migration force and: (a) friction coefficient.1 10 15 20 Friction coefficient Neck length (mm) (c) (d) 6 6 EVG displacement force (N) EVG displacement force (N) 5 5. we increased the neck diameter by 0 to 20% of its original value to simulate neck dilation. due to biochemical processes in the wall tissue.e. (c) blood pressure. Clinically. but the contact force is reduced (Fig. Thus. (b) neck length.9 1.10d) by up to 25%. In conclusion. Because of the pre-stress and blood pressure applied to the EVG.5 1 4 0 120 130 140 150 160 170 0 5 10 15 20 25 Blood Pressure (mm Hg) Neck dilation (%) Figure 3. Researchers found that the neck diameter growth rate is about .

15:1:0.e. tissue- related. Clearly. barrel. but the cause of dilation may not only be due to the placement of an EVG in the neck.Chapter III: Analyses of Arterial Diseases 295 0. specific EVG designs for groups of patients may prevent migration.9 1. The hourglass shape is the worst and its maximum fixation force is only 10% of the straight neck configuration (Fig. 3. the tubular neck appears to be best for EVAR placement and neck fixation because the other four shapes are irregular and may decrease the contact area to a small ring.85:1:1.11).5mm/year after open surgery.9 0. dilation may cause EVG migration.3.1 1. factors as well. the AAA region and all biomechanical parameters were identical for the five case studies.2.9mm/year in the distal region.7- 1.5 Conclusions The following conclusions can be drawn from this fluid-structure interaction analysis as applied to an axisymmetric stented AAA: .9 mm/year in the proximal region and 1. conical. Some researchers found that dilation also occurs in the super-renal artery without EVG placement.2. The lengths of the neck segments are the same for all neck shapes where the geometries are shown in Fig. a neck with an hourglass or conical shape is not suitable for EVAR. Therefore. 6 5 EVG displacement force (N) 4 3 2 1 0 Straight Barrel Inverted conical Conical Hourglass D1 D1 D1 D1 D1 D2 D2 D2 D2 D2 D3 D3 D3 D3 D3 1:1:1 0.3.2. after EVAR. Clinically. i. Based on clinical observations. there are other.2.3.9:1:0. As expected.15:1:1. However.11 Influence of neck configuration on EVG migration. Because different patients have different neck shapes.15 Figure 3.. neck shapes can be grouped into tubular.11. and hourglass. reversed conical.7-0. the neck diameter growth rate may reach 0.

for example. (2) Sac-pressure (non-endoleak) may be caused by fluid-structure interactions between EVG. stagnant blood. ultrasound. Taking coronary bypass as an example. as well as biomechanics followed by fast and accurate solutions of such a fluid- structure interaction (FSI) problem. intraluminal thrombus. can be shared around the world for expert interpretation and second opinions (see Rangayyan.3. The concentrated regions for low wall shear stress and high mechanical stress are eliminated. (3) Simulation results indicate that the maximum EVG fixation force is determined by multiple factors.1 Introduction Modern information technology is slowly advancing into the field of clinical medicine.4). While images of a patient’s disease. and powerful for optimal surgical recommendations. It should be noted that in this study AAA asymmetry. . planned optimal surgical intervention requires realistic descriptions of the patient’s coronary arteries in terms of geometry. and AAA wall. 5. the assessment of disease severity based on nonlinear multivariable computer programs or to follow the recommendations for optimal surgical intervention based on elaborate computer model simulations and analyses. echocardiography. is rare. iliac bifurcation. and proper EVG placement. improved EVG designs. wall motion. neck dilation. In patient-specific stented AAAs. Both the sac pressure and the ratio of sac-pressure-to-lumen-pressure change with time. and/or CT- scans. hemodynamics. mechanical stress. MRI machines. and neck shape. 3.3 EXAMPLES OF COMPUTERIZED DISEASE MANAGEMENT 3. and angled neck were not considered. there is a prevailing reluctance among surgeons to trust.296 Biofluid Dynamics (1) An endovascular graft (EVG) can reduce pulsatile sac-pressure. videos. For some good reasons. using computer programs and CFD simulations. (4) Coupled CFX-ANSYS simulations for blood flow and EVG- artery structure interactions are valid. these factors may significantly affect the fluid-structure interactions and hence possible EVG migration (see Sect. quantitative medical diagnostics and decision-making. generated via digital cameras. which may prevent AAA rupture effectively. and the maximum diameter in AAAs significantly. 2005). including EVG-neck friction coefficient. blood pressure. predictive. neck length. Quantitative medical diagnostic programs often require elaborate input data sets to be obtained from sophisticated CT-scan and/or MRI interpretations.

and drug-delivery devices. They can provide detailed physical insight and potentially optimal design solutions. endoleaks. cardio-pulmonary bypass systems.g. it is quite a challenging multi-step procedure. respectively. joint lubrication.and time-intensive solutions to patient-specific medical problems have to be demonstrated. e. possible stent-graft migration. hemodialysis machines. Building 3D geometric surface models from CT and/or MRI scanned image files is being widely used in biomedical engineering and other disciplines. Clearly. experimental methods are sometimes impossible to use or fall short in predicting local deposition concentrations of inhaled toxic solid particles. Steinman et al. experimentally validated computational cardiovascular fluid dynamics analysis.2. high-resolution imaging of a patient’s diseased system will allow for transient 3-D computer modeling of FSIs to simulate and analyze causes of the particular disease. can provide the surgeon with quantitative guidelines for better surgery planning. where to place a bypass graft. based on model geometries from Magnetic Resonance Angiography (MRA). (2003). which regions experience high pressure loads and hence wall stress concentrations. computer modeling and virtual prototyping are essential. and vapors.3. suitable stent-graft selection and secure placement. e..Chapter III: Analyses of Arterial Diseases 297 Specifically. including device implantation and organ/tissue transplantations. while Taylor & Draney (2004) reviewed experimental and computational methods in cardiovascular fluid mechanics applied to medical management.2 Image-File Conversion Steps Computed tomography (CT) and magnetic resonance imaging (MRI) are most suitable for obtaining good pictures of bones (hard tissue) and soft tissue. droplets. an artery. artificial valves and hearts. especially when the geometry of the object. References dealing with detailed computational fluid-particle dynamics of branching conduits include Zhang et al. etc. (2005). before such lofty goals can be achieved and hence computerized medical management will be widely accepted. (2003). to name a few. . much lower cost. at all stages of the development of medical devices. Section 3. etc.. illustrate the detrimental effects. should a stent be implanted or not. A second example is the computational fluid-structure-interaction simulation of a (stented) aneurysm to provide quantitative information on the need for stent-graft insertion.g. Other biomedical applications where CFD analyses play key roles include the respiratory system and device development.3 illustrated simple applications of medical diagnostics and designs which are expanded and discussed in more detail in Chapter V. and Lohner et al. stent-grafts. and come up with an optimal treatment in form of drug therapy or surgical interventions. for example. As a first step. Furthermore. Kleinstreuer & Longest (2003). 3. as needed in toxicology and health-effect studies. nano- therapeutics implants. Ultimately. However.

3. unsmoothed) Step 3 Magics 3D surface (STL file. Thus. each step is illustrated as follows. Gridpro.. smoothed) CFD mesh generation with ICEM. Step 4 CopyCAD. or tissue structure is complicated. For simpler geometries.298 Biofluid Dynamics organ. e.3. 3. Software: Photoshop (Adobe Systems Incorporated. and CAD software.1 Sequential multi-step procedure for image-file conversion to CFD Mesh. such as Geomagics. Step 1: Processing CT/MRI scanned images Goal: To obtain distinct grayscale-separated boundaries of the object. Photoshop Improved CT/MRI images (clear boundaries) Step 2 Amira or Mimics 3D surface (STL file. such as artery walls..1).g. San Jose. CA. Pro/E CFD mesh generation or 3D surface (CAD file. Using the human nasal cavity as an example.g.g. e. Original CT/MRI images Any image processing Step 1 software. e. iges format) other 3D modeling Fig.. USA). the necessary procedural steps and associated software required can be summarized in flowchart form (see Fig. 3. Matlab Toolboxes may be employed to generate CAD-like geometry files from DiCom image files. etc Inverse-engineering Inverse-engineering software. .

2). and make necessary modifications using image processing software like Photoshop. or cavities based on the varying tissue densities.2. View direction Fig.e. Step 2: Constructing 3D surface from CT/MRI images Goal: To obtain a 3D surface computer file from improved CT/MRI images .2. Those images are taken with a certain resolution. CT/MRI images often do not have well grayscale-separated boundaries of the organs.Chapter III: Analyses of Arterial Diseases 299 The original CT/MRI scanned data comes as a stack of images (Fig. To eliminate this problem. i. The final image for each slice was carefully examined and manipulated to assure good grayscale- separated boundaries as indicated with three image examples given in Fig.5mm. As indicated in Fig. body fluids.. 3.2 MRI slices and local image contour. Such unclear boundaries will lead to distortion of organ surfaces which have to be constructed later. However. it is recommended that CFD modelers examine each image with the help of research MDs or radiologists. for the present case of a human nasal cavity there are 69 MRI slices at intervals of 1.3. due to insufficient scanner resolution and unexpected tissue density changes. slice distances between each other from the start to the end of the organ. The images are gray-scaled to represent different organs. 3. 3.

3. Step 3: Smoothing surfaces Goal: To smooth the 3D surfaces Software: Magics Magics (Materialise. The smoothness parameter for a STL surface is often given as its maximum direction angle difference between any two neighboring facets. 3. CA) and Mimics (Materialise. Ann Arbor MI) was found to be the best software for 3D surface smoothing.3. The 3D STL surfaces are made up of three-sided “facets” or triangles (see Fig. MI) both are suitable for constructing 3D surfaces from CT/MRI images. the automatic smoothing function cannot accomplish the whole . The direction of each facet is given defined with its normal unit vector.3. and it is called feature angle β. with a maximum feature angle of β =160 ° .3). especially for rapid prototyping.3.3) from a stack of images (see Fig. They follow a very similar working procedure. the first-cut 3D surface of the human nasal cavity produced by Amira is too coarse. The computer file is in STL (stereolithography) format which is a widely used format in engineering. For this particular case study of a human nasal cavity. 3. Amira was used.2) in a minute. Its robust automatic surface generation routine produces rough 3D surfaces (see Fig. 3. Clearly. While Magics has its own automatic smoothing scheme. This does not reflect a realistic human nasal cavity surface and is also unacceptable for CFD mesh generation. Ann Arbor. 3. San Diego.300 Biofluid Dynamics Software: Amira or Mimics Amira (Mercury Computer Systems Inc.3 First-cut STL surfaces using Amira. Fig.

conversion from NURBS to STL is a one second job. if one wants to connect the human nasal cavity with oral airway models. Its maximum feature angle is only 35 degree. Such manual modification facet by facet is painstaking. or Geomagic Studio The 3D surface in STL format is made up of a great number of small triangular facets. but very effective. iges format Software: Imageware.. For example. Magics graphically indicates which facets have a bad smoothness quality and provides very convenient operation to make improvements locally. The relationship between STL surface and NURBS surface is the analogy between data points and their curve-fitted connections. It is different from the 3D surface in CAD format which is described by Nonuniform Rational B-Splines (NURBS). Step 4: Converting STL format to CAD file format (Inverse- engineering) Goal: To convert the 3D surface STL file to the CAD file. e. called inverse-engineering. Hence. it . Although the STL format file is acceptable for most mesh generators. CopyCAD. Thus. Fig. 2006). a CAD format file is still highly desirable because the CAD file will provide much more room for further geometric processing.4 Final 3D body surface of human nasal cavity (Shi.3.4 shows the final 3D surface definitions in STL format after smoothing with Magics. which is already within the range of good quality surfaces. 3. the perpetually running smoothing algorithm will not keep producing smoother surfaces. Interestingly enough. a proper approach was found to be a combination of auto- smoothing and user manual modification facet by facet. Specifically. Figure 3. the transfer from STL to NURBS is a quite challenging topic. but will lead to distortion or shrinkage of surfaces.g.3. Now the surface file is ready to be used in CFD mesh generators like ICEM or Gridpro.Chapter III: Analyses of Arterial Diseases 301 smoothing step because of mathematical limitations of auto-smoothing.

a certain airway length was added to the nasopharynx as an air outflow extension as well in preparation for future linkage to the oral airways. e. Windsor. human nasal cavity. it is recommended to use a structured mesh... Figure 3.. Step 5: CFD finite volume mesh generation Goal: To generate a finite volume mesh for numerical analysis Software: ICEM ((ICEM-CFD Engineering Inc. Berkeley. Plano. USA) are capable to perform inverse-engineering. For the geometries with highly- irregular geometric features.6 shows the final mesh for the human nasal cavities. Fig. generating tetrahedral elements with dense mesh resolution layers of prism elements would be more efficient and provides more flexibility. ON.302 Biofluid Dynamics would be very inconvenient to construct the geometric linkage between the two models if any one is in the STL format.3. . USA) ICEM has powerful functions to generate both structured and unstructured meshes for complicated biomedical 3D models.3. In this case.5 3D CAD surface model of a human nasal cavity (Shi. CA. CopyCAD (Delcam Inc. Imagware (UGS Inc.. TX.5 shows the 3D surface model in CAD (iges) format after completing inverse-engineering in Geomagic Studio.g. 2006). Quite a few software packages.g.3.g. Figure 3. 3. For the geometries with low-level complexity. e.. Research Triangle Park. Canada) and Geomagic Studio (Raindrop Geomagic Inc. NC. USA).. e. arteries and lung airways..

3.3. They incorporated an empirical nonlinear correlation for lumen pressure and local tube radius as well as minor loss coefficients for geometric non-uniformities. for species pathway analysis. such as outflow resistance. The major conditions were Poiseuille-type flow. 2002. 2004).. For illustration purposes the work of Wan et al. Asgharian et al. 2003. Alternatively. 3. (2002) and Steele et al. 2005). such dramatically simplified models require more empiricism in terms of input data sets.6 Unstructured finite volume mesh (Shi. pharmacokinetics modeling (Nestorov. junctions. Examples include global lung models for the evaluation of toxic particle impact (Phalen & Oldham.3 A Stenosed Artery Model for Surgical Bypass Planning As mentioned in Sect. 2003). Steele et al. 2006). Hofmann et al.3. and minor loss terms for the . algebraic global models or transient one-dimensional transport models have been proposed. and 1-D models for vascular cardiovascular planning (Wan et al.. Naturally. 3.g. changing cross- sectional areas due to the elastic tube walls. 3.. a straight human common carotid artery or a stenosed aorta with bypass graft.Chapter III: Analyses of Arterial Diseases 303 Fig. e. 2001. (2003) is discussed. 2002. experimental correlations. realistic computer models for medical planning are presently too slow and too expensive for routine use. focusing on a 1-D computer model describing pulsatile flows in blood vessels.. and stenoses..1. and measured transfer functions. Rodgers.

e.3. t ) = p0 + 1 −  (3. t ) . 1976 for stenoses) and tube junctions. i. t) is prescribed and at the outlets Q ( z = L. The inflow rate Q(z=0.3) h 1 − γ r0 where ∆p = p − p0 . The coupling between internal pressure p(z.3.3.3. r E r − r0 σ θ = ∆p = (3. selections of the “minor” loss coefficient are critical in providing accuracy to the 1-D modeling approach by realistically incorporating local obstructions (see Seeley & Young. The tube loss term is he = K   . Poisson’s ratio γ = 1 / 4 and E is Young’s modulus. t ) is given by Laplace’s law for thin-walled tubes of base radius r0 and wall thickness h . It is noted that E [r0 ( z )] is highly nonlinear for arteries. t ) = p L / R . p( r = r0 ) = p0 .4) ∂∀ 2hE Solving for the transmural pressure (see Eq.1) ∂t ∂t ∂Q ∂  Q2  A ∂p ∂ 2Q Q (z-momentum) + (1 + δ )  = − + γ 2 + he ∂t ∂z  A ρ ∂z ∂z A (3. where p L = p L (t ) and R is the lumped downstream resistance. t) is the axisymmetric cross section and for Poiseuille- Q velocity profiles δ = 1 / 3 .304 Biofluid Dynamics stenosis plus graft-to-artery junctions.3. where  2L   Q  2 K = ∆ p /  ρ / 2   .(3.   A   Clearly.2) where A=A(z.5) 3 r0  r where r0 / r ≡ A0 ( z ) / A( z. t) and wall stress σ θ ( r .3)) yields 4 Eh  r0  p ( z . 1973): ∂A ∂Q (continuity) + =0 (3.3. The resulting flow equations in terms of Q(z. Olufson (1999) developed a curve-fitted correlation for arteries as: ..t) are (see Hughes & Lublieu. following the arterial volume compliance as: ∂p 3r ≈ 0 (3.

3. 2004). Three cycles were simulated to avoid start-up effects on the flow field.2) with the prescribed flow rate at the inlet and resistance boundary conditions at the exit. following the given input pulse expressed as Fourier series. b = 2 × 107 g /(cm ⋅ s 2 ) . the 1-D approximate method can be used to predict flow distributions and pressure waves in cardiovascular bypass grafts. the lag of volume flow rate at the outlet does not appear in 3-D simulations.3..7.4cm and 16cm. were also given by Wan et al. 1-D and 3-D simulation comparisons for various volume flow rates and pressure waves in a straight vessel via s are displayed in Fig. The transient 3-D simulation results were achieved by solving continuity and momentum equations with the commercial finite volume software CFX. However. In the latter case.53cm −1 .65 × 105 g /(cm ⋅ s 2 ) .Chapter III: Analyses of Arterial Diseases 305 Eh = a + be cr0 (3. In summary. The large lag of both volume flow rate and pressure wave at the outlet can be observed in the 1-D modeling. Vignon and Taylor.3.6) r0 ( z ) where a = 8. which may be valuable for rapid evaluations of alternate treatment plans for patients with specific cardiovascular diseases.g. and other complex. the inlet velocity profiles were determined by the analytic expression of transient developed flow in a straight tube. 3. The dimensions of this vessel were chosen to approximate the human common carotid artery. the pressure wave at the inlet obtained from the 3-D simulation agrees very well with that from 1-D modeling with similar inlet flow waveform and outlet resistance boundary conditions.3. The results for 1-D finite element simulations were obtained by Wan et al. . Clearly. such as a porcine thoraco-thoraco aortic bypass graft. 3-D simulations are necessary for accurately analyzing realistic blood flow and associated particle dynamics as well as fluid-structure interactions in human blood vessels.1) and (3. The pressure at the outlet was specified as p(t)=Q(t)R with R=3000 dynes·s/cm5 . which perhaps is set up manually in order to exhibit the shifted wave because the same phenomenon is not shown in other 1-D simulations (e. secondary flows. Comparisons between 1-D and 3-D solutions for more complicated geometries. Clearly. They showed that the peak differences for time- dependent flow rates were on the the order of 50% in the native aorta. realistic flow phenomena. but less than 10% in the bypass graft. Thus. where the radius and length of the straight circular pipe were specified to be 0. the 1-D approximation of blood flow ignores flow separation. considering the mass balance at each time level. (2002). (3. The minor damping of the propagating pressure waves are shown in both 1-D and 3-D simulations. (2002) by solving Eqs. and c = −22. respectively.

3.6 1 time (sec) Fig.4 0.306 Biofluid Dynamics (a) Volume Flow 80 Inlet (1D) Inlet (3D) Outlet (1D) Flow Rate (cc/sec) Outlet (3D) 60 40 20 0 0.2 0.7 Variations of volume flow rate and pressure waves in a straight circular pipe.2 0.8 1 time (sec) (b) Pressure 200 180 Inlet (1D) Inlet (3D) 160 Outlet (1D) Pressure (mmHg) Outlet (3D) 140 120 100 80 60 40 0 0.6 0. key biomechanical factors and their .4 0.4 AAA-Rupture Prediction Focusing on a representative abdominal aortic aneurysm (AAA). Hence to assess the risk of AAA rupture. 3. it is of interest to determine quantitatively whether it is necessary to repair AAAs via open surgery or endovascular aneurysm repair (EVAR).

1 (Li.8).Chapter III: Analyses of Arterial Diseases 307 threshold values. we start with the original Laplace equation.85 p sys σ max = 0. Equation (3.3. where l p and l a are the AAAA.5 cm (male). the area ratio AILT . indicating possible AAA rupture.5cm/year Mechanical stress Threshold: 0. r is the radius. 2005a). pr σ= (3.1 Parameters used in AAA rupture prediction (see Fig. (3. and t is the wall thickness.8).3. i. 1.5 d AAA .0123 ( 0.7) ct where σ is the average wall stress. max β= = 0.125 where σ max is the maximum wall stress which occurs most frequently at a location two thirds of the maximum AAA diameter. Table 3.max la distances from the center point O to the posterior and the anterior (Fig.7) greatly overestimates or underestimates actual aneurysm wall stresses because of the many underlying assumptions. predictor of the maximum wall stress in common AAAs (see Fig. In order to provide a useful. which correlates (blood) pressure with the average wall stress (see Sect. 5..7) was extended based on observed clinical evidence and computational analyses (Li & Kleinstreuer.44MPa Intra-luminal Thrombus Threshold: V ILT / V AAA = 0.3.8) t 0.5). 3.006 in [MPa] (3.3. Maximum diameter Threshold: 5.63 β 0.3. have been surveyed as summarized in Table 3.e.65 L AAA Saccular index High risk for AAA with large curvature In order to estimate the actual wall stress.62 Diastolic pressure Threshold: 90 mmHg and 105 mmHg as the middle and high risk levels for AAA rupture Wall stiffness Rupture threshold: stiffness begins to decrease Asymmetry High risk: asymmetry index d AAA .3. p is the pressure load. accurate and easy-to-calculate. 2005). The functional form of +19.3.max ) (1 − 0. Eq. max lp α= .3.68a ) e 0.0 cm (female) Expansion rate Threshold: 0. while c = 1 is for cylinders and c = 2 for spheres. 3. . the asymmetry index β = .

the transverse area can be approximately calculated as πd AAA.max la is the distance from center point O to the AAA anterior side lp is the distance from center point O to the AAA posterior side Figure 3.3.max -measurement plane (see Fig. A AAA.max = (3. max H AAAA.3.3. max = AAAA.max [cm]. 3.level Lumen area Onset of AAA Infrarenal artery (dashed line) dAAA. max (3. then.max and AILT .max plane thrombus (ILT) area la lp O H t Iliac bifurcation d AAA . the ILT area is given as AILT .8). p sys is the systolic blood pressure (mmHg).308 Biofluid Dynamics 3.max are the transverse areas of the AAA and intra-luminal thrombus (ILT) at the d AAA.max may be calculated similarly.9a) 4 Renal artery AAA neck dAAA.max . d AAA. respectively.3. where H is the in-plane axis normal to the d AAA. Specifically.8 Geometric parameters of an AAA.8).max − Alumen . The lumen area Alumen .3.max -location. and t is the wall thickness [mm] at the location of d AAA.level Intra-luminal projected onto the dAAA.max is the maximum AAA diameter (cm).max . For imaging techniques not providing area measurements.9b) .

AAA-wall deterioration with time. whereas the Laplace equation generates a maximum error of 85. the breaking (or tensile) strength is closely related to age. AAA  b  VILT  =    (3. • Residual stress is neglected because its magnitude is much smaller (3%) than the mechanical stress caused by the blood pressure in the artery wall.10) σ break . AAA-wall properties alter gradually with age.ref   V AAA  . collagen-to-elastin ratio. for seriously distorted geometries the evaluation of maximum wall stress magnitude (and location) is too complicated to predict with this simple equation. Based on a literature review (see Li. using data from ten different clinical and numerical AAA models.3.3. sustainable wall stress is a function of time.max is in [mm].3. Clearly. and wall thickness.8) integrates the stress concentration effect caused by asymmetry to same extent.2892  d AAA. (3. Notes: • While Eq.3.5%. Even though the ILT close to the luminal surface is most newly formed and tends to be “older” away from the luminal surface toward the AAA wall. (3.. but also correlates the effects of ILT and AAA asymmetry. To predict AAA rupture more accurately.6%. assessing time-dependent wall deterioration is very important. i. 2005). the maximum error using the new wall stress equation is 9. it was found that the use of mean ILT properties as opposed to the exact patient-specific parameters would result in a maximum error of only 5% in wall stress predictions. wall stiffness. A nondimensionless model to estimate wall deterioration associated with time could be expressed as: a σ break . diameter. i. the average wall stresses were calculated.3. One can see that Eq. In order to test the broader validity of the new wall stress equation.normal  E p . (3.max  t = 3. (3.3.9c)  2  where d AAA.Chapter III: Analyses of Arterial Diseases 309 If the wall thickness t is difficult to obtain from CT-scans. As shown in Table 3. AAA  E p .e. and the results compared with Eq.9   in [mm].e.8). and ILT content.8) not only represents the nonlinear correlation between wall stress and blood pressure. • The ILT material property is assumed to be uniform. it may be approximated with a curve-fitted correlation −0.2..

32 0. and the annual AAA-growth rate.max t α β Stress equation (Eq. 120 6 0. where the exponents a and b may be deduced from clinical data sets.2% et al.19 0 0.343 2.4 0.310 Biofluid Dynamics where σ break . a reference pressure-strain elastic modulus.19 0 0.9% Vorp et al.0% 0.7% 0.335 4.4% et al.2 Comparisons with the new wall stress equation.208 9.37 0.25 16.7% 0. σ break .2 0.7 0.19 0. and AAA volume.21 8.3 0.3% (1998) Raghavan 115 5.54 0.299 6.ref .1 0. 2006).798 85. AAA .15 0 0. E p.9 0.10)).3.4 0. Table 3.412 4.43 0.19 0 0.0 0. (3.3 0.normal .104 0 0.278 7.11) i where α i are the weighting coefficients and Pi are the dimensionless biomechanics quantities (see Kleinstreuer & Li.5 0.3 0. .362 5.7% Wang et al.05 0. have been calculated based on CT measurements and expressed in terms of a time- dependent severity parameter (SP). 128 6.184 0.7% 0.5% 0.7% Li & Kleinstreuer 120 5.3. stress equation AAA model p dAAA.23 0. 3.19 0 0. AAA . respectively. (2003.43 0.175 0.5 0. the pressure-strain elastic modulus.449 21. New wall Laplace Max.6% (2005) AAA-rupture prediction.7 0.1% 0.2. In order to quantify AAA-rupture prediction for individuals. V ILT . E p .2002) 130 5.281 12.296 5.7% Thurbrikar 120 5.86 0.35 0.33 0.46 6.282 48.5 0.158 0 0. V AAA are the breaking strength in the AAA wall and normal abdominal aorta.2% 0.209 9.3 0.1) [mmHg] [cm] [cm] σ 3. the ratio of maximum AAA diameter to normal aortic neck diameter. SP = ∑ α i Pi (3.0% 0.389 5.86 0.277 20.34 3.2.28 0. (Eq.4 0. most notably the wall stress (see Eqs.2) (cylinder) [MPa] Stress Stress [MPa] Error [MPa] Error Fillinger 120 6.4% (2002) 155 6. ILT volume.319 3. key biomechanical parameters.5 0. (2000) 188 5.9% 0.15 1 0.3 0.8) and (3.1% et al. (2000) 120 5.

rather than requiring quantitative problem solutions. i.3. i.. leading to partial blood vessel occlusion (b) blood clot (or thrombus) formation leading to sudden downstream vessel occlusion (c) embolization . The latter is shown in both constant and time-dependent modes.2 indicates a low rupture risk. 3. critical assessments.1 Describe and illustrate: (a) stenotic development.4 HOMEWORK PROBLEMS Note: The homework assignment for Chapter III are more qualitative.70 surgical intervention should be considered. intimal thickening. Fig. and where SP(t) > 0. where SP(t) < 0.. i. literature reviews.” supported by bar-graphs for the two commonly employed predictors. 2005 and Kleinstreuer & Li.Chapter III: Analyses of Arterial Diseases 311 The severity parameter can vary between 0 and 1 (or 0 and 100%). 3.2 SP(t) < 70 a moderate to average rupture risk.3.9 depicts a hypothetical sample SP(t) for patient “Johnson.e. 2006). 0.9 Hypothetical example with time-dependent wall-stress threshold (Kleinstreuer & Li. AAA-diameter ratio and AAA-wall stress ratio.e. Figure 3. and comparisons..e.3.

5 In conjunction with Section 1.e. and time constants important in blood particle transport. i. define and discuss all Reynolds numbers. and wall migration. 40%.3 Graph Equations (3. 3.4. 3.3) which are supposed to indicate the onset and progression of arterial diseases.3).4 The hematocrit. t ) for a proper blood rheology.3.2 & 3. 3. say.1.8- 3.11 and suggest an “optimal” end-to-side configuration.4.6 Update the discussion of “near-wall forces” (see Sect. 3.11 Questions concerning aneurysms: (a) Who was the most famous person of the 20th century who died because of a sudden AAA rupture? (b) Why do the platinum-coils in a saccular brain aneurysm reduce the local (radial) blood pressure? (c) What are the differences and functions of stents in brain (or coronary) arteries with aneurysms and AAAs? (d) Why is the total cost for EVAR almost twice as much as that for open surgery? .7 Describe physically and mathematically blood-cell and vessel- wall interactions.3. resuspension. variable wall stress in critical junction areas.3). 3. 3. such as particle lift.1.23) for different sets of model parameters (see Fig. 3. 1. Discuss the merits of these two hypotheses. 3.1.1. Develop a conceptual model which takes into account r H = H ( x .4.2) with an emphasis on particle lift as important for micro-particle deposition. and (b) excessive. 3.4.2 Discuss the causes/differences between “atherosclerosis” and “intimal hyperplasia” and give examples of their occurrences. 3.5.3..2c) and (1.312 Biofluid Dynamics 3. attachment.10 Comment on the usefulness of the designs shown in Figs.1.8 Critically review the pros & cons of hemodynamics wall parameters (see Sect.1. differs between people and changes in the blood vessels with location and time. DAIH (distal anastomotic intimal hyperplasia) formation has been explained via two contrasting viewpoints: (a) aggravating multi-pathway particle-hemodynamics (see Sects. rolling. RBCs by volume. deposition. length scales.

PhD Dissertation. New York. Isn’t the inlet pressure needed as well? (f) What are the highest risks for EVAR vs. E. F. .9)) may be applicable. and comment.5c shows the outlet pressure waveform. (2002) “Transport Phenomena.2.. etc.G. (2002) Journal of Endovascular Therapy.” John Wiley.J. R. and computer modeling. and (c) collagen in arterial walls. stress-strain relations for arterial wall are nonlinear. e. 17: 213-224. hyperelastic. E. and Lightfoot.. Bird. and Miller. surgery. 9:458-463 Buchanan. (2004) Journal of Aerosol Medicine. N. R. as used for diagnostics. etc. W. ultrasound.13 Write a brief report on imaging/scanning techniques..11) be derived and is it the best indicator for possible AAA rupture? (e) Figure 3.D. J. properties.2. and functions of (a) SMCs. viscoelastic.. Menache M. however. B. J. why is it recommended that old AAA-patients undergo EVAR while relatively young ones should opt for open surgery? 3. List in detail more realistic constitutive models. What is their change in properties over time? 3. M. (d) How can Equation (3. Raleigh.. and Prinssen. Stewart...Chapter III: Analyses of Arterial Diseases 313 (e) What exactly causes aneurysm rupture and does this criterion change with time? (f) In general. in general. such as X- rays. MRI/MRA..12 Discuss the make-up. 3.3. how exactly are wall movement and blood flow in changing conduits calculated over time? (c) For low pressure loads with small variations Hooke’s law (see Eq.g. open surgery? References Asgharian B. Blankensteijn. Mechanical and Aerospace Engineering Department. NC. CT/CAT. (b) elastin. North Carolina State University. (2000) Computational Particle Hemodynamics in the Rab- bit Abdominal Aorta. NY. (3.14 Mathematical AAA modeling questions: (a) How are wall structure and fluid flow equations coupled? (b) In computational fluid-structure interactions of AAAs.

). W. Colman.. F. Ferziger. Dalman. C. (2000) Computers and Fluids. (1987) The rheology and tube flow of blood. R. Chervu.. and Kleinstreuer. and Peric. and Tsuji. Lippincott Company.. NY. Series B. H. C. Springer-Verlag. Corash. and Clowes. H. Jr. L. V. New York. H. 29:695-724. Philadel- phia. H. and Particles. J...314 Biofluid Dynamics Buchanan. Missouri. Hirsh. J. (1998) “Multiphase Flows with Droplets and Particles. R. B. 22: 1753-1777. Marder. J.. M. McGraw-Hill. M. Buchanan. Louis. Grace.. Drops. L. W. (1974) Journal of Colloid and Interface Science. W. (1977) Blood. Davies. NY. and Moore. Berlin. (1967) Chemical Engineering Sciences. M.C. and Friedman M. and Schroter. Quality Medi- cal Publishing Inc. and Taylor. S.. Lipencott. (1994) Over- view of Hemostasis. Crowe. V.... In: Handbook of Bioengineering.). 370: 53-72. and Weber..” Academic Press. .. and Hirsh. M. A. (1998) Journal of Biomechanical Engineering. (1977) Journal of Biomechanics. Philadelphia. Coimbra. M. 3-18. FL.. J.R. W. G. and McLaughlin.. (1978) “Bubbles. M. G. and Salzman. Marder. W. R. M.. (1993) Infrainguinal Revascularization Procedures. In Hemostasis and Thrombosis. (eds. J. L. (eds. M.. Hirsh. Sommerfeld. R. Cherukat. (1994) Journal of Fluid Mechanics.. Chien (eds. In: Basic Data Underlying Clinical Decision Making in Vascular Surgery. 263:1-18. and Rangel. J. R. J. (1983) Laboratory Investigation. R.R. J.. J. 49: 327-332. (1990) Gynecology and Obstetrics. New York. M. 1224-1237. and Brenner. (1971) Proceedings of the Royal Society of London. In: Hemostasis and Thrombosis: Basic Principles and Clinical Practice. Cox.. 120: 446-454. Jr. St. Caro. 3rd Ed. Reidy. B. Cokelet G. 171: 433- 447. Dahneke. and Harvey. Clowes. R. 49:71-87. Tan. J. (1999) Computational Methods for Fluid Dynamics. 2nd ed. B. J.). Taylor.. 10: 561-568... R. R.. J. C. Porter.. Salzman (eds.K.G. R. Skalak and S. C. L.. and E. 177:109-159. (1994) Mechanisms of thrombosis in atherosclerosis. Marder. Y.. W. 48:520-522. Colman. W.). and Comer. Salzman. Clift. E.. R. E. W. J. A. (1998) Journal of Fluid Mechanics. H.. Kleinstreuer. P. M. Colman. L. R. Ehrlich. A. C. V. E.” CRC Press. W. Boca Raton. Fitz-Gerald. J. J. M.

. and Ku.. H. R. M. M. and Brenner. He. K. 22: 653-660. V. V. Asgharian.. Cox. L... Keynton. 31:1008-1017. S. Karino..” PhD Dissertation. Kusaba. Sims... (2004) Lancet.. D. (1999) Critical Ischaemia.. and How. Kleinstreuer. Schreck. K. Wassmer. 89(4):514. H. Humphrey. Fry.. S. T. Gobin. T. V. Kina. J. J. Humphrey. R.. Hofmann. (1967b) Chemical Engineering Science.. Knez.Chapter III: Analyses of Arterial Diseases 315 Fisher. Rodway. 279: 413-445. Mechan- ical and Aerospace Engineering Department. C.. K. L. G. A. and Brenner. A. (1996) Journal of Biomechanical Engineering. M. and Harris. How.. (2004) ASME J.. and Lubliner. Aerosol Sci. 9:20-26. Rowe.. T. A. Goldman. 113:123-131.. Hyun. and Organs. S. 18:161- 170. (2000) Journal of Vascular Surgery.. Schmitz- Rixen. Brown. 33: 219-235. Biomechanical Engineering. Gilling-Smith.. Kwong. W. G. M. G.. Winter. Jaschke. V. G. Kamori. Longest. (2002) British Journal of Surgery. I.. Hyun. R... Goldman. NY. S. S. L.. Brennan. (1998) “Transient Particle-Hemodynamics Simulations in Three- Dimensional Carotid Artery Bifurcations. S. S.C. and Rittgers. 4:843-8.G. J. L... 22:637-651. X.P. G. and Goldsmith. M. 92:1006-1015. and Brunkwall. and Okadome. 27: 45-50. D.” Springer. and Nerem. J. E.. (2004) “An Introduction to Biomechanics. L. J. Series B. R. New York. J. T. (2002) “Cardiovascular Solid Mechanics: Cells. How.R. J.. Cox. and Winkler-Heil (2002) J. Powell. R.L. . P. H.. A. P. N. (2004) European Journal of Vascular and Endovascular Surgery.” McGraw-Hill. V. B. 118: 74-82. Raleigh. R.. G. 22: 165-197. Thompson.J. (1977) Philosophical Transactions of the Royal Society of London. 123: 464-473.. NY. C. P. L.. Tissues. A. Gilling-Smith. J. U..D. S. (1968) Circulation Research. Inokuchi. G. L.” Springer-Verlag. (1973) Mathematical Biosciences. (2001) Journal of Biomechanical Engineering. NC.. Fox. Harris. G. R. Helmlinger... L. S.. Evancho. North Carolina State University. and Harris.. MA. (1991) Jour- nal of Biomechanical Engineering... L. J. P. T. Gawenda. (1982) Surgery. Hughes. R.. C. Boston.M. and Chen.D. M..T. N. Geiger. Kirkpatrick. (1996) “Human Physiology. S. 126:188-195. (1967a) Chemical Engineering Science.W. and Delange... Greenhalgh. T.

(1996) ASME J. Raleigh. P. Raleigh. (2003). C. Truskey. Buchanan. Leuprecht. M. 29:1605-1614. Lambert. New York. and Kleinstreuer.. Ku. (1985) Arterioscle- rosis. Duran. D. Kleinstreuer.. Steinman. N. Hyun. and Longest. WIT Press. Perktold.” PhD Disserta- tion. 31: 69-83. Lei. and Schima. Kleinstreuer. UK... Ashurst. Berk. and Hobson. N. Kunov.. and Archie. Biomechanical Engineering. Ku. Longest. DeFouw. 33: 209-213. 17: 60-65. A.” PhD Dissertation. R. J. Z. W. P... and Horrocks.. C. Lei. P. 118: 158-164.. North Carolina State University.. P. and Archie.S. Prosi. H. and Li. Biomechanical Engineering. D. Biomechanical Engineering. Mechanical and Aerospace Engineering Department.Experiment and Modeling. C. 29: 1-64. R..). S. N. P. J. K.. Giddens. Z. S. . P.J. Archie. Zarins. and Li. 35:225-236.. R. Lei. C.. Budd. J.316 Biofluid Dynamics Kissin... NC Kleinstreuer. C. D... Kleinstreuer. W. (2005) “Patent Application for Computerized Medical Management Software.. C. J. (2002) Journal of Biomechanics.. Raleigh. A. Advances in Fluid Mechanics Series. and Glagov.. M. A. T. Li. M. Pappas. (2001) Critical Reviews in Biomedical Engineering... G.. D. A. (2006) “Analysis and Computer Program for Rupture-risk Prediction of Abdominal Aortic Aneurysms.. A. and Archie. “Particle-hemodynamics Anal- yses of End-to-side Anastomoses: A Computational Comparison Study.. Li. C. (2003) “Two-Phase Flow: Theory and Applications. K. J. North Carolina State University. C. D.W. M.” Biomedi- cal Engineering Online. Tura (ed..” North Carolina State University. and Ethier. (2005) “Computational Analysis and Simulations of Fluid-structure Interactions for Stented Abdominal Aortic Aneurysms... NC. Kleinstreuer.” In: Vascular Grafts . M. (2000) Journal of Vascular Surgery. (2005a) Annals of Biomedical Engineering... J. Williams. NY. Z. J. (1996) J. D. Kleinstreuer. M. (1999) Euro- pean Journal of Endovascular Surgery. M. C. NC. Z. Kleinstreuer. J. C. Kansal. 29: 399-434. (1996) Journal of Biomechanics. P.” Taylor & Francis. (1995) “Computational fluid dynamics and optimal design of bifur- cating blood vessels.. W. 118:506-510.. (1997) Annual Review of Fluid Mechanics. Kleinstreuer. 5(3):293-302. W. 119:343-348. C. (1997) ASME J.W. Trubel. O. M. Lei. N. P.

and J. R. Surgery. NC.. C. Mei. R. 270: 133-174.. I. M. S. P.. (1993) International Journal of Numerical Methods for Heat Transfer and Fluid Flow... (2004) Computers and Fluids. Malina M.. (2002) Journal of Biomechanical Engineering. (2000) Progress in Energy and Combustion Science. Nestorov I. 43: 637-650... C. Z. 26:161-223. North Carolina State University. Ivancev K. 54:283-285.. Cebral. 28: 1281-99.Chapter III: Analyses of Arterial Diseases 317 Li.. 30:1579-1582.R. E. (1994) International Journal of Multiphase Flow. and Kleinstreuer. J. H. R. Longest. Giddens. Zarins.. Loth. S. Jones. Longest. 127: 662-671. Raleigh. and Bassiouny.. (1987) Physics of Fluids.. J. D.. Meter.. J. and Fads. Loth.. Mechanical and Aerospace Engineering Department. C. 119:187-194. W. S. O. Nassar. R.” Ph. P. Maxey. Maxey. J. P. and Brunkwall J.. Z.” MS Thesis. F. Biomechanical Engineering.. and Zarins. H. .. M. Kleinstreuer. Numerical Methods in Fluids. and Kleinstreuer. Lindblad B. Giddens. J. H.. (2000) Critical Reviews in Biomedical Engineering. 124:44-51. (2003) Clinical Pharmacokinetics. Foreman. (1999) “Computational Hemodynamic Simulations and Comparison Study of Arteriovenous-access Grafts for Hemodialysis. Loth. 3:3-14. S... Ferguson.. (2004) “Computational Analyses of Transient Particle Hemodynamics with Applications to Femoral Bypass Graft Designs. 27:624-31. L. 33: 577-601. NC. (1998) Journal of Vascular Sugery..W. (2003) Int. P. N. Biomech. P. Lánne T. 25: 843-858. 26: 883-889. (2000) Journal of Medical Engineering and Technology. Jones. Lonsdale. C.. C. B. and Kleinstreuer. P. Longest.D. F.. Michaelides. and Buchanan. (1997) Journal of Fluids Engineering. 42: 883-908. and Andreotti. (2005b) ASME J. S. Glagov.. P. P. (2000) Annals of Biomedical Engineering. J. Engineering.. D. W. Glagov. R. North Carolina State University. Kleinstreuer. Soto. 24: 102-110. Longest. 119: 233-247. Longest. R. K. Riley (1983) Physics of Fluids. Raleigh. O. Bassiouny.. E. C. K. (1984) Aust. J.W. MAE Dept. and Burgess. 20: 211-232. (1997) ASME J. C. Longest. McLaughlin. R. P. E. Dissertation. (1994) Journal of Fluid Mechanics. J. J. F. 28:141-147. (2003) Medical Engineering and Phys- ics. Yim.E. Lohner. Miller.. W.W.. D.

and Sugimachi. S... F.. M. C. 340:115-126. 25:243-285.. Onohara. D. Rhie... Pappas. and Wright. Richardson. D. R. B. C. 30: 923-931. W.A. and Batson. and Born. T. R. V. (1989) Lancet. 5: 413-418.A. and O’Rourke.. B.. 93:809-817. K. Young (1976) Journal of Biomechanics. D. C. F. M. W. P. Rodgers. Yao.R. and D. J. K. Hobson.. Raleigh. J. Ku. 314:488-500.R.. Shi. Ross.. (1999) American Journal of Physiology. Schima. 114: 112-118. P. M. (1997) Critical Reviews in Biomed- ical Engineering..J... D. V.. J. (1999) New England Journal of Medicine. G.. Czerny. C. (1986) New England Journal of Medicine.. J. 8:1-10. M. R. Stonebridge.. P. Resch T. Prescott.V.S.. North Carolina State University.. of Biomechanical Engineering.C. Endovasc.. (1997) Journal of Vas- . K. D. K. P. University of Iowa. Steinman. Okadome. (1999) Eur.. Ross. Steele. L. (1983) AIAA Journal. G. J. G. Norley. NC. and Johnston. (1998) Cardiovascular Sur- gery.. J. Sottiurai. J. Vinh. R.S. V.. 94: 1259-1276. W. Experimental and Clinical Principles. and Padberg. C. S. Davies. P. and Chow. 9:439-448. Perktold. Silva. (2006) “Analyses and Simulations of Inhaled Particle Uptake and Droplet Spray Dynamics in Representative Human Nasal Airways. Neuroradiology. Shyy. Goldberg. Hughes.. Raghavan M (2002) “Lecture notes on cardiovascular bio-solid mechanics section. Phalen R. Lánne T. Trubel. J.. 24:559-566.. (2005) “Biomedical Image Analysis. and Holdsworth. M. W. Lownie.. (2003) IEEE Transactions on Biomedical Engineering. N.W. (1999) International Journal of Angiology. Polterauer. Z. J. (2005) Journal of Pharmaceutical Sciences. C. Flinn. H. B. Steinman. R. T.” PhD Dissertation. FL.. Surg..F.. Leahy.W..318 Biofluid Dynamics Nichols.M. B.” CRC Press. and Taylor. J. W. R. N. 21: 1527-1532. T. W. M. (1991) Annals of Vascular Surgery.. Ethier. Wan. 2:941.C. Sonesson B. (2003) Am. 6: 19-26. M.R. and Topoleski. and Oldham M. and Rowland. Jamil. P. R. Sottiurai. Yamamura. W. L. H. 28: 889-894. Milner. Cobbold. Karner. B.S. Thakur.” Dept. Ojha. and Ivancev..J.. D. S. S. Meyers.” Lea and Febiger. Boca Raton. Olufsen. R. 50:649-656. (1983) Surgery.. PA. Noori. S. (1992) AIAA Journal. M. D.. J. 18:191-200. (2001) Respiration Physiology.P. M.. Philadelphia. 128:119-130. Rangayyan.N. Scherer. R. Salunke.... K..T... and Ruckley. T. C.J. (1998) Journal of Vascu- lar Surgery. T... (1993) Journal of Biomechanical Engineering. (1998) “McDonald’s Blood Flow in Arteries: Theoretical.. Seeley.. 276: 257-268..A. V. Lee. A.

D. D.D. M. in: Hemostasis and Thrombosis. (1999) Annual Reviews of Biomedical Engineering. M.. J. T. Vasc. and Hanratty. B.. and Towne. Spicer. H. Vignon. F.. J. Feijoo..T. 79: 348-354.. (1982) “Experiments with higher-order finite difference formulae. (2004) Annals of Biomedical Engineering. J.K. R. J. and Wiles. (1987) Journal of Vascular Surgery. 3: 104-114. M. G.A. and E.. Wang. Woodard. Biomechanical Engineering.B. O.S. H. (1987) Eur.. S. 32(7): 947-960. J. J. Giddens. F. (2001) Cardiovascular and Interventional Radiology. Philadelphia. Zheng. B.. R.Chapter III: Analyses of Arterial Diseases 319 cular Surgery. S. S. J. (1994) Anatomy and structural organization of the platelet.. S.. D. and Glagov. NJ. 127:553-563. 5: 413-420. Tang. (2004) “Transport Phenomena in Biological Systems.. Maruyama. S. M. P.F. M.B. and Taylor.P. S. 7: 147-163. Young. and Glagov. and Taylor. White. Ascer... Surg. and Raithby.A.).. (1997) Int. T.. M. J. J.. W. (2001) . J...-N.J.J.R. K.. L. Van Doormal. and Yuan.. Salzman (eds. C. Iwai.. Womersley. Loth. K. Marder. and McLaughlin. S. 5:195-206. Yuan. pp.. P.. (1993) ASME J. A. G.. J. Sicard. R. 36: 197-231.. Bruneval. Yang. Cox. and Ishi- gaki. C... (2002) Computer Methods in Biomechanics and Biomedical Engineering. S. and Cox. E.. A. D. Suzuki. J. C. Samson. Veith. R. Loh.. Wan. C. Multiphase Flow. J. Report No.. Colman.J. S. AERE-R 10493. N. G... Bassiouny. D. 115: 104-111. White. Ponzio. K. et al.D. 26: 543-550. 39: 361-374.. Physiology. McFarland. UK: AERE Harwell. Zegdi. G. R. C. and Katz.. and Chester. McFarland... Truskey. Thompson. (2004) Wave Motion. Taylor. and Ku. 1:335. S. P.397-413. Ishiguchi. 10: 299-329.. I..A. (1986) Journal of Vascular Surgery. P. and Fabiani. R. D.A. Gupta. C. 24:94- 98. T.. Steele. S.R. and Draney..Lippencott.. D. T. (1984) Numerical Heat Transfer. J. White-Flores.. Zatina. R. Zarins. (1955) J. Upper Saddle River. W. Zarins. Squires. Hughes.N... K. (2004) Annual Review of Fluid Mechanics.. R. S. J. A. P. 37: 1529-1536...K. I. Saffitz. Taylor. Ku. Jones. C.A. Scher.. G. P.H.. F.. 23:749-763..” Harwell. A. (1992) British Journal of Surgery.F.E.. (1991) AIChE Journal. K. Hirsh. C. J..” Pearson Prentice Hall. N. Lajos. N. Taylor. B. J. V.. Q. Kawatsu. Wootton.. Chen. Giddens. I. Strohband.A.

. .F. Kleinstreuer. C. J. 36: 211-233.S. Donohue. Zhang. J. and Kim.. Z.320 Biofluid Dynamics ASAIO Journal. Aerosol Science. 47:329-332. C. (2005).

. respectively. Nevertheless. some intriguing convective and diffusive transport phenomena occur in the other organ systems as well (see Fox. 1996 . I nteractive or gans t hat perform primary functions can be grouped into systems (Table 4. They interact to pr ovide proper blood fl ow and blo od co nstituents.e. ep ithelial.. and fibrous. musc le. In terms of sur face area.1. are of special interest in bi ofluid mechanics. and regulatory systems. Ende rle et al . 1..e. nervous. t he largest or gan i s the ski n whi ch has var ious functions provided by al l f ive pr imary t issues. Chapter IV BIOFLUID MECHANICS OF ORGAN SYSTEMS Organs ar e gr oupings of pr imary t issues.. i. i. t he kidneys and the liver. this chapter pr ovides s ome background inf ormation on th e re spiratory sys tem. e. T ruskey et al . • large local surface areas. conn ective. • elevated membrane/wall permeability because of controlled openings via carrier-protons and ion-channels (Sect.1. 2000 .1. as we ll a s t he ur inary a nd regulatory systems. respiratory. i. As already alluded to in Chapter II. Clearly. i ncluding pul monary problems. the transport of fluids and species from o ne do main to a nother ac ross membr anes of ca pillaries and tub ules takes place on the microscale. due to: • driving forces such as hydrostatic (blood) pressure and concentration gradients l eading to o smotic p ressure a nd/or di ffusion ( see Sect .2). The first three. int o ana tomical and fu nctional units. circulatory. 2004 .2). among ot hers).. often enhanced by microvilli. W hile Chap ters I I and I II dea lt wit h various a spects o f the c irculatory s ystem. a strong knowledge base and elevated skill level of Chapter I and Appendix A mat erial are essential fo r deal ing wit h the r esearch projects discussed in this chapter.1). 321 .1.

gallbladder. Continuation of the associated glands and human species ducts After Fox (1996) 4. salt abs orption. and clearance Immune Bone marrow. nails Protection. stomach. and waste- product di scharge in the kidneys. pressure. Table 4.1. the main passage with the narrow olfactory region as well as the two meatuses.1 THE LUNGS The respiratory system consists of the conducting zone.e. i .322 Biofluid Dynamics The specific examples discussed in this chapter include CO 2 − O 2 gas exchange i n the lu ng. a nd the r espiratory zone. i.1). as wel l as re gulation of the blo od’s chemical composition in the liver. and their functions.1. (covering) thermoregulation Skeletal Bones. The nose with its two cavities separated by the nasal septum contains three airflow passages on each side.. t he bronchioles and al veolar sac s.. pl asma-water filtration. They are created by . nerves Regulation of other body systems Muscular Skeletal muscles Movements of the skeleton Digestive Mouth. spinal cord. an d l arynx (Fig.e. Secretion of regulatory (ductless such as the pituitary. pharynx.. Movement of blood and lymphatic vessels lymph Respiratory Lu ngs. i. intestine. ureter Regulation of blood composition. pancreas molecules that enter the body Reproductive Gonads. major organs. hair.1 List of body systems. airways O 2 − CO 2 gas exchange w/blood Regulatory Liver. tr achea pl us bronchial t ree. System Major Organs Primary Functions Circulatory Heart. the upper airways. external genitalia. lymphoid Defense of the body organs against invading pathogens Integumentary Skin. wher e the O 2 − CO 2 gas exchange takes place. Breakdown of food into liver. kidneys. and adrenals hormones Nervous Brain. cartilages Movement and support Endocrine Hormone-secreting glands. The u pper a irways include t he nose/mouth. molecules called glands) thyroid. blood vessels. 4.e.

4.Chapter IV: Biofluid Mechanics of Organ Systems 323 100 dp [µm] nose throat G0 10 trachea bronchia G1 alveoli 1 Fig. Fig.4.2 Nasal airways with specific flow regions (Shi.1 Hu man resp iratory sy stem an d lik ely lo cations of micr oparticle deposition.1. 2006).1. .

After the airway cons triction in th e larynx.e. the respiratory system also assures f or proper br eathing i n t he thorax ( or chest) a nd t he abdomen. Specifically.324 Biofluid Dynamics the p rojection of tur binates fr om the la teral wal l ( Fig. Specifically. a partial vacuum) by enlarging the thoracic space. together with the downward moving diaphragm act like a “pump” creating a negative-pressure field (i. T he regulator i s t he ep iglottis which c loses during s wallowing and is op en during br eathing. su rfactant production to control t he sac st ability vi a v ariable surface tension. the se bronchi bifurcate int o t hree se condary bronchi in t he right lung and into two in the left lung. where t he tw o ar e s eparated by a di aphragm. a nd scavenging of foreign material. in conjunction with the kidneys. Furt her repeat-bifurcations of the bronchi rapidly reduce the local airflow rate and geometric di mensions unt il th e te rminal bro nchioles a re rea ched whi ch become the alveolar ducts. during ex ercise. 1977). i. continuous a irflow r ate d ivision quickly reduces the lo cal Rey nolds nu mber all t he wa y to Re < 1 in th e al veolar region. At the end of the alveolar ducts millions of small sacs (the alveoli) are c lustered. i s neu rologically co ntrolled to ma intain t he r equired CO 2 /O 2 -levels and. 4. T he t rachea di vides at t he carina to form the left and right bronchi (or branches) leading to the lungs. s ay. sur rounded by ca pillary beds conn ected t o bot h t he pulmonary artery and the pulmonary vein.e. cartilaginous rings. The r ate a nd depth of ventilation. the diaphragm moves from its flat position upwards again. On expiration.e.. Mucou s layers on the passage walls condition the inhaled air and trap fine/ultrafine particles and microorganisms. blo od per fusion via t he bro nchial and pul monary . i.e. This portion of the respiratory tract is also lined with mucous membranes where the cilia (i. such as bacteria. The trachea (or windpipe) is a smooth-muscle tube supported by incomplete. and the lung deflates (Pedley. C-shaped. The la rynx (or voi ce b ox) is a car tilaginous structure connecting the p harynx and trachea. The pharynx (o r throat) is a muscular section where air flows from the nose to the larynx/ trachea a nd f ood sl ides from t he mouth t o t he esophagus/stomach. In addition to the upper airways and lungs. The turbinates are bones with warm mucous- membrane surfaces that also contain sensitive nerves that detect odors or in duce sneezing to ex pel irritating par ticles. Clearly. a normal body f luid pH of 7.. airflow in the tr achea and below may become turbulent a t e levated b reathing ra tes. The thin-walled alveoli consist of a la yer of e pithelial c ells f or t he O 2 − CO 2 exc hange. I t mainly contains th e v ocal c ords which p roduce s ound bec ause of vibr ation when air pas ses th rough t he opening. a ir move ment i n a nd o ut of t he respiratory tr act. the respiratory muscles relax. called the glottis. Neural inputs to an d feedback fr om these pu lmonary s tructures ar e c ontrolled by t he ne rvous system. fine hairs) sweep mucus an d embedd ed pa rticles towa rd t he pha rynx fo r cl earing.. I n addition to ventilation and subsequent gas diffusion in the alveolar re gion.4. on inspiration the respiratory muscles and chest wall. even when the airflow in the trachea is turbulent..1.2 ).

e. air volume at the start of tidal inhalation. t he largest inhaled volume (VC = TLC-RV ≈ 4 liters). FRC =ˆ functional res idual capacity (FRC ≈ 3 liters). T his di stensibility is c alled co mpliance which re lates ch anges in lung vol ume to net pre ssure cha nges.e. VT =ˆ tidal volume. branching f rom th e tho racic aorta an d int ercostal arteries.2). i. RV =ˆ residual vo lume at mi nimally i nflated l ung. wher e C= C(∀) a s indicated in Fig.1. C = ∆∀/∆p.1.e. 4. i. ..3 P ressure-volume cu rve o f t he l ung. TLC =ˆ t otal lu ng capacity (TLC ≈ 6 liters). i. Clearly. 4. The pul monary circulation is a lo w-pressure s ystem whe re ve nous bl ood is t ransported from the ri ght ve ntricle t o t he right an d l eft l ungs ( see Fi g. air - volume in max imally in flated lu ng.. and outer la yers of the pulmonary ar teries and veins.e.3) is largely due to the inelasticity..Chapter IV: Biofluid Mechanics of Organ Systems 325 circulation su pplies n utrients a nd oxygen t o the l ungs.3. N ote: MV =ˆ minimal vo lume. and IC =ˆ inspiratory capacity (IC = TLC – FRC ≈ 3 liters). In the pulmonary capillary bed the gas exchange takes place and then O 2 -rich bl ood is returned t hrough t he pulmonary vei ns t o t he l eft atrium. i. E = 1/ C. Specifically.2..1.. perfuse the trachea and bronchi as well as the lungs. i . of the collagen fibers when they are stretched.1. Lung compliance and elasticity are inversely proportional.e. The inflation limit (see Fig. They expand nonlinearly under transpulmonary pressure to a point. i. tissue.e. VC =ˆ vital cap acity. . elastic resistance. net average volume during periodic (or tidal) breathing as needed for metabolic requirements.e. due to a high content of elastin and collagen fibers.2 a nd Sect.1.. nerves. t he bronchial ar teries. i. 4. the lungs are very elastic. 30 20 p( ∀ ) p [mm Hg] 10 ~ E=C-1 0 ∀T ∀ [L] 0 MV RV 2 FRC 4 6 TLC ∀ C [%] 0 25 50 75 100 After Staub (1991) Fig.

and drive air into the larger sacs. Wall Velocity. U U R Surface Tension γ Fluid with Air bubble Viscosity µ δf δ(z) y z Fig. among ot hers). W hile nor mal i nfants ha ve de veloped a f ully functioning surfactant system assisting in inflating the fluid-filled lungs at birth. Two-phase flow in the lung is be st i llustrated cons idering t he important rol es surface ten sion and su rfactants pl ay in lung i nflation and function of newbo rns. In turn.. pro duced by al veolar ce lls. which could be e stimated from fluid statics as p = 2 σ/r. Ghadi ali & Gaver. the air bubble is stationary and generates a liquid film δ f = δ f ( µ . S. .1.transport) and absorption (active transport of Na+) is a thin aqueous film in the alveoli. 20 03.4. one s hould r ecall t hat pul monary surfactants exhibit non-uniform (bulk) concentrations and nonlinear surface tension properties (Morris et al. This problem and its sol ution was pro vided by Prof. even collapse. Whe n an alyzing t he fluid dyn amics an d ph ysiology o f collapsible respiratory ai rways. γ . generating a surface tension σ [F/L]. thus requiring a mechanical ventilator for ox ygen supply (s ee Ghad iali et al . a s urface-active agent (s urfactant). 2002. T his implies that smaller alveoli (without film surfactant) with higher pressure levels would reduce in size. Ghadiali (L ehigh University). 2001). pr emature inf ants c annot i nflate their lungs and breathe on their own. 1994).U ) .326 Biofluid Dynamics The net result of fluid secretion (active Cl . However. this creates a pressure in the alveolar sacs of radii r. Air bubble progression and surface film dependence in lung-airway opening. The steady opening of a l iquid-filled airway can be modeled as the pr ogression of a semi -infinite bubbl e i n a r igid ax isymmetric 2-D conduit. In letting the channel move to the left with constant velocity U. gradually l owers t he su rface tension as t he a lveoli get smal ler du ring expiration and henc e pr events the sac s fr om col lapsing ( see Grot berg..4 System schematic. premature babies lack this surfactant system and hence exhibit very high s urface-tension le vels. As a result.

Th is yi elds the following governing equation for the film thickness ∂ 3δ 3µU 3µU δ3 + δ= δf (4.3) with these boundary conditions yields: γ ∂ 3δ  y 2  u( y ) = U − 3  − δy  (4.5) However.1.Chapter IV: Biofluid Mechanics of Organ Systems 327 Assuming steady. (4. where Ca is the capillary number defined as Ca=µU/γ. low-Reynolds-number flow with no gravity effect. whi ch i s a f orce-balance r elationship. three boundary conditions have to be specified plus a 4th condition in order to determine the film thickness.2). the Navier-Stokes equations in the thin film reduce to: ∂p ∂ 2u =µ 2 (4.1): ∂ 3δ ∂ 2u γ = − µ (4.1.7) ∂ζ 3 For this third-order differential equation. t he flow be comes un iform a nd Q= Uδf. • The first condition is that as ζ → ∞ .1.2) − p = γκ = γ ∂z 2 Here κ i s t he i nterfacial c urvature and d ifferentiating Eq .1. the dimensionless governing equation is: ∂ 3η η = 1 −η ( 4. we c an write with Eq. Note that we are assuming a constant surface tension at the air- liquid interface and no Marangoni stresses.1. δf . can be used to express the fluid pressure p in terms of the surface tension γ ∂ 2δ (4.1.4) µ ∂z  2  We can cal culate t he fl ow r ate t hrough a ve rtical section i n t he thin film as: δ (z) γ ∂ 3δ δ 3 Q= ∫ u( y )dy = Uδ + 0 µ ∂z 3 3 (4. (4 .6) ∂z 3 γ γ Using the following scales z = ζδ f (3Ca ) −1 / 3 and δ = ηδ f .3) ∂z 3 ∂y 2 The boundary conditions for this problem are u(y=0) = U and du/dy=0 at y=δ(z). Integration of Eq.1. as z → ∞ .1.1.1) ∂z ∂y The Young-Laplace E quation.(4. η → 1 . incompressible.1. .

e.1. The particular s olution to t his e quation can b e w ritten a s: η=1+Aexp(-ζ)+Bexp(ζ). (4 . This integration yields a value of α=0. they end vi a the terminal bronchioli and alveolar ducts into the millions of sacs with a .1. i. • The 4th and last condition is that there should be a smooth transition in interfacial curvature from the thin film region to a region near the bubble cap.1 Respiratory Tract Geometry In or der to si mulate an d anal yze air flow as well as inh aled par ticle transport and deposition in the lung. The assumption is that the curvature in the bubble-cap region is constant and therefore at some value of ζ. and trachea of the upper airways. As the tubes get smaller and shorter.643(3Ca ) 2 / 3 (4.643.. and then pr ogressing t hrough 23 g enerations of the l ungs.2). The co efficient A may be chosen arbitrarily because the solution should be invariant to a shift in the origin.1. so B=0. 4. ending in t he alveolar regions with millions of al veoli of different shapes and diameters of Ο(100µm).1. F urthermore.e. realistic geometries of the respiratory system are needed.1.1.9) R This ex pression hol ds fo r lo w ca pillary numbers and has been val idated with experiments (Bretherton. However.8) in dimensional form and setting the curvature for a projected two-dimensional spherical bubble to d2δ/dz2 = 1/R. • However. By writing Eq. In ge neral. th e s ystem mor phology is ver y c omplex. 4. ∂ 2η =α (4. 1961).7) re duces t o d 3η/dζ3=1-η. ( 4.328 Biofluid Dynamics • In th e ne ighborhood o f η= 1. w here each bifurcation leads to a new lung “generation” numbered 1-23 (see Fig. we obtain: δf = 0. e ach i ndividual h as a different re spiratory t ract morphology whi ch c hanges with age and may become locally obstructed because of pulmonary diseases. i.8) ∂ζ 2 where α is a constant that is related to the curvature in the bubble- cap re gion. there are several obstacles which have to be addressed be fore computer model ing or la boratory testing c an b egin.7) by starting at a large value of ζ and decreasing ζ until d2η/dζ2 is constant. Clearly. η cannot grow exponentially. t he geometries of t he smaller airways and alveoli change nonlinearly with time as a natural part of breathing. there is t he problem of s ubject v ariability. (4. Eq. where R is the channel radius.1.. larynx. One ca n per form a stepwise numer ical integration of Eq.1. Finally.1 and Table 4. th e lungs co nsist o f a series of bi furcating t ubes. r anging fr om i ntricate nasal airways to the pharynx.

039 0.035 total surface area for rapid diffusion of nearly 100 m 2.171 0.056 16 0.046 18 0.06 0.07 1.037 0. 091 0. 20 00.e. an d dis tensible. amon g othe rs. 134 0. Scaled Weibel Generation model model length (cm) length (cm) diameter (cm) diameter (cm) 0 12. 1968.624 1.565 0.706 0.121 0. 126 0.393 0. The underlines in the table indicate the bor der b etween t he al veolar and t racheo-bronchial reg ions i n t he models (after Finlay et al.033 0. Ph alen et al.038 21 0.043 2 2. Scaled Weibel Finlay et al. and Fi nlay et al.282 0. 068 0. 357 0.035 23 0. 745 0.65 0. 197 0.414 1. 2000).1 0.081 13 0.364 0.65 0.26 1.093 12 0.197 8 0. It should be not ed th at the c onducting airways.218 0. 068 0.028 0.2 cont rasts t he traditional Weibel A m odel with t he upd ated geome tries of Fi nlay et al . known as the symmetric planar Weibel A model. 862 1.071 0. 614 4.132 10 0.115 0. the airways are asymmetric.539 (trachea) 1 3. (2000). i .121 0.78 1.915 0. (2000). ex trathoracic airways plus tracheobronchial region. 454 0.035 0.479 41 .299 6 0.2 Dimensions of the Weibel A lung geometry (Weibel.03 0.71 3 2. 065 0. 281 0.769 0. Table 4.885 0. 653 0.. 897 0. 048 0 .. 828 0.043 0.024 0.05 0.085 0.1..043 19 0.141 0. 277 0. 1963) scaled to a 3 ltr lung volume.456 1 0. 081 0.061 0.086 0.159 9 0.333 0. Finlay et al. 109 0. In reality.454 0. 073 0. 555 0. and using a vol ume of 10 -5 ml per alveoli.073 0. 068 0. The first and still widely use d m apping of the human l ung was done by Weibel ( 1963).547 0.239 7 0. i mprovements wer e published by Hor sfield & Cumming s.462 0.049 0.385 5 1. 219 0. 063 15 0. 1991.231 0.07 14 0. have an air volume of only about 150 . is compared to the symmetric lung geometry used by Finlay et al. out -of-plane.092 0.037 22 0.81 1.04 20 0. 19 78.051 17 0.111 11 0.Chapter IV: Biofluid Mechanics of Organ Systems 329 Table 4. Thus.162 0..1 . Weibel.286 0.

Kle instreuer & Zhang (2003) a nalyzed the air flow pattern and local micro-particle depositions for different inlet conditions at G3 for this Weibel A model (see Table 4. (20 02) hav e f ound a matching fl ow rat e.2 Pulmonary Disorders and Treatment Options Lung diseases can drastically change the airflow mechanics because of local airway constrictions (e. where bronchogenic carcinoma accounts for more than 90% of all lung tumors.2). e. lung cancer causes 28% of all cancer deaths. I nspired by the clinical observation that the effect of in haled drug-aerosol t reatment diminished as t he tumor gr ew beyond a cer tain s ize.0. emphysema).5). che motherapy wi th (a ggressive) dr ug a erosols is a viable treatment option.50(Re mean + Re max ) re presenting t he in halation c ycle. Bronchitis and emphysema. we co nsider a hemispherical l ung t umor of radius r situated in t he s econd branch o f generation 5 (G5. tumors and bronchitis) and destruction of alveolar tissue (i. I n varying t he tumor radius in t he r ange o f 0. w hich generates very s imilar p article de position ef ficiency values ( < 5% e rror) and d eposition pa tterns as in equivalent pulsatile flows.. pa rticle dep osition data f or cy clic flow c annot be s imply evaluated assuming steady flow at the mean flow rate given by the average Reynolds n umber. 4. Fluid-particle dynamics surrounding a hemi-spherical lung tumor. 4. the two most common cau ses of r espiratory failure. an inl et Reynolds numbe r Re match ≈ 0.1.330 Biofluid Dynamics ml. i. As a spe cial cas e of loc al ai rway obst ruction. The e nd e ffect i s t hat t he airway f luid i n t he al veolar r egion b ecomes ex cessively vi scous a nd difficult t o cl ear. Howeve r. Another pulmonary disorder because of a genetic defect in the Ca ucasian pop ulation i s cy stic fibrosis. In order to illustrate some of the difficulties encountered when targeting drug aerosols. t he bra nch area can be bl ocked anywhere from 2% to 100%..1. are tog ether called chr onic obstructive pulmonary disease (COPD). types of inhalers and aerosol dynamics are discussed in Hickey (2004) and by Finlay (2001). Next to surgery and r adiation t herapy. Al though ai r flow and particle transport can be considered as quasi-steady under normal breathing conditions. wi th t he matching Reynolds and Stokes numbers representing resting and moderate .e.2≤r/R≤2. which is the fifth leading cause of death in the US. whil e t he a lveolar r egion enc ompasses 2-6 li ters. T ypical i nhalation flow rates range from 10 to 60 l/min for adults.1. Breathing pa tterns ar e pu lsatile in na ture. Henc e. A mod ern app roach t o c ombat lu ng a nd ot her di seases starts w ith the inhalation o f ta rgeted drug aerosols. Re mean .. among others.2) with r adius R ( see Fig. F or example. Zh ang e t al. computer simulation r esults o f e ffective dr ug a erosol d eposition on a bronchial tumor are presented next.g. Presently. th e characteristics of drug aerosols.

w e expect tha t the computationally efficient constant-flow deposition data duplicate the cyclic-flow scenarios.1 .1.4 G3 G5.11) ρ r where v is the fluid velocity vector. p is the pressure.5 Schematic o f a s ymmetric trip le b ifurcation airway model (generations G3-G6) with a hemispherical tumor. and ν is the fluid kinematic viscosity. lar ge pa rticle-to-air density ra tios. the describing equations are: (Continuity) r ∇⋅v=0 (4.1. t hese pa rticles ar e relatively l arge. Fir st. and ∑ Fp a re t he forces act ing on t he pa rticle. Assuming s teady. G6. 200 3).1 G5 . s o t hat Brown ian m otion c an b e ne glected. Se cond.4. x p is the displacement of the r particle.4. The motion of liquid or solid particles suspended in air is governed by Newton’s Second Law (see Sect. t he particulate m aterial considered is far dens er tha n ai r.1.Chapter IV: Biofluid Mechanics of Organ Systems 331 exercise bre athing (see Table 4.3) .1.2 z=0 Plane G6 .∇p + ∇ ⋅ [ν (∇v + (∇v ) tr )] (4. laminar in compressible ai r f low. no nrotating spheres.12) dt where m p is the mass of a sin gle particle.3): r d 2x p r mp 2 = ∑ Fp (4.1. most of t he k nown particle forces other than the drag force can be discounted using order of magnitude ar guments ( Kleinstreuer. ca using t erms that .1. mon odispersed non-interacting spherical m icron-size pa rticles in smoo th r igid impermeable conduits. ρ is the fluid density. a nd relatively low a ir shear stress v ariations. Consi dering micron-size par ticles.2 side wall x tumor Fig.10) (Momentum) r r 1 r r ( v ⋅ ∇) v = .3 G6 y r G4 2R G6 .

l ow-level fl uid s hear f ields ( Saffman lift).017- 0. Finally. where U is the mean velocity and D is the tube diameter b Stokes number St=ρ p d p 2 U /(18µD ) . such as pressure force.12 0.102 M atching Stokes number range at G3 0.5 M ean Reynolds number a at G3 388 1586 M atching Reynolds number at G3 463 1882 Particle diameter (d p . and it may be stated as: Table 4.3 R epresentative resp iration d ata an d p article c haracteristics with matching Reynolds numbers and Stokes numbers. Eqs.02 – 0. from generation G9 on).12 Representative N ormal particle inlet 4. µm) 3 –7 3 –7 Particle density (ρ p .13). see Longest & Kleinstreuer (2003). to be very small. which produced excellent particle deposition results when compar ed with meas urements (Zh ang et al.25×10 -5 particle concentrations Controlled inlet 5. For near-wall particle drag modifications. the resulting simplified particle equation of motion. 41×10 -4 stream a Reynolds number Re=U D /υ. T hird. the lift forces in the present Stokes flow limit (Re p ≤ 1) are negligible because of a lack of measurable particle spin (Mag nus lift) and the l aminar. i.1. 2002). .e.332 Biofluid Dynamics depend on the density ratio. virtual mass effect. will all ow elucidation of th e physical pro cesses at w ork. the drag can be considered as the dominant point force..63×10 -4 1.1. More specifically.1. 47 0.5×10 -5 1. Hence.02 – 0.017-0.12) and (4. l/min) 15 60 Tidal volume (ml) 500 1153 Time ratio of inspiratory phase (t in /t total) 0.474 Breathing frequency (cycles/min) 14 24. t he ef fects of gr avity ca nnot be co nsidered bec ause gravitational settling is only important for large particle sizes and low flow rates which occur at the lower airways (say.102 0. kg/m 3 ) 540. and Basset force. (4. Physical State Resting M oderate exercise Respiration rate for inhalation (Q in . 6480 600-5000 b Range of mean Stokes number at G3 0.

C DP is the drag coefficient given as (Clift et al.16a) dt subject to the initial conditions r x p ( t 0 ) = ( x0 .646 (4. . The numer ical so lution of t he fl uid fl ow equa tions was car ried ou t with a us er-enhanced f inite-volume ba sed pr ogram CFX4. low mass loading ratios.13b) where  24 / Re p . Fig. z 0 ) (4. parabolically distributed monodispersed particles was specified at the i nlet.1.1. 4. Su ch a t ype of i nlet pa rticle dis tribution was ver ified to be suitable for e xperimental a nalyses and data c omparisons.1..058 exp − 0..3 ( see Kleinstreuer & Zhang. cf .. 5).13) for each non-interacting particle yields r dx p r = vp (4.1. 4. For t he drag f orce. respectively.1. λm is the mean free path in air.e.12. The in itial particle velocities were set equal to that of the fluid and one-way coupling was a ssumed b etween the a ir an d p article f low fields due to t he dilute particle suspension.e. Integration of Eqs.b) 24 / Re p for 1.1. Pa rticle de position an d wi thdrawal occur when t he particle’s center comes within a radius from the wall. The boundary conditions for the governing equations include symmetry with respect to the plane of the bi furcation (i . and no fluid sl ip at the rigid impermeable walls. A uni form pressure con dition was employed at the outlets.14a.16b) For the steady inhalation phase a parabolic fluid velocity profile with random.1. for 0.0 < Re p ≤ 400 and 2λ m   d p  C slip = 1 + 1.1. 2003 for details).0 < Re p < 1.1. the Fa xen cor rection was al so ignored b ecause t he particle diameter is much less than the characteristic length associated with the air flow field velocity distribution.142 + 0.0 CD =  0.15) dp   2λm  Here. 1978): C DP = C D / C slip (4. The local particle Reynolds number is r r Re p = ρ | v − v p | d p / µ . where µ is the dynamic viscosity of t he fluid. z = 0. i.13a) r where d p and v p are the particle diameter and velocity. (4. which mimics the fact that s olid particles or liquid droplets “de posit” a s long a s t hey t ouch t he moist epithelial or mucus lay er.Chapter IV: Biofluid Mechanics of Organ Systems 333 r r 1 r r r r ∑F p = FDrag = πd p2 ρC DP ( v − v p ) | v − v p | 8 (4.999  (4. y 0 .

Such fl ow sep aration-and-reattachment eve nts are a function of the l ocal geometric f eatures. 200 1. particle dep osition pa tterns. 4. T he branch flow rate ratio declines with tumor size. ver y l arge recirculation zone s ar e f ormed at t he b ack o f t he tumor a t l ow inh alation r ates ( cf. a nd de position efficiencies (Comer et al. which pushes high-velocity fluid to the low- velocity region.1. Fig. Due to the relatively high Peclet nu mbers present.1. 2000. Specifically. and recirculating flow in C-C′. a nd t he l ocal Reynolds nu mber.1. However. hence. the graphical results are only d epicted fo r t he t umor a rea as wel l as t he down stream r egion.6) . 2002 ). ai r an d particle fl ow fields upst ream of sma ll to medium-size tumors are almost the same. At cross section D-D′. The secondary flow depicted in Slice A-A′ will move the particles from the outside (A′) to the inside (A). the particle distribution was captured just when t hese particles passed through this section. . some particles move from the inside (D) to the outside (D′) carried by secondary flows. The good ag reements between e xperiments and th eory i nstilled co nfidence t hat t he present computer simulation model is sufficiently accurate to analyze fluid-particle dynamics a nd a erosol dep osition i n a fou r-generation ( triple b ifurcation) lung model segment with local tumors.. these particles may convect downstream into generation G6. the maximum velocity has slightly shifted back towards the center of the tube due to the upstream secondary motion. directly down stream of the par ticle occlusion. Fo r example. 4. 4.1. It is expected that t he p articles n ear t he tumor s ide ( A′) may i mpact th e tumor di rectly and deposit on the tumor . because of the reduction of the airway lumen for both inhalation conditions.5) further amplifies the a symmetry of t he f low fi eld bet ween br anches G5.7 show airflow fields and particle transport in the vicinity of the tumor for different inlet flow rates and tumor sizes. Figs. Fig. Both mid-plane velocity vector plots demonstrate at va rious degr ees th e bas ic phen omena o f st agnation p oint flow after A-A′.1. Particles deposit on the tumor surface and avoid entrainment into the recirculation zone so that there are no particles near the out side of th e tube (C′). 4.1. The tumor obstructing the G5-airway (see Fig. and the upstream effect of the tumor is small. 4. I n cr oss section A-A ′ of Fi g.2 an d G5. However.6) and h igh flow rates with an expanding cr oss sec tion d ownstream of th e l arge tumor ( cf.5 . and as a result two distinct particle swirls appear. accelerating flow in B-B′.334 Biofluid Dynamics The present computational fluid-particle dynamics (CFPD) model has been validated wi th various e xperimental d ata se ts fo r s ingle as wel l a s double b ifurcations u nder s teady or cy clic flow c onditions i n t erms o f velocity pr ofiles.6 and 4. Zhan g et al . d ownstream geo metric tr ansitions.1. For a given cross section. some par ticles may fol low the streamlines and bypass the tumor so that the particle concentration seems to be higher at cross section B-B′. .3. the particle concentration is quite uniform for this case.

6 Mid-p lane velocity vectors . ax ial v elocity contours.Chapter IV: Biofluid Mechanics of Organ Systems 335 ReG3=463 Mid-Plane Velocity Vector Profiles r/R=1 (Uniform Vectors) D′ A′ C′ B′ A B C Speed (cm/s) D 14 43 72 100 129 158 186 215 cm/s cm/s 0 40 80 120 160 A-A′ 50cm/s 4 48 92 136 180 224 B-B′ 50cm/s A A′ B B′ cm/s cm/s -4 40 84 129 173 217 C-C′ 50cm/s 50cm/s -1 38 77 117 156 195 D-D′ C′ C D′ D Fig. 4. . an d secondary vectors (upper half of cross-sectional views) as well as particle d istributions and flow directions (lower half of cross-sectional views) at selected cross sections near the tumor with r/R=1.1.0 at ReG3=463 and StG3=0.12.

5 at ReG3=1882 and StG3=0.12. 4.5 (Uniform Vectors) D′ A′ C′ B′ A B C Speed (cm/s) D 0 129 257 386 514 643 771 900 cm/s cm/s A-A′ 200cm/s 18 200 382 563 745 927 B-B′ 200cm/s -83 36 154 273 392 510 A A′ B B′ cm/s cm/s C-C′ 200cm/s 200cm/s -10 175 359 544 728 912 -66 123 313 502 691 880 D-D′ C′ C D′ D Fig.7 Mid-p lane velocity vectors .336 Biofluid Dynamics ReG3=1882 Mid-Plane Velocity Vector Profiles r/R=1. an d secondary vectors (upper half of cross-sectional views) as well as particle d istributions and flow directions (lower half of cross-sectional views) at selected cross sections near the tumor with r/R=1. .1. ax ial v elocity contours.

1 and branch G5.e. In o rder to pr operly e valuate the overall concentration of p articles appearing on a t umor s urface. This is demonstrated for the case r/R=1. attributed to the secondary flow and blockage effect of the obstruction. Particle d eposition on t he t umor s urface fo r di fferent (Re− St ) match pairs may be i nfluenced by t he l ocal o cclusion (i . At c ross section A-A′. where most particles land on the front surface of the tumor and some deposit on the tubular wall.e . a cr itical tumor size with maximum DF-values can be identified ( Fig. which indicates that very low particle depositions can be expected at the downstream generations G6.2 change significantly for the large-size tumor case as well.3 and G6. 4.e.1. C -C′. The flow ratios between branch G5. i. the maxi mum ax ial velocity zone has shifted to the outside (A′) due to the blockage effect. particles landing on the tumor surface are most desirable for targeted drug aerosol delivery. occurring mainly along the front surface of the tumor due t o i mpaction. 4. In general. The . However. 4.2 due t o t he bl ockage effect. a nd D. m ass of deposited particles to total particle mass entering the inlet tube (here G3).1.e. the recirculation zone behind the tumor extends further into G6. so that only a few particles exist near the outside of the tube (A′). some particles deposit along the top/bottom and inside wall of th e tube in th e geometric t ransition zo ne.2 bec ome larger for l arger tumor s pr esent at branch G5.1. t he critical tumor s ize i s r/R≈1. it may n ot only influence the flow fields d ownstream b ut also to a s mall e xtent up stream ( Fig. The strong secondary flow drives particles from the outside (A ′) to the inside (A).6).. the number of particles at cross section A-A′ is obviously smaller than that for the medium-size tumor case (r/R=1) because of the decreasing flow rate. i. i. t umor si ze a nd location)..e. When plotting the DF as a function of tumor radius for different (Re− St ) match pairs. Clearly.2 i s occ luded by t his type of obs truction. the situation cha nges somewhat because the high inlet Reynolds number has a measurable effect on the flow field approaching the obstruction. T he flow f ields ne ar t he second car inal r idge and in generations G5.25. i ncreases wi th increasing Stokes number a nd Reynolds number.D′ is ver y small.Chapter IV: Biofluid Mechanics of Organ Systems 337 For the large-size tumor (r/R=1.4 . very few particles enter and deposit at the downstream bifurcations because of the tumor blockage effect. a s w ell as in let velocity p rofile a nd p article d istribution.7). Re match = Re G 3 = 1882) . an d G6. G6. 8). For this relatively large-tumor-size case.7 s tagnation po int f low and a recirculation z one ar e generated in front of t he tumor. 4.1 . t he nu mber o f pa rticles at sec tions B-B ′. Fig.St combinations. At t he same time. 1.5) under moderate exercise breathing condition (i . As shown in Fig. F or t he present s ystem and most in let Re. Sin ce most particles near the tube center impact on th e tumor surface. In addition.25.. th e depo sition fr action (DF) . ab out ha lf of t he daughter bra nch G5. is employed.1..3 when compared to the medium- size tumor case (r/R=1.. particle deposition.

4 Maximum DF-values on variable tumor surface (Re=1882.25<r/R≤2.25) results in a reduction of flow rate and particle number entering the diseased branch.3 .1 1.e.5 7. However. t argeted dr ug a erosol de livery i s necessary (see Sect. Tumor size: r/R 0.0. St=0.. St=0.5 Critical tumor radius r/R Fig.0 1. Hence.8 Variations of DF-values on the tumor surface vs tumor size. Table 4. St=0. 1.12). 4.08 Re G3=1882. Table 4. 5.2). the number of deposited particles. St=0.5 2. DF. wh en t he p article numbe r as sociated wit h t he f low r ate approaching the tumor bra nch decreases dra stically i n the cas es o f l arge- size tumors.4 s ummarizes t he max imum de position fractions.8 7. 1.7 10. there are still a few particles depositing at t he tumor s ite i f the br anch i s c ompletely bl ocked by t he tumor.1. Although the presence of small-to-medium size tumors (e. ReG3=463.1.5 1 1.6 1.08 ReG3=463.2 0. Clearly. 4. i.12 Re G3=1882. The effect of tumor size may vary somewhat with the inlet Reynolds and Stokes numbers as indicated in Fig. In addition. St=0.0 Maximum DF (%) 0. wh ere fo r ( r / R ) critical DFmax ≈ 11% .2 1.. St=0. i. St=0.5 2 2.04 ReG3=1882..04 ReG3=463.3.1.g.25 1.e.8. This can be explained by the existence of a low positive flow near the inside wall which moves particles to the inner tube wall or tumor site. decreases a lthough the probability of par ticle impa ction i s hi gher. the likelihood of particle deposition increases with tumor growth due to inertial impaction. 0<r/R<1.338 Biofluid Dynamics occurrence of this phenomenon can be attributed to the variations of branch air/particle flow rates as well as local flow fields with tumor size. DF-v alues con tinue to increase wit h tumor size.12 Deposition Fraction (%) 10 0 0 0.

5. tu mor locations. in let Reynolds numbers.. large- size tumors m ay chan ge th e fl ow features ups tream an d in the sister branches. the kidneys filter out and discharge via the urinary tract biochemicals as wel l as water to cl ean the bl ood an d to ach ieve constant o smolarity. Transport of f luid (i . wa ter reabsorption. as wel l as up stream flow a nd particle distributions as a function of Reynolds nu mber.. W hile medium- size tumors generally affect the nearby and downstream flow fields. Particle deposition on the tumor surface may b e influenced by the local occlusion ( tumor s ize). breathing condition) as well. Separately it was shown that the release positions for aerosols which land on the tumor surface concentrate for high Re-St pairs in a specific area of the inlet plane. findings of the present study can be summarized as follows: 1. Thus . the blood’s pH is regulated to stay at 7. an d so lute di fferences i n solutions. DFs on t he tu mor surface increase first and then decrease with the tumor size. and molecules . i. As expected. S tokes nu mber. conc entration g radients. cel ls. W hen o ther c onditions (e .Chapter IV: Biofluid Mechanics of Organ Systems 339 Assuming symmetrically bifurcating rigid lung airways. pr oteins. e. controlled particle release could dramatically increase the deposition of drug aerosol at desired lung target sites and at the same time decrease the influence of aggressive drugs on the healthy airway epithelium (see Sect. The effects of tumors on the flow characteristics in the bronchial airways are related to the inlet flow rate (i. and the blood pressure is controlled at a normal arterial mean of 100 mmHg. an d con stant-diameter non.e. though this zone may vary somewhat with the particle release conditions and tumor size. Smooth inhalation wave forms and dilute aerosol uptake can be accurately simulated with matching Re-St pairs. the presence of a singular tumor in bronchial airways changes the flow distributions among different branches.2 THE KIDNEYS The kidneys are an amazingly efficient pair of filters and regulators. 4. eve ry 4 0 minu tes or s o t he en tire blo od vol ume ( ≈5-6 lit ers in adults) is cleared of the blood’s waste products.3.interacting mi cron-size pa rticles. Maximum DF- values are generated for a he mispherical tumor size where the lumen area of the afflicted tube is approximately half-way occluded by the tumor.4. and se cretion o f ADH a s wel l as to xins an d i ons include pressure di fferentials. ( The influence of t umor shape wou ld be important as wel l. In fa ct. Hence. 2.g. The driving forces for b lood f iltration. and tu mor s ize.. w ater). h omeostasis of t he p lasma c oncentration. and Stokes nu mbers) are f ixed. Secretion of the a nti-diuretic ho rmone (ADH) re gulates the c orrect wa ter balance i n t he bod y..) The par ticle depo sition fr action (DF) on a tu mor increases si gnificantly whe n s witching b reathing pat terns f rom r esting t o moderate exe rcise.e. t umor loc ation.2). 3. hemispherical tumors.

s upplied b y t he r enal artery. From the loop of Henle. The renal capsule covers the medulla which houses 8 to 15 conical renal pyramids. 4. osmolarity. The tubular walls have millions of microvilli (tiny fingerlike projections of wall-cell m embranes) whi ch fo rm la rge e ffective surface ar eas for re - absorption of wat er.. against a concentration gradient and hence requiring metabolic energy. The fl uid (n ow cal led ur ine) i n t he . NaCl) is immediately returned to the vascular system via reabsorption to maintain a normal wat er bal ance and henc e nor mal blood osmol arity an d blo od pressure (see Fig. diffusion. 4.g. Approximately 99% of the filtered water (plus salt. osmosis.1b). c oiled) t ubules ( see Fi g. Leaving th e renal cor tex.. the remaining water flows via the distal convoluted tubules to the collecting duct.. as nee ded by the body. It is a net filtration pr essure ( p = 1 0 mmHg). a n ant i- diuretic hormone (ADH. and ultrafiltration.e.e. Bec ause 99% o f t he f iltered wat er has t o be reabsorbed. Thus .2. the fluid descends the “Loop of Henle” (see medulla region in Fig. 1. whi ch t hen migr ates i nto t he lumen of the pr oximal convoluted ( i. 200-500Å) of the glomerular capillaries to the inside of the capsule and the lumina of the nephron tubules. Each renal pyramid consists of “nephrons. 4. sa lt. the ascending limb i s i mpermeable to water but ac tively tr ansports Na +. filtration takes p lace in the c apillary be d ( or gl omerulus) which i s covered by t he Bowman capsule. blood cells.340 Biofluid Dynamics occurs se lectively acr oss tub ular wal ls whi ch ac t li ke se mi-permeable membranes (see Sect.1b). which generates across large capillary surface areas a high v olume of ult ra-filtrate. cal led ul trafiltrate.” i.1 Kidney Structure and Functions Each adult kidney weighs about 160 grams and is about the size of a fist (see Fig. 4..e. i. pr oper ADH rel ease a nd i ts con centration c ontrol collecting-duct permeability and hence normal water balance. t ubules plus capillary beds . The net effect is that salt is extruded into the surrounding tissue fluid so th at water is dra wn out of the des cending limb by t he osmotic pressure and collected by the capillaries. a nd other mole cules i nto t he surrounding capillaries.e. c alled th e gl omerular f iltration r ate (GFR ≈ 180 liters/day). The filtrated water passes through filtration slits into Bowman’s ca psule. Water and dissolved solutes (minus proteins. through rather large pores (i.. fol lowed passively by Cl-. e .2).1b) where most o f the f iltered wat er i s r eabsorbed vi a osmosis in to the c apillaries.2. and pl atelets) pas s f rom the bl ood plasma.2. 4 .1a).1. Such membranes feature variable pores with micro-channel s izes wh ere th e so lute t ransport ca n be pas sive.2). hyd rostatic bl ood pr essure minu s osmotic pressure..e. the functional uni t of the kidney th at i s responsible for the formation of urine (Fig. 4. While the descending limb is passively permeable to water. and ul timately uri ne d ischarge.2. whi ch i n conjunction w ith th e osmotic pr essure secures s ufficient water re- absorption. Specifically. or active. i.2.2. or anti water-loss hormone) triggers formation of water ch annels in the ce ll m embranes of the co llecting duc ts.

1996).2. a) Kidney midplane Renal pyramid (b) Blood flow and plasma clearance Proximal convoluted tube R en a l cortex Renal artery Renal vein R en a l m e du lla towards Ureter Fig. 4. The ureter terminates in the urinary bladder to which the urethra is connected. 4. . i s th en drained in to the re nal pe lvis (see F ig.2. located in the r enal cor tex and medulla r egion.Chapter IV: Biofluid Mechanics of Organ Systems 341 collecting duc ts.1a) from whic h i t leaves t he kidney via the ureter.1 Kidney structure and nephron function (after Fox.

re-absorption. t o di ffuse eas ily t hrough an artificial semi- permeable m embrane. pl asma i s c leansed o f j ust t hese wa stes (e. 1996).2. can cause hy pertension and high urea concentrations in the blood. t o maintain homeostasis. Re nal ins ufficiency. inside a cylindrical housing (or shell) of D≈3- 4 cm and L = 20 – 30 cm are several thousands of hollow fibers (see Fig. of Cl.2 Fluid Flow and Mass Transfer in an Artificial Kidney Model As discussed i n Sec t. ca lled u ltrafiltration.2 Microscale kidney function: Filtration.g. 4. called dialysate. and secretion in a nephron (after Fox. proper vascular access to withdraw blood and return it is ve ry i mportant. called uremia.. . e. u rea.2.g. Hemodialysis is the separation of unwanted molecules on the basis of size. Uremia patients ( 350k in the US al one) hav e t o be o ften c onnected t o di alysis machines.342 Biofluid Dynamics Secretion of Reabsorption Filtration Bowman's Capsule addit. accompanied by elevated plasma H+. 2003) .2. washes the wastes and extra fluid out of the kidney machine. Thus . Long est & Kl einstreuer (2000) a nalyzed acc ess-point problems and discussed solutions mitigating stenotic developments.. 4. d ue t o destroyed ne phrons or r educed p lasma fi ltration. just as it occurs ac ross t he gl omerular capillaries. u rea an d ot her was tes.2. 4. one of t he k idney’s k ey f unctions is t he removal of was te pro ducts f rom t he blo od. A hollow fiber is a long tiny tube of inner diameter d≈ 200 µm and semi-permeable membr ane th ickness t = 15.50 µm ( Liao et al. The dialysis solution.e.and K+-ions which may lead to uremia coma. These m achines are bas ically co nvective-diffusive co untercurrent mass exchangers. 4..1. It contains chemicals which bind to the waste products and has to be prescribed for every patient to be most effective.) as they migrate in a di alyzer from the blood across the por ous m embrane in to t he di alyzing fl uid bec ause of a pr essure difference. re -absorption of water a nd needed species is not possible with dialysis machines (see Fig. Ho wever. etc. 4. i. cr eatinine potassium. (GRF) waste Na+ Afferent T ubule products water Arteriole Excretion towards Collecting Ducts Glomerulus Efferent Arteriole Fig.2. People wit h ki dney pro blems oft en have to und ergo hemodialysis t hree times per week.3b). kno wn a s artificial kidney mac hines. Typically. fluid. .3a).

Focusing on a bundle of hollow fibers (say. (2003). and (b) Single hollow fiber model.2. three flow processes ha ve t o be cons idered. The incoming bl ood wi th it s waste pr oducts i s considered to be Newt onian.1. assumptions and approaches for fluid flow and mass transfer in this representative dialyzer unit are: • Steady 2 -D la minar di lute s uspension f low in t he f ilter l umen. Following Liao et al.4.2.2. Kidney machine model. Lumenal Blood Flow..e .2b) .3. fl ow t hrough a p orous medium formed by these numerous hollow fibers. l umen-side.b). employing the Kedem-Katchalsky equations (see Sect. Thus. 1..2.2a) ∂r ∂z ρ ∂z  r ∂r  ∂r  ∂z  r-Momentum: ∂v r ∂v r 1 ∂p  ∂ 1 ∂ 2  vr + vz =− +ν   (rv r ) + ∂ v2r  ∂r ∂z ρ ∂r  ∂r  r ∂r  ∂z  (4. 4. using Darcy’s law (see Sec.3a.4). 4.2). a nd shell-side flows. 1.3 Hemodialysis: (a) A typical hollow fiber dialyzer (k idney machine). • Steady 2-D di alysate. the governing equations and boundary conditions read: 1 ∂ ( r v r ) ∂v z Continuity: + =0 (4.2.1 and 1. i. t rans-membrane. she ll-side. e. i.2).1) r ∂r ∂z z-Momentum: ∂v z ∂v z 1 ∂p  1 ∂  ∂v z  ∂ vz  2 vr + vz =− +ν  r  + 2  (4. entering ax isymmetric and f ully d eveloped each filter lumen at a low Reynolds number. • Steady radial trans-membrane mass fluxes for solution and solute.000) as the key element of a dialyzer (see Figs. using the reduced Navier-Stokes and mass transfer equations (see Sects. 1.Chapter IV: Biofluid Mechanics of Organ Systems 343 (a) (b) out h Shell c Transmembrane d Rs flux pd Dialysate Hollow flow Fibers Membrane cb t Lumen r Rl Blood flow p z b Fig.3. n = 12.

∂c and J s = − D + cJ v ∂r Membrane Mass Transfer.1 Model parameter values. the semi-permeable m embrane (s ee Fig. D M = 9. Specifically. k r =. the reflection coefficient. (4.2.3) r ∂r ∂z  ∂r ∂z  Boundary Conditions:  r 2  2Qtotal • Inlet (z = 0) v z ( r ) = v max 1 −    .7 cm. t = 15 µm Shell-side: N = 12. 76 × −6 10 m / s .2.1).344 Biofluid Dynamics v r ∂ ( rc) ∂c  ∂ 2c ∂ 2c  Mass Transfer: + vz = D 2 + 2  (4. s. and dif fusive per meability D M are intrinsic membrane properties and have to be measured (see Table 4.416 −11 2 −9 2 × 10 m . L=20.4)). i.000 hollow fibers.15 × 10 −10 m / s / Pa . σ = 0. we have: Solution Flux: J v = LM ( pb − p d ) − σRTLM ( cb − C d ) (4. D = 1. which indicates the s olute fr action bl ocked b y th e membr ane. v r = J v (see Eq.81 × 10 −9 m 2 / s 2 . Hollow Fiber R = 100 µm. 0 • Outlet (z = L) p=0 ∂v z ∂c • Centerline (r = 0) = 0. v max =  R  πnR 2 vr = 0 and C = C b.2. k z = 1. The hyd raulic membrane permeability L M .5) Note th at the di mension of the sol ution fl ux [L3 /( L2 T)] is t hat of a velocity [L/T].4) Solute Flux: J s = C M (1 − σ ) J v + DM (C b − C d ) (4..29 × 10 m Membrane-Urea: LM = 1. Table 4. 4. e. while that of the species flux is [M /( L2 T )] .2. = 0. Bec ause of fluid pr essure an d spec ies concentration differentials between lumen area and shell side both solution and solute cr oss the lu men w all. v r = 0 ∂r ∂r • Membrane surface (r = R) v z = 0. The av erage s olute c oncentration in side th e membra ne i s C M [M / L3 ].2. R s = 1.2.10b).32 cm.

Focusing on ur ea c learance <te st sol ute> a nd usi ng a cel lulose triacetate membrane.1. can be neglected under th e as sumption of l ow ult rafiltration a nd ba ck filtration rates.2. (2003) suggested correlations to e stimate these source terms when trans-membrane fluxes are significant. as given by Liao et al.7b) µ ∂r ∂c ∂c Mass Transfer: v +u = Ss ( = 100 mg/dL. . a nd ur ea co ncentration distribution at the shell-side as well as the urea clearance [ml/min] from the solution (blood) for different flow rates.8) ∂r ∂z Boundary Conditions: Qd • Inlet u = 0 and v = − 2πRs h • Outlet p=0 • Walls u=v=0 Note that the source terms S v and S s for solution and solvent.4. Assuming ra dially-directed uni form inl et and out let velocity profiles o f t he di alysate th rough o uter-shell ar eas Ain = A out = 2πR s h .2. The flow chart for the sequential calculation steps is given in Fig. Of in terest ar e th e velocity. pr essure.7a) µ ∂z 1 ∂p r-Momentum: v = − kr (4. are summarized in Table 4. the system parameter values.2. the “porous medium” flow equations read: 1 ∂ ( r v ) ∂u Continuity: + = Sv (4. Li ao et = 120 m l/min and C b. (2003). Specifically.6) r ∂r ∂z 1 ∂p z-Momentum: u = − kz (4.2. respectively.9) Typical lumen inlet values are Q b.Chapter IV: Biofluid Mechanics of Organ Systems 345 Dialysate (4. Maximizing waste product clearance is of course the goal of optimal dialyzer design and operation.2. 4. 200 ≤ Q d ≤ 1000 ml/min and 100 ≤ Q b ≤500 ml/min. say. κ ≡ (QbCb / in − QbCb / out ) / Cb.

2.2. (4. Figure 4.(4.4 Computational flow chart for urea clearance in a kidney machine model.4) and (4.2.2.). 4. and urea concentration distributions a t t he s hell s ide for fl ow r ates Q b= 400 ml /min.1-4. Cl early. Because of th e much lar ger re sistance in t he rad ial di rection (k rr<<kzz). t he ax ial .2.3 and 4.9) as a fu nction of d ialysate a nd blood f low r ates.2. Then.8) Update boundary to calculate the distribution conditions and source of velocity. 4.3) to calculate the distribution of velocity.2. Figures mome ntum. Eqs. pressure and terms for Eqs. 4.346 Biofluid Dynamics Given initial data Solving Eqs.5 de picts the variations of the urea clearance rate (see Eq.6-4. the solution and solute fluxes were updated ac cording to the Kedem-Katchalsky equations f ollowing t he simulation steps (see Fig.3) and (4. Qd= 400 ml / min and an inlet urea concentration at the lumen C b.2. and mas s transfer equations in both lu men and she ll sides (i . I nc. pressure.8 show the velocity.2.4.6-4.8) were solved numerically. (4.2. 4. t he urea clearance increases with an increasing dialysate or bl ood flow rate due to the enhanced mass fluxes. Results and Discussion..2. pressure and concentration at the lumen side Solving Eqs.1- concentration at the shell 4.e. the c oupled c ontinuity. Using the commercial finite volume code CFX4 (Ansys.8) side Solving to calculate the transmembrane flux of solution and solute Meet convergence criterion? Generate output Fig. (4.6-4.4).6 - 4.1.

350 300 250 κ 200 Simulation (Qb=300 ml/min) 150 Exper. The ra dial flow is onl y s ignificant nea r t he inl et an d out let due t o i nlet/ outlet effects. Moreover.2.3b). 4.2.. data (Qb=400 ml/min.60 2. ..80 1. 2003) Simulation (Qb=400 ml/min) Exper. Liao et al. 4.6. Liao et al.4.6 Velocity vectors an d ax ial velocity conto urs at the s hell s ide of an artificial kidney model (see Fig.2. 2003) 100 100 200 300 400 500 600 Qd (ml/min) Fig.00 0. The pressure distribution is non-uniform near the dialystate inlet and outlet and becomes uniform in mid-region due to t he interaction between isotropic permeability and entrance/outlet effects (see Fig.2.2. data (Qb=300 ml/min.20 1.Chapter IV: Biofluid Mechanics of Organ Systems 347 flow is dominant in most areas of the dialysate side.40 1.5 Computational and measured urea clearance rates in an artificial kidney model. flow becomes weak near the centerline below both inlet and outlet.20 Fig. Dialysate outlet Dialysate inlet u (cm/s) 2.7). 4.60 0. 4. as shown in Fig.

00 1050. Dialysate outlet Dialysate inlet c (mg/dl) 0.00 1810. A more uniform distributed dialysate velocity and concentration may contribute t o a hi gher p erformance of a n ar tifical kidney.2. 4. The co ncentration in the center is larger than that in the periphery and the maximum concentration appears at the cen terline bel ow t he ou tlet bec ause th e so lute c an r emain longer in these low-velocity regions.00 670.8). effective biofluid analyses are very helpful to evaluate the urea clearance rate an d opt imally des ign the ar tificial k idney i n or der to achi eve hi gh clearance of urenary solutes of toxins.00 1430.16 0.00 1620.00 100. .11 Pressure distributions in the shell side of an artificial kidney model.00 1240.34 0.45 0.39 0. Associated with both axial and radial velocities.22 0. 4.8 Distribution of urea concentration in the shell side of an artificial kidney model.00 290. The refore. 4.00 480.00 860.00 Fig.348 Biofluid Dynamics Dialysate outlet Dialysate inlet p (Pa) 2000. the urea concentration in the shell is also non-uniform (see Fig.28 0.05 Fig.

Category Processes Detoxication of Blood • Digestion of toxins by Kupffer cells • Chemical alteration of biologically active molecules (hormones and drugs) • Production of urea. i t ca n re move hormones. Table 4. 4.3. the liver has the largest variety of functions. drugs. compared with all other organs. lactic acid) by gluconeogenesis • Secretion of glucose into the blood Lipid Metabolism • Synthesis of triglyceride and cholesterol • Excretion of cholesterol in bile • Production of ketone bodies from fatty acids Protein Synthesis • Production of albumin. Spec ifically.3 THE LIVER The li ver. w hich is th e la rgest i nternal o rgan.1a).3. uric acid.3 .1).1 Liver functions (after Fox. re gulates t he c hemical composition of the blo od (Fi g.Chapter IV: Biofluid Mechanics of Organ Systems 349 4. and other compounds via different biochemical processes (see Table 4. and other molecules that are less toxic than parent compounds • Excretion of molecules in bile Carbohydrate Metabolism • Conversion of blood glucose to glycogen and fat • Production of glucose from liver glycogen and from other molecules (amino acids. Clearly. a major component of the plasma proteins • Production of plasma transport proteins • Production of clotting factors (fibrinogen. prothrombin. 1996). and others) Secretion of Bile • Synthesis of bile salts • Conjugation and excretion of bile pigment (bilirubin) .

nutrients. 1996).1 Blood and bile circulation in the liver (after Fox. toxins. etc. 4.350 Biofluid Dynamics (a) The liver’s central role in regulating blood chemical composition Renal Artery Ureter Capillaries Blood. . (b) Blood and bile flow in a liver lobule Fig.3.

Chapter IV: Biofluid Mechanics of Organ Systems 351 4. as well as mass transport in blood . 4.3.e. cal led hepatocytes. li ke whi te blo od cells. Based on surface appearance. fluid flow in the acinus. re moval.2 a. can “de vour” toxins. bile i s p roduced by t he hepatocytes and se creted into microchannels. 4.1. The rate at whi ch the l iver up take a nd transport various chemical species.1). 4. a hi ghly t oxic chemical produced in t he l iver f rom ami no ac ids and the action of ba cteria i n t he intestine. where they are excreted into the ureter (see Sect.3.1a). double-layered membrane that reduces friction against other organs. An ex ample whe re l iver and kidney fu nction t ogether is t he conversion. and terminal hepatic venule.1 Liver Structure and Functions Apart from a patch where it is connected to the diaphragm. Hence. 3. w hich em pty in to th e bi le d uct (s ee Fig.1a). The central veins of all the lobules converge and deliver the blood to the inferior vena cava. t hree t ransport pro cesses hav e to be considered. In contrast. The li ver has the ne cessary enz ymes t o conv ert ammoni a int o less toxic urea molecules. Mas s transport of ch emical spe cies between bl ood and th e h epatocytes ( liver cells) m ay o ccur a t th e le vel of t he he patic ac inus. ea ch with a ce ntral ve in which c ollects t he a rterial/venous blood mixture flowing through the sinusoids (Fig. These plates are sep arated by mi crochannels.3.. one of the liver’s key functions is effective uptake.3. the liver is covered entirely by a t hin.3. As in dicated i n Fig. sinusoids. i. etc. 4 .3.1b). a nd d ischarge of ammonia. The cha nnel wal ls are li ned with K upffer c ells wh ich.1a.2 Fluid Flow and Mass Transfer in a Liver Model As discussed in Sect. Focusing on a hepatic acinus as the key element of the l iver ( see Fi gs.1b). is of i mportance in stu dying li ver physiology a nd pha rmacology a s wel l a s tr eating l iver d iseases. which are secreted into the blood and carried to the kidneys.b ). T he l iver c ells. t ransport. As shown i n Fig 4. 4. ca lled b ile c analiculi. Hepatic acinus model. a s it rec eives arterial bl ood vi a t he hepatic artery.3. nutrients in the blood from the digestive system are carried via the hepatic partial ve in t o cap illaries i n the li ver.2a. 4. Surprisingly.2 and Fig. form hepatic plates that are one to two cell-layers thick. the hepatic acinus consists of the bile duct. 4. a nd d istribution o f c hemical s ubstances to t he blood an d b ile.3. si mulating th e fluid flow and mass transfer in the hepatic acinus is useful in evaluating the uptake a nd transport of c hemical s pecies in t he liver. su ch as a dru g. it can be divided into major folded lobes (see Fig.3. 4. terminal portal venule. terminal hepatic arteries. 4.3. The hepatic plates are arranged into active units called liver lobules. storage. ba cteria. 4. there are only two types of ce lls f orming t he s tructure of th e l iver. called si nusoids (s ee F ig.

(a) Hepatic acinus (after Gumucio. the a ssumptions and app roaches f or f luid fl ow an d mass t ransfer i n t his representative acinus model are: • Steady. neglecting any metabolism.3.2c) Outlet p = p0 (4.2d) Walls vr = v z = 0 (4. • Simple dif fusion of c hemical spe cies a cross the cel l me mbrane. l aminar.1) r ∂r ∂z 1 ∂p z-momentum: vz = − kz (4.2e) .2a) µ ∂z 1 ∂p r-momentum: vr = − kr (4.3.3. Fol lowing Lee & R ubinsky ( 1989).3.352 Biofluid Dynamics and ti ssue and ac ross th e membrane.3. the “porous medium” flow equations read: 1 ∂ (rv r ) ∂v z Continuity: + =0 (4.3.2 Hepatic acinus. 1983) (b) Schmetic of hepatic acinus model terminal hepatic terminal arteriole r hepatic terminal terminal terminal venule portal hepatic portal venule venule z venule bile duct portal vein hepatic artery Fig. incompressible po rous f low i n the a cinus with sinusoids.3.2b) µ ∂r Boundary conditions: Inlet p = p1 (4. Thus. • Transient di ffusion-convection i n the bl ood ve ssels a nd surrounding tissue. The incoming blood with chemical species is considered to be Newtonian. entering t he s inusoids at a low Rey nolds nu mber ( Re < 1) . Fluid Flow. 4.

is treated as a source term in Eq..3. con centration i n t he t issue (i .e. t ) − ct (r. Once t he velocity fi elds were ob tained. the tr ansient sp ecies transport in t he f luid (i. The pa rameter values used for this simulation are given in Table 4. the net change in the fluid chemical species concentration exchanging with the ∂  Vt  tissue.3.3d) ∂n Mass transfer across the cell membrane.e. t = 0 ) = c0 .3. After the ∂t  V f  species concentration in th e fl uid. The co mmercial fi nite volume software CFX4 was employed to solve the fluid flow and chemical species di stribution i n t he t hree-dimensional a xisymmetric ge ometric representation of t he l iver ac inus. z.3b) Inlet: c f = cin (4. (4. z.3. Vt and V f are volume of tissue and sinusoids.4) ∂t Vt where Ps is the species permeability across the cell and S A is the surface area of the tissue. was obt ained by sol ving Eq.Chapter IV: Biofluid Mechanics of Organ Systems 353 Mass transfer in blood and tissue.e. The st eady f luid flow e quations (Eqs. For a certain time instant. 4..2.. . t )) (4.3c) ∂c Outlet and walls: =0 (4. c f . c t) wa s upd ated with Eq. Numerical method and model parameters.3. Cons idering t he mac roscopic mas s diffusion in a control volume which contains blood vessels and surrounding tissue.3. i.2) were solved fi rst t o obt ain the f low f ields.3. i. t ) Ps S A = (c f ( r. i.3. the convection-diffusion equation for chemical species reads: ∂  V   1 ∂  ∂c f  2 ∂c ∂c  ∂ cf c f + t ct  + v r f + v z z = D   r  + 2  ∂t  V f  ∂r ∂z  r ∂r  ∂r  ∂z  (4. respectively.3. Initial condition: (4 c f ( r .4.1 and 4.3a) where c f and ct are the concentration in the blood and tissue.3.4) by forward time differencing. Th e si mple di ffusion o f chemical species tr ansfer ac ross th e ce ll membrane fr om the fluid i s governed by Fick’s law. z . (4. respectively.e. (4.3). ∂c f ( r. The time step used was small enough so th at a converged solution co uld be achi eved. ct .e.. Eq.3) was a lso solved with CFX4.3.3. z.3) a t a c ertain t ime..

0272 cm2/s Diffusion coefficient.3.2 Parameter values (adapted from Lee & Rubinsky.4 depicts the fluid velocity distribution in terms of velocity vectors an d ax ial v elocity contours. S A / Vt 556 cm-1 Volume tissue/volume fluid. t he velocity in creases r apidly from t he i nlet to the out let du e t o th e co ntraction.11 Length of sinusoid. 1989). the converging flow path and increasing c ross-sectional area of t he s inusoids towa rds t he hepatic vein would change the hydraulic permeability.3 shows a pressure distribution in the acinus assuming a constant permeability. ν 0.8×10-5cm/s Kinematic viscosity.e. Vt / V f 8 Permeability in r-direction. The pr essure distribution in th e rad ial direction is also non-uniform because of the geometry effect. Ds 6. t he max imum velocity may s hift a bi t f rom the center near the outlet. p 0 2 cm H2O Porosity. In addition.03 cm Diameter of sinusoid. Clearly. I n mos t pa rts of t he acinus. Ps 4. k r<kz).3. Figure 4.3×10-6cm3g-1s Permeability in z-direction. Due to symmetry. d s 7 µm Surface area/volume tissue. Parameter Va lue Pressure portal vein. . hence. the velocity at the center is larger than that at the periphery because of the wall viscous effects and larger flow resistance in the radial direction (i. th e “s queeze” of the con tracting wal l pu shes t he fluid downwards. the pressure and velocity distributions within the acinus are non-uniform and functions of position. Ls 0.354 Biofluid Dynamics Table 4. k z 2. p1 6 cm H2O Pressure hepatic vein. ε 0. However. the pressure gradient becomes steeper close to the hepatic venu le due to the co ntraction of t he fl ow domai n st arting from th e terminal por tal venu le. Figure 4. In summary.7×10-7 cm2/s Pressure and velocity fields. which may also have a significant ef fect on pr essure and flow distributions in t he l iver ac inus (Lee & Rubinsky.6×10-6cm3g-1s Membrane permeability. Again.3.. k r 4. only half of the cone is displayed. 1989).

Fi gure 4.1 cm/s 1cm/s Axial Velocity [cm/s] 11 3. A t t he in itial s tage o f th e tra nsport pr ocess (e .02 3 0.3.68 2 0.04 3 4 6 1. th e species concentration i n t he ti ssue is m uch l ower th an that in the f luid.3 Pressure distribution in liver acinus. 9 0.38 5 7 2. the fluid concentration at the ce nter ch anges more rapidly than that at the periphery with a more rapid fluid flow.4.72 8 2. The upper half shows the v elocity vectors while the lower half depicts the axial velocity contours.40 10 3.00 Fig.Chapter IV: Biofluid Mechanics of Organ Systems 355 Pressure [Pa] 15 580 13 529 11 477 9 426 7 374 5 323 3 271 1 220 15 14 13 11 13 12 9 5 7 Fig. The t ransient responses of the f luid a nd tissue concentration i n t he liver a cinus to a s tep c hange i n t he i nlet species concentration h ave a lso b een simulated.70 2 5 1.36 4 1.5 s hows bo th the species concentration distributions in the fluid and tissue after 5s and 300s.4. which .34 1 0.06 9 2.4 Velocity dist ribution in liver acinus . t= 5s). Transient mass transfer with a step change in inlet chemical species concentration.g. Associated with the low-velocity re gion. As e xpected.3. a dis tinct low co ncentration zone ap pears ne ar t he wall.3..

00 0. The e ffluent c oncentration is onl y a sy nthesis o f t he different fluid concentrations at the outlet.40 0.4.10 0.90 0. 4.07 1.80 0.3. 0.4 0.90 0.00 1.05 0.00 0.02 0.00 0.04 1. The average fluid concentration at the outlet (i.01 0.5 Distributions of s pecies concentration in the fluid and tiss ue for a step change in the inlet concentration at t=5s and t=300s. 0.00 0.90 0.03 0.3. t=5s cf/cin t=300s cf/cin 1.06 Fig.20 8 5 0. effluent concentration) is also plotted in Fig.10 0.10 0.80 0.90 0.95 t=5s ct/cin t=300s ct/cin 0.65 0. After 300s.90 1.20 0.40 0.06 0.3.70 0.75 0 0.50 0.356 Biofluid Dynamics indicates a del ay fo r ch emical s pecies transferring f rom th e fl uid to t he tissue.30 0. 1989 ).40 0.00 0.02 01 0. which cannot reflect the multi- dimensional.95 0. 70 0.4 0.8 as a function of ti me. the comprehensive biofluid dyn amics simulations ar e ver y hel pful to ana lyze t he he patic uptake pr ocess of various chemical substances.e.70 0.05 0.70 60 0.70 0.80 0. transient m ass t ransfer pr ocesses in t he li ver (Lee & Rubinsky.30 0..00 0. Ther efore.00 0.30 0 0.3.00 0. T he concentration variations as a function of time and position can be observed more clearly from Figs.6-4.8.80 1.6 and 4. 4.90 0.85 0.00 0.20 0.03 0.3.20 0.80 0.30 0.60 0.60 0.60 0.10 0. 1 0.60 0 0.7.3. .80 0.90 0.50 0. the chemical species has already convected and diffused into most r egions of th e f luid and tissue wi thin the ac inus. As shown in Figs. both the f luid a nd t issue con centrations along the cen terline and wall i nitially have a rapid decay and then tend to be flat. 4.04 0. 0.95 0.50 0.

3.03 z (cm) Fig.8 t=50s 0.02 0.3. 4.7 Variations of tissue concentration as a function of time and position.03 z (cm) Fig.02 0.4 t=200s t=50s 0.8 0. .6 Variations of fluid concentration as a function of time and position.01 0.4 t=5s 0. Centerline 1 Wall t=200s 0.2 0 0 0.6 t=200s cf/cin t=5s t=50s 0.6 ct/cin t=50s 0.Chapter IV: Biofluid Mechanics of Organ Systems 357 Centerline 1 Wall t=200s 0.2 t=5s t=5s 0 0 0.01 0. 4.

Dep ict oxygen as well as particle transport.6 cf/cin 0. ur ine. 4. 0. l ymph. i. and tidal volume.1 Air and water-like liquids (e . re quiring l iterature rev iews. bl ood. i.5≤∀T≤2 l. plot inhaled particle depositions as a function of v olumetric l ung region. suc h as MATLAB or studentFluent <www. 1nm ≤dp≤100µm.. uptake.3.) ar e the bod y's biofluids.4 0.8 Variations of mean fluid concentration at the outlet as a function of time. 0 ≤∀T≤0. net air- volume>. 4. and migration. in halation fl ow rate. 4. 4.4 HOMEWORK PROBLEMS Note: As the C hapter II I assignments.2 Considering particle siz e. w hat are the par ticle tr ansport a nd deposition mechanisms as a f unction of particle size range and inhalation flow rate? .358 Biofluid Dynamics 1 0..2.fluent.2 0 0 100 200 300 time t (s) Fig. 4. and per haps the use of ba sic software.3 In ext ension of HWP4. Design a sys tematic bio fluid flow chart showing i ndividual pa thways and biofluid ex changes.5 l. t he following H WPs a re somewhat ope n-ended. etc.e.8 0. alo ng a pathway from mouth/nose to terminal alveoli.. cr eative thinking.g.e. 10≤Q≤60 l/min.

The pre sence of sur factant in t he pul monary an d re spiratory systems is well known to reduce the surface tension.2) From continuity. a g radient in s urface concentration may develop and therefore ∇ s ⋅ Γ may be non zero. which according t o t he Y oung-Laplace e quation wi ll re duce t he pressures required to di splace fluid with air bubble