Finite-Deformation Modeling of Elastodynamics and Smart Materials with Nonlinear Electro-Magneto-Elastic Coupling
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By
Robert Lindsey Lowe, B.S., M.S.
Graduate Program in Mechanical Engineering
The Ohio State University
2015
Dissertation Committee:
Prof. Sheng-Tao John Yu, Advisor Prof. Marcelo J. Dapino Prof. Daniel A. Mendelsohn Prof. Amos Gilat Dr. Kelly S. Carney c Copyright by
Robert Lindsey Lowe
2015 Abstract
Eulerian formulations of the equations of finite-deformation solid dynamics are ideal for numerical implementation in modern high-resolution shock-capturing schemes.
These powerful numerical techniques – traditionally employed in unsteady compress- ible flow applications – are becoming increasingly popular in the computational solid mechanics community. Their primary appeal is an exceptional ability to capture the evolution and interaction of nonlinear traveling waves. Currently, however, Eulerian models for the nonlinear dynamics of rods, beams, plates, membranes, and other elastic structures are currently unavailable in the literature.
The need for these reduced-order (1-D and 2-D) Eulerian structural models moti- vates the first part of this dissertation, where a comprehensive perturbation theory is used to develop a 1-D Eulerian model for nonlinear waves in elastic rods. The leading- order equations in the perturbation formalism are (i) verified using a control-volume analysis, (ii) linearized to recover a classical model for longitudinal waves in ultrasonic horns, and (iii) solved numerically using the novel space-time Conservation Element and Solution Element (CESE) method for first-order hyperbolic systems. Numeri- cal simulations of several benchmark problems demonstrate that the CESE method effectively captures shocks, rarefactions, and contact discontinuities.
ii The second part of this dissertation focuses on another emerging area of finite- deformation mechanics: magnetoelectric polymer composites (MEPCs). A distin- guishing feature of MEPCs is the tantalizing ability to electrically control their mag- netization, or, conversely, magnetically control their polarization. Leveraging this magnetoelectric coupling could potentially impact numerous technologies, including information storage, spintronics, sensing, actuation, and energy harvesting. Most of the research on MEPCs to date, however, has focused on optimizing the magnitude of the magnetoelectric coupling through iterative design. Substantially less activity has occurred in the way of mathematical modeling and experimental characterization at finite strains, which are needed to advance fundamental understanding of MEPCs and encourage their technological implementation.
The aforementioned need motivates the second part of this dissertation, where a finite-strain theoretical framework is developed for modeling soft magnetoelectric composites. Finite deformations, electro-magneto-elastic coupling, and material non- linearities are incorporated into the model. A particular emphasis is placed on the development of tractable constitutive equations to facilitate material characteriza- tion in the laboratory. Accordingly, a catalogue of free energies and constitutive equations is presented, each employing a different set of independent variables. The ramifications of invariance, angular momentum, incompressibility, and material sym- metry are explored, and a representative (neo-Hookean-type) free energy with full electro-magneto-elastic coupling is posed.
iii Dedicatedinlovingmemoryofmyfather
iv Acknowledgments
This dissertation represents the culmination of a rich and rewarding journey. I have reached this point by standing on the shoulders of others. I will never be able to repay my debt of gratitude in kind, but I will do my best to pay it forward.
I would like to begin by expressing my sincere appreciation to my advisor, Pro- fessor Sheng-Tao John Yu, for his guidance, advice, support, and friendship through the years. I thoroughly enjoyed our conversations – both technical and otherwise – and the wisdom and perspective I gained through our interactions. Prof. Yu’s words of encouragement always gave me hope and inspiration when times were challenging.
I profited greatly from his expertise in numerical methods and programming.
I would also like to thank my former advisor, Professor Stephen Bechtel, who was forced to retire far too young due to the devastating effects of Parkinson’s dis- ease. Prof. Bechtel taught me the importance of rigour, clarity, and perfectionism in scientific work as well as the art of technical writing. These lessons have had an indelible impression on me. I feel privileged to have been selected by him to convert his prized set of continuum mechanics course notes into a published textbook. Prof.
Bechtel’s unique perspective on the topic and legacy as a continuum mechanician live on through our book.
I would like to express my appreciation to Professor Marcelo Dapino, Professor
Dan Mendelsohn, Professor Amos Gilat, and Dr. Kelly Carney for serving on my
v doctoral committee. Each of them has willingly offered support, guidance, and ad- vice when I needed it. Prof. Dapino’s input on the smart materials portion of this dissertation brought a much-needed shift toward developing theories that would be useful in practical applications. I am thankful for the opportunity to have taken Prof.
Mendelsohn’s course in elastic wave propagation and Prof. Gilat’s course in plasticity, both of which were foundational in my research. Their help and friendship over the years are also acknowledged. Dr. Carney’s comments on the computational aspects of my work are also much appreciated; I look forward to working together more in the coming years.
I would like to acknowledge several faculty members at OSU who have provided mentoring, guidance, support, and friendship, including Professors Carlos Castro,
Brian Harper, Gary Kinzel, Cheena Srinivasan, and Denny Guenther. A hearty thank you is also due to Ms. Janeen Sands and Mr. Nick Breckenridge in the Graduate
Advising Office. Briefly stated, both are tremendous assets to our Department. I am particularly thankful for Janeen’s compassion, willingness to listen, and helpful advice, which helped me weed my way though several challenging situations.
I would like to express my gratitude to several friends with whom I’ve had the pleasure of collaborating: First, I’d like to thank Dr. Po-Hsien Lin for performing the numerical computations in Chapter 2. I’ve enjoyed and profited greatly from our conversations about hyperbolic systems and numerical methods. Secondly, I’d like to thank Professors Monon Mahboob and Md. Zahabul Islam for a fruitful inter- national collaboration that resulted in three journal publications on the mechanics of nanoscale structures. Thirdly, I’d like to thank Professor Sushma Santapuri for
vi numerous discussions on the nuances of continuum electrodynamics and an enjoy-
able collaboration in this field. I would also like to thank Dr. Lixiang Yang for a
productive collaboration during his time at OSU and for many helpful conversations.
Next, I would like to thank Professor Prasad Mokashi for his close friendship and support throughout the years. I would also like to thank Dr. Hafez Tari, who has become a dear friend to my wife and I in but a few short years in Columbus. I extend many thanks to my amigos Professor Chris Cooley, Mr. Tom Walters, Dr. Jeremy
Seidt, Dr. David Bilyeu, and Dr. Krista Kecskemety for sharing the joys (mostly) and frustrations (some) of graduate life over hot and cold beverages alike. A special thanks as well to my friends outside of graduate school – Shawn, Kyle, Nate, Brad, and many others – for road trips, rounds of golf, and chatting about sports and other topics over the occasional frosty beverage.
I acknowledge the generous financial support I received throughout my doctoral studies, most notably a Presidential Fellowship from the OSU Graduate School, a
Graduate Teaching Fellowship from ASME, and a research assistantship from NSF. I am grateful to Department Chairman Ahmet Selamet and Associate Chairman Dan
Mendelsohn for their financial support during several times of need. I would also
like to thank former Department Chairman Cheena Srinivasan and former Associate
Chairman Gary Kinzel for the opportunity to serve as an independent course in-
structor. I also gratefully acknowledge travel support from The Honor Society of Phi
Kappa Phi and the OSU Council of Graduate Students.
I would like to extend a special thanks to my family, particularly my wife Karen
and my mother Kristine. Karen – words cannot convey how grateful I am for your
enduring love, support, patience, understanding, and encouragement during this long
vii journey. We did it! Mom – I will never be able to repay you for your unwavering
love, devotion, and selflessness, but know that it is felt and cherished. I truly hope
I can be half the parent you are. A loving thank you is also due to my Aunt Doey,
whose weekly phone calls always provided me with a few good laughs, some lively chat about sports, and all the latest news in Bryan, Ohio. A heartfelt thank you to my Uncle John & Aunt Diane and cousins Kirk & Amy, who lovingly looked out for me during my years in Columbus. And last, but certainly not least, much love and
many thanks to my in-laws Ed, Barbara, Thomas, and Kim for their love, support,
and unwavering belief in me.
I am saddened that my father Bob, who passed away in 1997, is not here to
celebrate this milestone with me; I dedicate this dissertation to his memory.
Lastly, I thank God for His presence, for always being by my side, and for bringing
all of these outstanding people into my life. I am truly blessed beyond words.
viii Vita
October 16, 1980 ...... Born- Toledo,OH
June 1999 ...... Diploma, Bryan High School, Bryan, OH
September 1999 - May 2003 ...... Presidential Scholar, T.J. Smull Col- lege of Engineering, Ohio Northern University, Ada, OH
May 2003 ...... B.S. Mechanical Engineering, with High Distinction, Ohio Northern Uni- versity, Ada, OH
August 2005 ...... M.S. Mechanical Engineering, The Ohio State University, Columbus, OH
September 2006 - present ...... Ph.D. Student, ASME Graduate Teaching Fellow, and Presidential Fellow, The Ohio State University, Columbus, OH
Publications
Books
S.E. Bechtel and R.L. Lowe. “Fundamentals of Continuum Mechanics: With Appli- cations to Mechanical, Thermomechanical, and Smart Materials.” Elsevier Academic Press, San Diego, Nov. 2014.
ix Refereed Journal Articles
R.L. Lowe, S.-T.J. Yu, L. Yang, and S.E. Bechtel. “Modal and Characteristics- Based Approaches for Modeling Elastic Waves Induced by Time-Dependent Boundary Conditions.” Journal of Sound and Vibration 333(3), pp. 873-886, Feb. 2014.
S. Santapuri, R.L. Lowe, S.E. Bechtel, and M.J. Dapino. “Thermodynamic Mod- eling of Fully Coupled Finite-Deformation Thermo-Electro-Magneto-Mechanical Be- havior for Multifunctional Applications.” International Journal of Engineering Sci- ence 72, pp. 117-139, Nov. 2013.
M.Z. Islam, M. Mahboob, R.L. Lowe, and S.E. Bechtel. “Characterization of the Thermal Expansion Properties of Graphene Using Molecular Dynamics Simulations.” Journal of Physics D: Applied Physics 46(43), 435302, Oct. 2013.
M. Mahboob, M.Z. Islam, R.L. Lowe, and S.E. Bechtel. “Molecular Dynamics and Atomistic Finite Element Simulation Studies of the Effect of Stone-Wales Defects on the Mechanical Properties of Carbon Nanotubes.” Nanoscience and Nanotechnology Letters 5(9), pp. 941-951, Sept. 2013.
S.-T.J. Yu, L. Yang, R.L. Lowe, and S.E. Bechtel. “Numerical Simulation of Linear and Nonlinear Waves in Hypoelastic Solids by the CESE Method.” Wave Motion 47(3), pp. 168-182, Apr. 2010.
L. Yang, R.L. Lowe, S.-T.J. Yu, and S.E. Bechtel. “Numerical Solution by the CESE Method of a First-Order Hyperbolic Form of the Equations of Dynamic Nonlinear Elasticity.” ASME Journal of Vibration and Acoustics 132(5), 051003, Oct. 2010.
Fields of Study
Major Field: Mechanical Engineering
Studies in: Theoretical & Applied Mechanics
x Table of Contents
Page
Abstract...... ii
Dedication...... iv
Acknowledgments...... v
Vita...... ix
ListofTables...... xv
ListofFigures...... xvi
1. AnIntroductiontoEulerianApproachestoElastodynamics...... 1
1.1Overviewandresearchopportunity...... 1 1.2Backgroundandpreliminaries...... 3 1.2.1 Eulerian forms of the fundamental laws of continuum mechanics 3 1.2.2 ConstitutivemodelinginEulerianfiniteelasticity...... 6 1.2.3 A brief review of Eulerian and Lagrangian approaches to com- putationalmechanics...... 10 1.2.4 AbriefoverviewoftheCESEmethod...... 13 1.2.5 Existing 1-D models for the nonlinear elastodynamics of rods 16 1.3Objectives,structure,andnovelcontributions...... 17
2. An Eulerian Model for Nonlinear Waves in Elastic Rods, Solved Numeri- callybytheCESEMethod...... 20
2.1Introduction...... 21 2.2The2-Dmathematicalmodel:Eulerianformulation...... 27 2.2.1 Descriptionoftheproblem...... 27 2.2.2 3-Dgoverningequations...... 27
xi 2.2.3 3-D boundary conditions at the lateral surface ...... 29 2.2.4 Specializationofthemodelto2-D...... 30 2.3Derivationoftheleading-order1-Dmodel...... 33 2.3.1 Thedimensionless2-Dmodel...... 33 2.3.2 Perturbationformalism...... 36 2.3.3 Theelementary1-Dtheory(leading-ordermodel)...... 39 2.4Analternativedevelopmentoftheelementary1-Dtheory...... 43 2.4.1 Conservationofmass...... 46 2.4.2 Balanceoflinearmomentum...... 47 2.5Linearizationoftheelementary1-Dtheory...... 49 2.6Numericalimplementationoftheelementary1-Dtheory...... 52 2.6.1 Mathematicalstructureofthemodel...... 52 2.6.2 Conservativeform...... 55 2.6.3 TheCESEmethod...... 57 2.6.4 Benchmarkproblems...... 58
3. Modal and Characteristics-Based Approaches for Modeling Elastic Waves Induced by Time-Dependent Boundary Conditions ...... 69
3.1Introduction...... 71 3.2Illustrativeelastodynamicproblem...... 74 3.3Solutionoftheproblembymodalanalysis...... 76 3.3.1 Theconcentratedbodyforcemethod(CBFM)...... 77 3.3.2 The homogeneous eigenfunction expansion method (HEEM) 81 3.3.3 TheMindlin-Goodmanmethod...... 82 3.4Solutionoftheproblembythemethodofcharacteristics...... 85 3.4.1 Mathematical structure of the first-order linear system . . . 86 3.4.2 Implementationofthemethodofcharacteristics...... 88 3.4.3 Resultsanddiscussion...... 91
4. ConclusionsandFutureWorkinEulerianElastodynamics...... 93
4.1Conclusions...... 93 4.2Futurework...... 98 4.2.1 Finite-deformationelasticity...... 98 4.2.2 Fluid-structureinteraction...... 100 4.2.3 Finite-deformationplasticity...... 100
5. AnIntroductiontoSmartPolymers...... 106
5.1Overviewandresearchopportunity...... 106 5.2Abriefreviewofseveralclassesofsmartpolymers...... 108
xii 5.2.1 Magnetorheologicalelastomers...... 108 5.2.2 Magnetoelectricpolymers...... 111 5.3Objectives,structure,andnovelcontributions...... 114
6. Revisiting the Fundamental Laws of Continuum Electrodynamics .... 116
6.1Introduction...... 116 6.2 Primitive, material, and integral versions of the fundamental laws . 119 6.2.1 Notationandnomenclature...... 120 6.2.2 Conservationofmass...... 121 6.2.3 Balanceoflinearmomentum...... 123 6.2.4 Balanceofangularmomentum...... 125 6.2.5 Firstlawofthermodynamics...... 127 6.2.6 Secondlawofthermodynamics...... 129 6.2.7 Conservationofelectriccharge...... 131 6.2.8 Faraday’slaw...... 134 6.2.9 Gauss’slawformagnetism...... 135 6.2.10Gauss’slawforelectricity...... 136 6.2.11 Amp`ere-Maxwelllaw...... 138 6.3Thelocalizationtheorem...... 140 6.4 Pointwise versions of the fundamental laws ...... 140 6.4.1 Eulerianforms...... 141 6.4.2 Lagrangianforms...... 145 6.5 Transformations between Eulerian and Lagrangian quantities . . . 149 6.6Someobservations...... 150 6.7 Two-way coupling between thermomechanics and electromagnetism 158 6.7.1 Effectiveelectromagneticfields...... 158 6.7.2 Electromagneticallyinducedcouplingterms...... 161 6.8MaxwellstressandtotalCauchystress...... 162
7. A Finite-Deformation Framework for Nonlinear Electro-Magneto-Elastic Materials, with Application to Magnetoelectric Polymer Composites . . . 164
7.1Introduction...... 164 7.2 Kinematics, fundamental laws, and boundary conditions ...... 170 7.2.1 Kinematics...... 171 7.2.2 Fundamental laws ...... 172 7.3Constitutiveequations...... 178 7.3.1 EnergyformulationswithEulerianIVs...... 179 7.3.2 EnergyformulationswithLagrangianIVs...... 189 7.3.3 Augmentedfreeenergyformulations...... 199 7.3.4 Invariancerequirements...... 204
xiii 7.3.5 Incompressibility ...... 205 7.3.6 Materialsymmetry...... 206 7.3.7 Arepresentativefreeenergyfunction...... 211
8. ConclusionsandFutureWorkinSmartPolymers...... 213
8.1Conclusions...... 213 8.2Futurework...... 214
Appendices 217
A. Supplement to Chapter 2 ...... 217
A.1 Extended mathematical structure of the perturbation formalism . . 217 A.2Thelinearizedtheory...... 221 A.2.1Mathematicalstructure...... 222 A.2.2Theequationofmotionforaninitiallystraightrod..... 224 A.2.3Thed’Alembertsolutionoftheinitial-valueproblem.... 224 A.2.4Velocity,stress,strain,density,andarea...... 227 A.2.5Impact...... 229 A.2.6Separation...... 232
B. Supplement to Chapter 3 ...... 234
B.1ReformulationoftheproblembytheCBFM...... 234 B.2ThespecializedCBFM/HEEMsolution...... 236 B.3ThespecializedMindlin-Goodmansolution...... 237 B.4Convergenceandterm-by-termdifferentiation...... 238
C. Supplement to Chapter 6 ...... 241
D. Supplement to Chapter 7 ...... 260
Bibliography...... 283
xiv List of Tables
Table Page
3.1 Material properties, geometric parameters, and boundary conditions . 76
6.1 Units for various electrical, magnetic, mechanical, and thermal quantities.154
7.1FreeenergiescorrespondingtodifferentsetsofEulerianIVs..... 180
7.2 Constitutive equations for energy formulations with Eulerian IVs . . . 181
7.3 Constitutive equations for energy formulations with Lagrangian IVs . 190
7.4 Constitutive equations for ‘augmented’ energy formulations with La- grangianIVs...... 201
7.5Invariantconstitutiveequations...... 205
xv List of Figures
Figure Page
2.1 A control volume (dashed lines) occupied at time t by a differential element of a tapered rod (solid lines). The axial control surfaces are fixed in space but undergo cross-sectional area variation. The top and bottom lateral control surfaces co-deform with the lateral boundaries ofthedifferentialelement...... 44
2.2 Low-speed (20 m/s) impact problem. Snapshot t =3μs after impact illustrating the analytical (solid) and numerical (circles) wave profiles for (a) density, (b) velocity, (c) area, and (d) stress. Both the striker bar (initially moving at 20 m/s) and the incident bar (initially stationary) are composed of 7075-T651 aluminum alloy. The impact interface is located at z = 5 cm. Two weak shocks travel away from the impact interface, both propagating compression and lateral expansion in their wake...... 60
2.3 High-speed (6000 m/s) impact problem. Snapshot t =4.8 μsafter impact illustrating the analytical (solid) and numerical (circles) wave profiles for (a) density, (b) velocity, (c) area, and (d) stress. Both the striker bar (initially moving at 6000 m/s) and the incident bar (initially stationary) are composed of 7075-T651 aluminum alloy. The impact interface is located at z = 5 cm. Two strong shocks travel away from the impact interface, both propagating compression and lateral expan- sion in their wake. A contact wave (or contact discontinuity) is also clearly discernible in the density, area, and stress profiles. Substantial discrepancies in wave amplitudes and wave speeds are observed, owing totheeffectsofnonlinearity...... 61
xvi 2.4 A comparison of the numerical predictions of three different objective rates – Jaumann-Zaremba (circles), Oldroyd (squares), and Cotter- Rivlin (diamonds) – with an analytical solution (solid line) of a low- speed (20 m/s) impact. As expected, good agreement is observed at lowspeeds(linearregime)...... 64
2.5 A comparison of the numerical predictions of three different objective rates – Jaumann-Zaremba (circles), Oldroyd (squares), and Cotter- Rivlin (diamonds) – with an analytical solution (solid line) of a high- speed (6000 m/s) impact. As expected, different rates yield substan- tially different physical predictions, and all deviate appreciably from theanalyticalsolution,owingtotheeffectsofnonlinearity...... 64
2.6 Low-speed separation problem (Vo = 10 m/s). Snapshot at t =4μs il- lustrating the analytical (solid) and numerical (circles) wave profiles for (a) density, (b) velocity, (c) area, and (d) stress. The rod is composed of UNS C15720 copper alloy. Two weak rarefactions travel away from the separation interface at z = 5 cm, both propagating decompression, tension,andlateralcontractionintheirwake...... 66
2.7 High-speed separation problem (Vo = 1000 m/s). Snapshot at t = 6 μs illustrating the analytical (solid) and numerical (circles) wave profiles for (a) density, (b) velocity, (c) area, and (d) stress. The rod is composed of UNS C15720 copper alloy. Two strong rarefactions travel away from the separation interface at z = 5 cm, both propagating decompression,tension,andlateralcontractionintheirwake...... 67
3.1 (a) A snapshot at t = 240 μs illustrating a propagating stress wave generated at the forced boundary traveling leftward prior to reflection. (b)-(e) Zoomed regions near the forced boundary, indicated by the box in (a). Solutions were obtained via the concentrated body force method and the homogeneous eigenfunction expansion method (solid lines), the Mindlin-Goodman method (dashed lines), and the method of characteristics (squares). The mode superposition solutions were truncated at (b) 500 modes, (c) 1000 modes, (d) 2500 modes, and (e) 5000 modes...... 79
xvii 3.2 (a) A snapshot at t =4.1 ms illustrating the stress profile resulting from the interference of leftward-propagating incident waves from the forced end with reflected waves traveling inward from the boundaries. (b)-(e) Zoomed regions near the forced boundary, indicated by the box in (a). Solutions were obtained via the concentrated body force method and the homogeneous eigenfunction expansion method (solid lines), the Mindlin-Goodman method (dashed lines), and the method of characteristics (squares). The mode superposition solutions were truncated at (b) 500 modes, (c) 1000 modes, (d) 2500 modes, and (e) 5000 modes. Note that the stresses are still well within the elastic regimeforthisparticularmaterial...... 80
3.3 The characteristic mesh for two time steps, illustrated for the special case N = 4. Along the line t = 0 spanning 0 ≤ ξ ≤ L, we introduce N nodes uniformly spaced Δξ apart. A right-running characteristic emerges from the left boundary node (i =1), a left-running char- acteristic emerges from the right boundary node (i = N), and both left-running and right-running characteristics emerge from the interior nodes. As these characteristics advance in time, pairs of them inter- sect, first at time t =Δt,thenatt =2Δt,andsoon,withthetime step Δt dictated by the nodal spacing Δξ and the constant slope ±c of the characteristics. Arrows denote the direction that information propagates along the characteristic lines, and filled circles designate nodes...... 91
6.1 Subsets S1 and S2 as seen in the present configuration of body B. Subset S1 is an open volume P bounded by a closed surface ∂P, while subset S2 is an open surface Q bounded by a closed curve ∂Q..... 121
xviii Chapter 1: An Introduction to Eulerian Approaches to Elastodynamics
This chapter begins the first part of the dissertation, where the focus is on Eulerian
approaches to mathematically modeling and numerically simulating inertia-driven
problems in solid mechanics.
1.1 Overview and research opportunity
Finite-element codes that employ a co-deforming mesh (Lagrangian approach) cur- rently dominate the landscape in computational solid mechanics. However, numerical algorithms that employ a fixed mesh (Eulerian approach) are beginning to attract sig- nificant attention – particularly for solving inertia-driven (dynamic) problems such as impacts and blasts in finite-deformation elasticity [1–16] and plasticity [17–32].
Generally speaking, Eulerian approaches employ a deformed-configuration (Eulerian) formulation of the governing equations, which are then discretized on a fixed mesh and solved numerically by a finite-difference or finite-volume scheme.
Eulerian approaches are appealing for several reasons. First, the mathematical structure of the governing equations enables the use of modern numerical methods for first-order hyperbolic systems, which are explicitly designed for resolving highly
1 nonlinear, inertia-driven physics. Consequently, Eulerian approaches are adept at tracking the evolution and interaction of nonlinear waves such as shocks and rar- efactions. Secondly, fixed grids are generally regarded as the most natural setting for shock capturing [30]. Thirdly, since the continuum is deforming through a sta- tionary grid, arbitrarily large strains and rotations can be accommodated without excessive grid distortion or remeshing. Finally, Eulerian approaches to solid dynam- ics have opened the door for solving multiphase problems on a single fixed mesh with a single solver, greatly simplifying the numerical treatment of solid-fluid interfaces in
fluid-structure interaction problems [33–40].
Rod-like structures are used in dynamic applications such as ultrasonics, acous- tic wave sensing, ballistic impact, structural analysis, nondestructive evaluation, flaw detection, high-strain-rate material testing, and percussion drilling. To model the nonlinear elastic waves that often arise in these applications, approximate 1-D rod theories are developed from the full 3-D dynamic finite elasticity theory [41–66]. The primary appeal of specialized 1-D theories over the general 3-D theory is the reduced order of the governing equations, which depend on just one spatial coordinate and time. Practically, this translates to more straightforward computational implemen- tation and reduced code execution times.
To the best of our knowledge, elastodynamic rod theories formulated in the present configuration (i.e., the Eulerian representation) are presently unavailable in the lit- erature, with the exception of some related ideas set forth by Yang and cowork- ers [67, 68]. An Eulerian rod theory is attractive for several reasons: First, it could be implemented in a computer code employing a fixed grid and a modern numerical
2 method for capturing hyperbolic wave physics. The resulting framework could poten-
tially achieve unprecedented shock capturing and wave tracking capabilities in elastic
rod-like structures. Secondly, it could be useful for validating new finite-difference
and finite-volume codes in a simple solid-dynamics setting. Lastly, it could be in-
corporated within a unified Eulerian computational architecture for fluid-structure interaction, wherein the dynamics of the solid and fluid phases are simultaneously
tracked on a single fixed mesh.
The remainder of this chapter is structured as follows: In Section 1.2, we pro-
vide some relevant background from continuum mechanics, finite elasticity, numerical
methods, and structural vibrations, with a particular focus on Eulerian aspects of
modeling and computation. Then, in Section 1.3, we highlight the objectives and
novel contributions of this part of the dissertation.
1.2 Background and preliminaries
1.2.1 Eulerian forms of the fundamental laws of continuum mechanics
In continuum mechanics, the motion
x = χ(X,t) (1.1)
maps each particle in the body B from its reference position X to its present position
x. Consequently, the body itself is mapped from its reference (or undeformed) config-
uration to its present (or deformed) configuration. In the deformed configuration, the
body occupies open volume R bounded by closed surface ∂R, and subset S occupies
open volume P bounded by closed surface ∂P.
Conservation of mass postulates that the mass of every subset of the body is
constant throughout its motion, or, equivalently, the time rate of change of the mass
of every subset is zero. This can be expressed in Eulerian integral form as [69] d ρdv =0, (1.2) dt P where ρ is the mass density and dv is a volume element, both in the present config- uration. Note that to perform the integration in (1.2), ρ is regarded in its Eulerian description, i.e., as a function of present position x and time t.
Provided the density is sufficiently smooth, the Eulerian differential form of con- servation of mass follows from (1.2) as [69]
∂ρ + div(ρv)=0, (1.3a) ∂t or, alternatively,
∂ρ +(v · grad)ρ + ρ divv =0. (1.3b) ∂t
In (1.3a) and (1.3b), v is the velocity, “grad” denotes the Eulerian gradient operator
(i.e., the gradient calculated with respect to the present configuration), “div” denotes the Eulerian divergence operator, and ( ) · ( ) denotes an inner product. Equation
(1.3a) is referred to as the conservative form of conservation of mass, while (1.3b) is called the non-conservative form [70].
4 Balance of linear momentum
Balance of linear momentum postulates that the time rate of change of the linear
momentum of any subset of the body is equal to the resultant external force acting
on that subset. This can be expressed in Eulerian integral form as [69] d vρdv = f m ρdv + t da, (1.4) dt P P ∂P where f m is the mechanically induced body force (per unit mass), t is the traction
(per unit deformed area), and da is a surface element in the present configuration.
Note that f m accounts for long-range forces such as gravity, while t accounts for
short-range contact forces acting on the surface of the subset.
Provided the fields are sufficiently smooth, the Eulerian differential form of balance
of linear momentum follows from (1.4) as [69]
∂(ρv) +divρv ⊗ v = ρf m +divT, (1.5a) ∂t or, alternatively, ∂v ρ +(v · grad)v = ρf m +divT, (1.5b) ∂t where T is the Cauchy stress and ( ) ⊗ ( ) denotes the dyadic (tensor) product of two vectors. Equations (1.5a) and (1.5b) are referred to the conservative and non- conservative forms, respectively, of balance of linear momentum.
Balance of angular momentum
Balance of angular momentum postulatesthatthetimerateofchangeofthe angular momentum of any subset of the body about the origin is equal to the resultant
5 external moment acting on that subset about the origin. This can be expressed in
Eulerian integral form as [69] d x × vρdv = x × f m ρdv + x × t da, (1.6) dt P P ∂P
where x is the present position of a continuum particle and ( ) × ( ) denotes the cross
product of two vectors. The local form of (1.6) is [69]
T = TT. (1.7)
Hence, the Cauchy stress tensor is symmetric.
1.2.2 Constitutive modeling in Eulerian finite elasticity
In finite elasticity, hyperelastic (or Green-elastic) materials are those for which the
Cauchy stress T is derivable from a strain energy potential W . In the isotropic case, the strain energy W (per unit reference volume) is a function of the deformation only through the principal invariants I1, I2,andI3 of the left Cauchy-Green deformation tensor B, i.e.,