H. Eschenauer, N. Olhoff, W. Schnell Applied Structural Mechanics Fundamentals of Elasticity, Load-Bearing Structures, Structural Optimization Including Exercises With 179 Figures e SpringerProf. Dr.-Ing. H. Eschenauer TA University of Siegen 64 Research Center for Multidisciplinary Analyses eo and Applied Structural Optimization FOMAAS cts Institute of Mechanics and Control Engineering 1907 1) - 87068 Siegen / Germany ¢ Prof. Dr. techn. N. Othoff Aalborg University Institute of Mechanical Engineering DK - 9220 Aalborg East / Denmark Prof. Dr. Dr.-Ing. E. h. W. Schnell Technical University of Darmstadt Institute of Mechanics D - 64289 Darmstadt / Germany ISBN 3-540-61232-7 Springer-Verlag Berlin Heidelberg New York Dic Deutsche Bibliothek - CIP-Einheitsaufiahme Eschenauer, Hans A.: Applied structural mechanics: fundamentals of elasticity. load bearing structures, structural optimization: including exercises / H. Eschenauer: N. Olhoff. W. Schnell, ~ Berlin; Heidelberg; New York; Barcelona, Budapest; Hong Kong; London; Milan; Paris; Santa Clara; Singapur; Tokyo: Springer, 1997 1232-7 Schnell, Walter ‘This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuseof illustrations, recitation, broadcasting, reproduction on microfilm or in other ways, and storage in data banks, Duplication ofthis publication ‘or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer- Verlag, Violations are liable for prosecution act under German Copyright Law. © Springer-Verlag Berlin Heidelberg 1997 Printed in Germany ‘The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, cven in the absence of a specific statement, that such names ate exempt from the relevant protective laws and regulations and therefore free for general use. Produet liability: The publisher cannot guarantee the accuracy of any information about dosage and application contained in thie book. In every individual case the user must check such information by consulting the relevant literature. ‘Typesetting: Camera-ready by editors SPIN:10508157 61/3020-5 432 10 - Printed on acid -free paperPreface The present English-language work is a compilation of the two-volume 3rd edition (in German ) of ® Elastizitatstheorie” (1993, 1994) published by BI- Wissenschafteverlag Mannheim, Leipsig, Wien, Ziirich. Since the first edi- tion of this book had appeared in 1983, the fundamental concept of this book has remained unaltered, in spite of an increasing amount of structu- ral-analytical computation software (eg. Finite Element Methods ). The importance of computer-tools, may this be supercomputers, parallel compu- ters, or workstations, is beyond discussion, however, the responsible engineer in research, development, computation, design, and planning should always be aware of the fact that a sensible use of computer-systems requires a re- alistic modeling and simulation and hence respective knowledge in solid mechanics, thermo~ and fluiddynamics, materials science, and in further disciplines of engineering and natural sciences. Thus, this book provides the basic tools from the field of the theory of elasticity for students of natural sciences and engineering; besides that, it aims at assisting the engineer in an industrial environment in solving current problems and thus avoid a mere black-box thinking. In view of the growing importance of product lia- bility as well as the fulfilment of extreme specification requirements for new products, this practice—relevant approach plays a decisive role. Apart from a firm handling of software systems, the engineer must be capable of both the generation of realistic computational models and of evaluating the computed results. Following an outline of the fundamentals of the theory of elasticity and the most important load-bearing structures, the present work illustrates the transition and interrelation between Structural Mechanics and Structural Optimization. As mentioned before, a realistic modeling is the basis of every structural analysis and optimization computation, and therefore nu- merous exercises are attached to each main chapter. By using tensor notation, it is attempted to offer a more general insight into the theory of elasticity in order to move away from a mere Cartesian view. An arbitrarily shaped” solid described by generally valid equations shall be made the object of our investigations (Main Chapter A ). Both the condi- tions of equilibrium and the strain—displacement relations are presented for large deformations (nonlinear theory ); this knowledge is of vital import- ance for the treatment of stability problems of thin-walled load-bearingstructures. When deriving the augmented equations as well as the corre- sponding solution procedures, we limit our considerations to the most essen- tial aspects. All solution methods are based on the HOOKEAN concept of the linear-elastic solid. As examples of load-bearing structures, disks, plates and shells will be treated in more detail (Main Chapters B,C). Fi- nally, an introduction into Structural Optimization is given in order to illus- trate ways of determining optimal layouts gf load-bearing structures (Main Chapter D). In the scope of this book, the most important types of exercises arising from each Main Chapter are introduced, and their solutions are presented as comprehensively as necessary. However, it is highly recommended for the reader to test his own knowledge by solving the tasks independently. When tueating structural optimization problems a large numerical effort generally occurs that cannot be handled without improved programming skills. Thus, at corresponding tasks, we restrict ourselves to giving hints and we have consciously avoided presenting details of the programming. ‘The authors would like to express their gratitude to all those who have as- sisted in preparing the camera—ready pages, in translating and proofreading as well as in drawing the figures. At this point, we would like to thank Mrs A. Wachter-Freudenberg, Mr K. Gesenhues, and Mr M. Wengenroth who fulfilled these tasks with perseverance and great patience We further ac- knowledge the help of Mrs Dipl.-Ing. P. Neuser and Mr Dipl.-Ing. M. Seibel in proofreading. Finally, we would aleo like to express our thanke to the publicher, and in particular to Mrs E. Raufelder, for excellent cooperation. Hans Eschenauer Niels Olhoff Walter Schnell Siegen/ GERMANY Aalborg / DENMARK Darmstadt / GERMANY April 1996Contents List of symbols 1 Introduction A Fundamentals of elasticity ~ Chapter 2 to 7— A.1 Definitions — Formulas — Concepts 2 Tensor algebra and analysis 21 ‘Terminology - definitions 22 Index rules ahd summation convention 23 Tensor of first order (vector ) 24 Tensors of second and higher order 25 Curvilinear coordinates 3 State of stress 31 Stress vector 32 Stress tensor 33 Coordinate transformation - principal axes 34 Stress deviator 35 Equilibrium conditions 4 State of strain 41 Kinematics of a deformable body 42 Strain tensor 43 Strain-displacement relations 44 Transformation of principal axes 45 Compatibility conditions 5 Constitutive laws of linearly elastic bodies 51 Basic concepts 52 Generalized HOOKE-DUHAMEL's law 5.3 Material law for plane states 54 Material law for a unidirectional layer (UD-layer) of a fibre reinforced composite xm 18 18 20 2 2a 26 29 3L 3 31 3L 32 35 37VIII Contents 6 Energy principles 39 61 Basic terminology and assumptions 39 62 Energy expressions 40 63 Principle of virtual displacements (Pvd ) 44 64 Principle of virtual forces ( Pvf) 44 65 Reciprocity theorems and Unit-Load -Method 46 66 Treatment of a variational problem 46 67 Approximation methods for continua aT 7 Problem formulations in the theory of linear 48 elasticity a Basic equations and boundary-value problems 48 72 Solution of basic equations 49 73 Special equations for three-dimensional problems 49 74 Special equations for plane problems 50 75 Comparison of state of plane stress and state of plane 51 strain A.2 Exercises 53 A-2-1: Tensor rules in oblique base 53 A-2-2: Analytical vector expressions for a parallelogram disk 60 A-2-3: Analytical vector expressions for an elliptical hole in 63 elliptical-hyperbolical coordinates A-3-1: MOHR's circle for a state of plane stress 66 A-3-2: Principal stresses and axes of a three-dimensional state 67 of stress A-3-3: Equilibrium conditions in elliptical-hyperbolical coordi- 70 nates (continued from A-2-3) A-4-1: Displacements and compatibility of a rectangular disk a A-4-2: Principal strains from strain gauge measurements B A-4-3: Strain tensor, principal strains and volume dilatation of “4 a three-dimensional state of displacements A-4-4: Strain-displacement relation and material law in ellipti- 6 cal-hyperbolical coordinates (continued from A-2-3) A.5A: Steel ingot in a rigid concrete base 78 A641: Differential equation and boundary conditions for a 80 BERNOULLI beam from a variational principle A.6-2: Basic equations of linear thermoelasticity by HELLIN- 82 GER / REISSNER's variational functional A-6-3: Application of the principle of virtual displacements for 83 establishing the relations of a triangular, finite element A-T1: Hollow sphere under constant inner pressure 86 A-7-2: Single load acting on an elastic half-space - Applica- 89 tion of LOVE's displacement functionContents x B_ Plane load-bearing structures 93 — Chapter 8 to 10 — B.1 Definitions — Formulas — Concepts 93 8 Disks 93 &1 Definitions - Assumptions - Basic Equations 93 82 Analytical solutions to the homogeneous bipotential 95 equation 9 «Plates 99 91 Definitions — Assumptions - Basic Equations 99 92 Analytical solutions for shear-rigid plates 107 10 Coupled disk—plate problems ua 101 Isotropic plane structures with large displacements 3 102 —_ Load-bearing structures made of composite materials us B.2 Exercises 123 B-8-1. Simply supported rectangular disk under constant load 123 B8-2 Circular annular disk subjected to a stationary tempera- 128 ture field B-8-3: Rotating solid and annular disk 131 B-8-4: Clamped quarter—circle disk under a single load 133 Semi-infinite disk subjected to a concentrated moment 137 Circular annular CFRP-disk under several loads 139 Infinite disk with an elliptical hole under tension M45 Infinite disk with a crack under tension 151 Shear-rigid, rectangular plate subjected to a triangular 153 load B-9-2; Shear~stiff, semi-infinite plate strip under a boundary 155 moment B-9-3: Rectangular plate with two elastically supported bound- 187 aries subjected to a temperature gradient field B-9-4: Overall clamped rectangular plate under a constantly 167 distributed load B.9-5; Rectangular plate with mixed boundary conditions un- 170 der distributed load B-9-6: Clamped circular plate with a constant circular line load 172 B-9-7; Clamped circular ring plate with a line load at the outer Ww boundary B-9-8: Circular plate under a distributed load rested on an ela- 179 stic foundationx Contents B-9-9: Centre-supported circular plate with variable thickness under constant pressure load. B-10-1: Buckling of a rectangular plate with one stiffener B-10-2 Clamped circular plate under constant pressure consi- dered as a coupled disk-plate problem C Curved load-bearing structures — Chapter 11 to 14 — C.1 Definitions — Formulas — Concepts 11 General fundamentals of shells ll Surface theory - description of shells 12 Basic theory of shells . 13 Shear-rigid shells with small curvature 12 Membrane theory of shells 121 General basic equations 122 Equilibrium conditions of shells of revolution 123 Equilibrium conditions of translation shells 12.4 Deformations of shells of revolution 12.5 Constitutive equations — material law 126 Specific deformation energy 13 Bending theory of shells of revolution 131 Basic equations for arbitrary loads 13.2 Shells of revolution with arbitrary meridional shape — Transfer Matrix Method 133 Bending theory of a circular cylindrical shell 14 Theory of shallow shells 41 Characteristics of shallow shells 142 Basic equations and boundary conditions 143 Shallow shell over a rectangular base with constant principal curvatures C2 Exercises C-11-l: Fundamental quantities and equilibrium conditions of the membrane theory of a circular conical shell C-12-1: Shell of revolution with elliptical meridional shape sub- jected to constant internal pressure C-12-2: peherical boiler under internal pressure and centrifugal rce 183 188 195 199 199 199 199 209 213 214 214 215 218 220 221 221 222 222 228 241 zai 242 245, 27 247 251Contents XI Spherical shell under wind pressure 255 : Hanging circular conical shell filled, with liquid 258 : Circular toroidal ring shell subjected to a uniformly dis- 260 tributed boundary load C-12-6: Circular cylindrical cantilever shell subjected to a trans- 264 verse load at the end ©-12-7; Skew hyperbolical paraboloid (hypar shell) subjected to 267 deadweight G-l3-I: Water tank with variable wall thickness under liquid 212 Pressure C-13-2: Cylindrical pressure tube with a shrinked ring 276 Pressure boiler 281 13-4: Circular cylindrical shell horizontally clamped at both 283 ends subjected to deadweight C-13-6: Buckling of a cylindrical shell under external pressure 288 C-13-6: Free vibrations of a circular cylindrical shell 290 C-14-L Spherical cap under a concentrated force at the vertex 293 14-2: Eigenfrequencies of a hypar shell 296 D_ Structural optimization 301 — Chapter 15 to 18 — D.1 Definitions — Formulas — Concepts 301 15 Fundamentals of structural optimization 301 15.1 Motivation — aim — development 301 15.2 Single problems in a design procedure 302 15.3 Design variables — constraints — objective function 303 15.4 Problem formulation — task of structural optimization 306 155 Definitions in mathematical optimization 307 186 Treatment of a Structural Optimization Problem (SOP) 309 16 Algorithms of Mathematical Programming (MP) 310 16.1 Problems without constraints 310 16.2 Problems with constraints 314 17 Sensitivity analysis of structures 321 171 Purpose of sensitivity analysis 321 12 Overall Finite Difference (OFD) sensitivity analysis 322 173 Analytical and semi-analytical sensitivity analyses 322XI Contents 18 Optimization strategies 18.1 Vector, multiobjective or multicriteria optimization — PARETO-optimality 182 Shape optimization 18.3 Augmented optimization loop by additional strategies D.2 Exercises D-15/16-1: Exact and approximate solution of an unconstrain- ed optimization problem D-15/16-2: Optimum design of a plane truss structure — sizing problem D-15/16-3: Optimum design of a part of a long circular cylin- drical boiler with a ring stiffener — sizing problem D-18-1: Mathematical treatment of a Vector Optimization Problem D-18-2: Simply supported column — shape optimization pro- blem by means of calculus of variations D18-3: Optimal design of a conveyor belt drum - use of shape functions D-18-4 Optimal shape design of a satellite tank - treat- ment as a multicriteria optimization problem. D-18-5: Optimal layout of a point-supported sandwich pa- nel made of CFRP-material - geometry optimiza- tion References A Fundamentals of elasticity B Plane load-bearing structures Cc Curved load-bearing structures D Structural optimization Subject index 325 325 329 334 337 342 347 355 360 364 370 315 375 376 3m7 378List of symbols Note: The following list is restricted to the most important subscripts, notations and letters in the Look. Scalar quantities are printed in roman letters, vectors in boldface, tensors or matrices in capital letters and in boldface. 1. Indices and notations ‘The classification is limited to the most important indices and notations. Further terms are given in the text and in corresponding literature, respectively. iis... @, Buy... ot ANB ACB latin indices valid for 1,2,3 greek indices valid for 1,2 Index for a layer of a laminate subscripts for covariant components superscripts for contravariant components indices in brackets denote no summation prime after index denotes rotated coordinate system eg. 6,4, comma denotes partial differentiation with respect to the quantity appearing after the comma, eg. u,x superscript prime before symbol denotes deviator, eg. vertical line after a symbol denotes covariant derive ting to curvilinear coordinates #, eg, vil bar over a symbol denotes virtual value, eg. F, roof over a symbol denotes the reference to a deformed body tilde denotes approximation asterisk right hand of a small letter denotes physical compo- nent of a tensor, e.g. a¥ asterisk right hand of a letter denotes extremal point asterisk right hapd of a capital letter denotes the cormple- mentary of work or energy, eg. U* nabla—operator differential operator intersection of A and B A is a subset of B for allXIV_List of symbols 2. Latin letters a a By determinant of a surface tensor radius of a spherical or a circular cylindrical shell co- and contravariant base vector of a surface in arbitrary coordinates normal unit vector to a surface co- and contravariant components of a surface tensor semiaxis of an ellipse determinant of the covariant curvature tensor co-, contravariant and mixed curvature tensor volume dilatation orthonormalized base ( Cartesian coordinates ) permutation symbol volume force vector objective function, ~ vector weight per area unit determinant of the metric tensor imequality constraint function, — vector co— and contravariant base (arbitrary coordinates ) co— and contravariant metric components, metric tensor height of a boiler equality constraint function, - vector core height, distance of the k. layer from the mid-plane buckling value shell parameter normal unit vector parabola parameter vector of external loads ; vector of control polygon points pseudo-load matrix preference function circumferential and normal loads external loads of a cylindrical shell distance perpendicular to axis of rotation radius of curvature distance to axis of rotation along the curvature radius load vector position vector to an arbitrary point of the mid-surface or a bodyList of symbols XV a A A B, orthogonal vectors coordinate in meridional direction vector of search direction wall thickness, layer thickness (k stress vector components of a stress vector state vector of a cylindrical shell, state variable vector ional and in circumferential direction displacements in me displacement vector displacements tangential to the mid-surface displacement perpendicular to the mid-surface weighting factors, penalty terms approximation for deflection design variable vector Cartesian coordinate system, EUCLIDIAN space shape parameter transformed variables complex variable state vector at point i of a shell of revolution area, surface; concentrated force at a corner strain-stiffness matrix; matrix of A—conjugate directions B-spline base functions, BERNSTEIN- polynomials rotation matrix; coupled stiffness matrix transfer matrix of a shell clement, total transfer matrix elasticity tensor of fourth order elasticity matrix flexibility matrix tension stiffness of an isotropic shell strain- or shear stiffnesses of an orthotropic shell flexibility tensor YOUNG's modulus, elasticity matrix plane elasticity tensor objective functionals concentrated forces; load vector implicit representation of a surface symmetrical flexibility matrix — mixed transformation tensor — system matrixXVI List of symbols IpIL Roa mom 1K yg Ky 1 Kyy Kl Ky H K x Nags Nay N, hex zzz & 2 & WARP Pm OO » shear modulus penalty function operator of inequality constraints elasticity tensor of a shell mean curvature operator of equality constraints HESSIAN matrices integral function invariants unity matrix, — tensor JACOBIAN matrix compression modulus bending stiffness of an isotropic shell GAUSSIAN curvature bending and torsional stiffness of an orthotropic cylindrical shell stiffnesses of an orthotropic plate bending stiffness matrix differential operators LAGRANGE — function boundary moment moment tensor bending and torsional moments pseudo-resultant moment tensor membrane force tensor normal and shear components of membrane forces effective inplane shear force polynomials transverse shear forces effective transverse shear force boundary force penalty parameter polynomial approximations radii of principal curvatures shape function of a shell surface n-dimensional set of real numbers stress tensorList of symbols XVIT HAA qa rr xs q e 2 g 3 8 £ z B i 3 pi wy ay aa Tye Yon Yop Ya Ys5* To0 8 88 85 ee sik Bigg fap Eg yy Fae } Fea Egy £66 "e c shear stiffness matrix tensor of n-th order (n = 0,1,2,3,4...) CHEBYCHEY polynomials or functions transformation matrix specific deformation energy specific complementary energy potential for field of conservative forces volume tensor of deformation derivatives, deformation gradient strain tensor (symmetrical part of V) tensor of infinitesimal rotations (antisyrnmetrical part of V) weight external work, complementary work feasible design space, subset semi-angle of a cone optimal step length coefficient of thermal expansion strains of the mid-surface of a shell LAGRANGE multipliers distortions of the mid-surface of a shell componente of a rotation matrix: thermal-elastic tensor strain tensor shear strains in Cartesian coordinates strains, distortions shear deformation shear strain variational symbol KRONECKER's tensor in curvilinear and Cartesian coordi- nates MAXWELL's inuence coeffients factor of the step length, strain vector permutation tensors strains in Cartesian and spherical coordinates vector of free thermal strains coordinate perpendicular to mid-surfaceXVIMT_ List of symbols MN 9 Ky Ky Ky ky xx 1 Oyy 1 One 1 On Sun 1, .0, mig, ni slack variable coordinate in latitude direction, latitude angle decay factors principal curvatures, variable exponents of an ellipse function tensor of curvatures eigenvalue vector of auxiliary variables LAME constants decay factor for a conical shell POISSON's ratio curvilinear coordinate system, GAUSSIAN surface parameters mass density tensor of curvatures (shallow shell ) normal stresses in Cartesian coordinates principal stresses stress vector mean value of norma] stresses time stress tensor shear stresses in Cartesian coordinates coordinate in meridional direction , meridional angle physical components of the bending angles of a shell of revolution LOVE function slope of cross~sections, bending angle eigenfrequency, eigenfrequency parameter coordinates of a spherical shell (starting from the bound- aries ) GREEN-LAGRANGE's strain tensor CHRISTOFFEL symbols of first and second kind LAPLACE-operator, modified LAPLACE operator temperature distribution thermal elastic tensor external, internal potential external, internal complementary potential total potential, total complementary potential AIRY's stress function modified function, penalty function1 Introduction ‘The classical fundamentals of modern Structural Mechanics have been founded by two scientists. In his work ” Discorsi”, Galileo GALILEI (1564 — 1642) carried out the first systematic investigations into the fracture pro- cess of brittle solids. Besides that, he also described the influence of the shape of a solid (hollow solids, bones, blades of grass) on its stiffness, and thus successfully treated the problem of the Theory of Solids with Uni- form Strength. One century later, Robert HOOKE (1635 - 1703) stated the fundamental law of the linear theory of elasticity by claiming that strain (alteration of length) and stress (load) are proportional ("ut tensio sic vis”). On the basis of this material law for the Theory of Elasticity, Edme MARIOTTE (1620 — 1684), Gottfried Wilhelm von LEIBNIZ (1646 - 1716), Jakob BERNOULLI (1654 - 1705), Leonard EULER (1707 — 1783), Charles Augustin COULOMB (1736 - 1806) and others treated special pro- blems of bending of beams. Until the beginning of the 19th century, the Theory of Beams had almost exclusively been the focus of the Theory of Elasticity and Strength. Claude - Louis - Marie - Henri NAVIER (1785 - 1836) developed the general equations of elasticity from the equilibrium of a solid element, and thus augmented the beam theory. Finally, he also set up torsion theory of the beam. Hence, he may quite justly be seen as the actual founder of the Theory of Elasticity. NAVIER's disciple Barré de DE SAINT-VENANT (1797 - 1886) augmented the work of his teacher by contributing new the- ories on the impact of elastic solids. His contemporary, the outstanding scientist and engineer Gustav Robert KIRCHHOFF (1824 — 1887), derived with scientific strictness the plate theory named after him. The first math- ematical treatments of shell structures were contributed by mathematicians and experts in the theory of elasticity as Carl Friedrich GAUSS (1777 - 1855), CASTIGLIANO (1847 ~ 1884), MOHR (1835 - 1918), Augustin Louis Baron CAUCHY (1789 - 1857), LAME (1795 - 1870), BOUSSINESQ (1842 - 1929), and, as mentioned above, NAVIER, DE SAINT-VENANT and KIRCHHOFF. A complete bending* theory of shells was derived systemati- cally by Augustus Edward Hough LOVE (1863 - 1940) on the basis of a publication by ARON in 1847. During the 19th century, numerous works have been published in the field of Structural Mechanics which cannot be described in detail here. However, based on the above-said one might assume that this discipline is an old one, the problems of which have largely been solved. As a matter of fact,2 1_Introduction this surmise may have been true until recently. However, the continuous de- velopment of the sciences and the technology, especially during recent years, calls for au increased exactuess of computations, in particular in the construction of complex systems and plants and in lightweight construc- tions, respectively. Owing to the introduction of duraluminium and other advanced materials like composites, ceramics, etc. into the lightweight con- structions, the number of publications in the field of shell and lightweight structures has witnessed a substantial increase. In [C.6] it is shown that the amount of publications has doubled per cach decade since 1900. Proceeding from about 100 papers in the year 1950, one counted about 1000 publications in 1982, ie. three per day. Thus, the references to this book can only com- prise a very limited selection of textbooks and publications. The still continuing importance of Structural Mechanics also stems from the fact that the relevance of structures that are optimal with respect to bearing capacity, reliability, accuracy, costs, etc., is becoming much more apparent than in former times. Especially in the field of structural optimi- zation, considerable progress has been achieved during recent years and this has prompted increased research efforts in underlying branches of solid mechanics like fracture and damage mechanics, viscoelasticity theory, pla~ stomechanics, mechanics of advanced materials, contact mechanics, and sta- bility theory. Here, the application of computers and of increasingly refined algorithms allows treatment of more and more complex systems. In this determination of an initial design intuitive modification of structural parameters mathematical structural optimization procedures MSOP. ‘ificationS fullfilled? objectives met? yes integration of mathematical structural optimization procedures (MSOP ) into the design process1_Introduction 3 context, one should mention the large amount of novel finite computation procedures (eg. Finite Element Methods [A.1, C.25]) as well as the Algo- rithms of Mathematical Programming applied in structural mechanics. One can thus justly claim that all of the above-named more novel fields and their solution approaches are all based on the fundamentals of elasticity without which the currently occurring problems cannot be solved and evalu- ated. The field of Structural Optimization increasingly moves away from the stage of a mere trial-and-error procedure to enter into the very des- ign process using mathematical algorithms (Fig. 11). This development roots back to the 17th century, and is closely connected with the name Gottfried Wilhelm LEIBNIZ (1646 - 1716) as one of the last universal scho- lars of modern times. His works in the fields of mathematics and natural sciences may be seen as the foundation of analytical working, ie. of a cohe- rent thinking that is a decisive assumption of structural optimization. LEIBNIZ provided the basis of the differential calculus, and he also inven- ted the first mechanical computer. Without his achievements, modern opti- mization calculations would yet not have been possible on a large scale. Here, one must also name one of the greatest scientists Leonard EULER (1707 - 1783) who extended the determination of extremal values of given functions to practical examples, The search for the extremal value of a function soon led to the development of the variational calculue where entire functions can become extremal. Hence, Jakob BERNOULLI (1655 — 1705) determined the curve of the shortest falling time (Brachistochrone), and Issac NEWTON (1643 - 1727) found the solid body of revolution with the smallest resistance. Jean Louis LAGRANGE (1736 — 1813) and Sir Wil- liam Rowan HAMILTON (1805 - 1865) set up the principle of the smallest action and formulated an integral principle, and thus contributed to the perfection of the variational calculus that still is the basis of several types of optimization problems. Many publications on engineering applications over the previous decades utilize the variational principle. LAGRANGE, CLAUSEN and DE SAINT-VENANT had already treated the optimal shape of one-dimensional beam structures subjected to different load conditions. Typical examples here are the huckling af a column, as well as the can lever beam for which optimal cross-sections could be found using the vari- ational principle. This requires the derivation of optimality criteria as ne- cessary conditions; these are EULER ’‘s equations in the case of unconstrai- ned problems. If constraints are considered, as, eg, in solution of an iso- perimetrical problem, LAGRANGE’s multiplier method is used.A Fundamentals of elasticity A.l Definitions — Formulas — Concepts 2 Tensor algebra and analysis 21 ‘Terminology — definitions The use of the index notation is advantageous because it normally makes it possible to write in a very compact form mathematical formulas or systems of equations for physical or geometric quantities, which would otherwise contain a large number of terms. Coordinate transformations constitute the basis for the general concept of tensors which applies to arbitrary coordinate systems. The reason for the use of tensors lies in the remarkable fact that the validity of a tensor equa- tion is independent of the particular choice of coordinate system. In the fol- lowing we confine our considerations to quantities of the three —dimen- sional EUCLIDEAN space. We introduce the following definitions : A acalar characterized by one component (eg. temperature, volume) is called a tensor of zeroth order. A vector characterized by three components (eg. force, velocity ) is called a tensor of first order. The dyadic product of two vectors, called a dyad (eg. strain, stress), is a tensor of second order characterized by nine components. Tensors of higher order appear as well. Notation of tensors of first order: 2 a) Symbolic in matrix notation: a =| a]. 3 a b) Analytical: a=a,e, + ae, + a6, 3 or a=aeteetae= > ee a with e,, ey, €, as base vectors in a Cartesian coordinate system. The sub- scripts are indices, and not exponents. In index notation the expression a (or a, ) (i = 1,2,3) denotes the total vector (see Fig. 2.1). Notation of tensors of second order: ty fiz ts | a) Symbolic in matrix notation: T -| tor tee tes ts, tsz tag6 2 Tensor algebra and analysis Fig. 2.1: Definition of the base vectors b) Analytical: Tatcettyee +... +tsee = igi or T= 22he e, where cle! is the dyadic product of the base vectors. In index notation the expression t, denotes the total tensor. 2.2 Index rules and summation convention (i) Index rule If a letter index appears one and only one time in each term of | an expression, the expression is valid for each of the actual va- | lues, the letter index can take. Such an index is called a free indea. EINSTEIN ‘s summation convention Whenever a letter index appears twice within the same term, as subscript and/or as superscript, 2 summation is implied over the range of this index, ic, from 1 to 3 in the three ~dimen- sional space (Latin indices used), and from 1 to 2 in the two- dimensional space (Greek indices used ), Such an index is called dummy. (i) Mazimum rule Any letter index may never be applied more than twice in each term. Examples of (i): al + o2b'=0, asapha0 <> vobrao, a+ ap=0.2.2 Index rules and convention 7 =T, <=> ty Note: Comma implies partial differentiation with respect to the coord- nate(s) of succeeding indices. The rules (i) - (#i) apply for these in- dices as well. Bramples of (ii): a =ae= ale, ta7e,+a°e,, — three~dimensional space , e+ ase’, two - dimensional space (surface) , Tatjed=telet tee t.ttgee, a=ae = 12,38 i hr+ hth, _ i of 1 of 2 of. 3 df = fydxt = 26 ax! + OE dx! + 28 dx’. Attention: As it is of no importance which notation 2 doubly appearing index possesses, this 30 called duramy index can be arbitrarily renamed: =ae =ae=atk asae sae ay Exception: No summation over paranthesized indices, ie. fei) _, at=a, be etc i): Following expressions are meaningless: Examples of (i gt =0 , bicose,=1 The following expressions are also meaningless, as the free indices have to be the same in each term: ht be =o , Ag, = BZ 2.3. Tensor of first order (vector) Base vectors (Fig. 2.2) e, = orthogonal base with the unit vectors e,, e,, e;, 8, = base in arbitrary coordinates with the base vectors g,, £), By Measure or metric components GB = Bj = Gj BB (2.10)8 2. Tensor algebra and analysis Fig. 2.2: Orthonormalized aud, arbitrary base xt Metric tensor Bn Bi Bs (g;) =| ° S22 &s | —* Due to (2a) the metric (2.10) f tensor is symmetrical. 33 (2.1c) Determinant of the metric tensor —_ det (g:) (22a) (226) Angle y between vectors x and y cose BT - xl-lyl Venn ¥" ¥" Ya Covariant and contravariant base An arbitrary base g;(i = 1,2,3) is given in the three~ dimensional EUCLIDEAN space. We are searching a second base g’ so that the following relation exists between the base vectors B84, (230) (22c) where KRONECKER's delta is defined by 6 1 for i=j, (238) 2 Yo for i Fj.23 Tensor of first order 9 If the base g; is known, the base gi can be determined by means of the nine equations (2.8). The base g; is called the covariant base and gi the contra- variant base. Covariant metric components Bj = 8B = Bi- (24a) Contravariant metric components gag gi=g". (246) Rule of exchanging indices g =e'g . (25a) 6 = 88> (2.56) 4 =e e- (25¢) Other determination of the contravariant base vectors &2* Bs [81 +8285 | where [1,62 183] is the scalar triple product of the three covariant base vectors 8}, 2, &3 63 * Br. &.* 2 — = 26, [Teese] © 7 [eee]? & 18 Transformation behaviour A fundamental (defining) property of a tensor is its behaviour in connec- tion with a coordinate transformation. In order to investigate this transfor- ion behaviour, the following task shall be considered: itial base g, or gi (i = 1,2,3) is given together with a "new” base g,, An or gi (i' = 1,2,3) generated by an arbitrary linear transformation with prescribed transformation coefficients (],. Additionally, a vector be given in the initial base by its components v! or v;. Its components v;- or v'’ shall now be determined in the "new” base. Rules of transformation ‘Transformations of bases & =hig . 6 HBB, (270) k iv & =F.& . B= Bie (2%) The following relations are valid Ab = we > BY BY = (28)10 2 Tensor algebra and analysis Transformations of tensors of first order (25a) (2.9) (2.10) 2.4 Tensors of second and higher order Definitions: ‘Two vectors x and y are given in the EUCLIDEAN space. With that we are forming the new product T=xy. (2.11) ‘The notation without dot or cross shall indicate that it is neither a scalar product nor a vector product. Depending on whether the covariant base vectors g, or the contravariant base vectoro g! are applied here, one obtains four kinds of descriptions for a tensor of order two. T-Uge-tige tes, -tee’- (2.12) According to the position of the indices one denotes tj as covariant: components , wv as contravariant components , '; as mixed contravariant-covariant components , t? as mixed covariant-contravariant components of the tensor 'T . Formal generalization of tensors tensor of zeroth order (scalar) 3° = 1 base element, & tensor of first order (vector) 3' = 3 base elements, 7 = gg. tensor of second order (dyad) 3” = 9 base elements, 818) &, tensor of third order 3° = 27 base elements, a lg, 8 8, 8; tensor of fourth order 3* = 81 base elements.2.4 Tensors of second and higher order ety Transformation rules For a transformation of a vector base g, into 2 new vector base g,, equa- tions (2.7) and (2.76) are used: B= Oey and g, = Bg). The tensor T can be given either in the old base g, or in the new base g,, — er Pp Tat gg. = g.g,- (213) The transformation formulas read as follows ape! or ee = pi gi, (214) (215) From T= tye'g’ follows t= BBY tye oF thy = BY Bh ty (216) In a similar way one obtains the transformation formulas of the mixed components of the tensor T. Note: It is worth mentioning that tensors are actually defined by the rules by which their components transform due to coordinate transforma- tions. Thus, any quantity T with 3? = 9 components is then and only then a second order tensor if its components transform according to (2.14) or (2.16) in connection with an arbitrary coordinate transfor- mation, Physical components of a tensor of second order The physical components for orthogonal coordinate systems can be deter- mined as follows (for non - orthogonal coordinate systems see [A8]): (2.17) Symmetrical and antisymmetrical tensors of second order Any tensor of second order can elways be presented as a sum of a symime- trical and an antisymmetrical (or skew - symmetrical ) tensor: ay Du (2.182) a ON Legis i with PUL y ty, (2186) aij ij ji PH = Ei y (2.18c)12___2_Tensor algebra and analysis Permutation tensor or e- tensor As permutation tensor a tensor of third order is defined f= /bem » oh =tel™ (219) with the permutation symbol -1 » {i,j,k} anticyelic , (2.20a) { +1 for fiat cyclic , 0» {i,j,k} acyclic Permutation symbol in two dimensions i (2.206) e,=-1, Vector product as application of the e~ tensor k tom XX Y = Fig YB (221) Eigenvalues and eigenvectors of a symmetrical tensor ~ Principal azis transformation Lemma: | For any symmetrical, real valued, three—column matrix ‘T there always exist three mutually orthogonal principal directions (eigenvectors) a and three corresponding real eigenvalues » (which not necessarily have to be different from each other). These eigenvectors and eigenvalues are governed by the follow- ing algebraic eigenvalue problem, where I is the unit tensor: (TB-AL)a=o0 or (t}-A6j)a, =o. (2.220) Determination of the eigenvalues: 1 1 1 t,- A ty ty det(h-asiy=| tf B-a 8 [=o (£22) “ th th-a Characteristic equation of (2.22): N-LN+LA-1, =0. (228c) The roots A = 4, Ay and Any of this cubic equation are invariant with respect to transformations of coordinates. Substituting sequentially these cigenvalues into (£220) and solving for a, we obtain ay, ay and ayq-25 Curvilinear coordinates 13 The quantities 1,, Ip, Is in (2.28c) are invariants defined by | A.8]: L=%, (2.280) =dcid- gy), (2256) I, = det (th). (2.28) 2.5 Curvilinear coordinates Base vectors - metric tensor In the three-dimensional space a vector r can be presented in Cartesian coordinates x and in curvilinear coordinates indicated by é'(i = 1,2,3) (Figs. 2.3 and 2.4) e g Fig. 2.3: Position vector in Fig. 2.4: Curvilinear coordinates orthonormalized base and base vectors Position vector r of a pomt P r=r(x) | r=r(é) (224) Base vectors or =°ra,. 2Fer,. 2.25, eB | san (225) Relation between base g,(¢') and orthonormalized base e, i ox 6 = ks (226) Length of a line element ds’ = dr - dr —+ First fundamental form of a surface. (22%) Indicating the derivative with respect to the curve parameter t by a dot, the length of the curve between t, and t, is given by: ty s= fvasF Bae (22%) You 2 Tensor algebra and analysis Volume element av = fede alae (228) Partial base derivatives - CHRISTOFFEL symbols &a = rig, , (22%) ex =-Tus- (229) CHRISTOFFEL symbols of the first kind 1 Ti = y (Gini + 8hi,5 ~ Sia) > (2.30) CHRISTOFFEL symbols of the second kind rp =a Ly. (231) Rule: {The CHRISTOFFEL symbols can be expressed alone by the | metric tensor and its derivatives. / Note: The CHRISTOFFEL symbols do not have tensor character. For the CHRISTOFFEL symbols of the first kind (2.50) the following rela- tions hold: 1) Tye interchangeability (£380) of the first two indices , ° 2) Dang + Pinan =e! interchangeability (2.32) 28 of the last two indices . For the CHRISTOFFEL symbols of the second kind, the following relations are derived from (2.31) using (2.50): i 1) v5 ji interchangeability of subscripts (symmetry), (2.32c) ok, 2) riod ge 28m _ oltn/g ) (252d) HT 2° og Ya Covariant derivatives ‘Tensor of first order ay =ale (233) with al, =ajt hat. (24a) By anology al) =a, ,~ Thay. (281b)25 Curvilinear coordinates 15 Tensor of second order m m ithe = Apa 7 Pie Ami ~ Pie tim + (23a) + rigava rial. (2.356) Gradient of a scalar funktion & v= grad = VG = 5 |, (2560) Gradient of a vector v Grdv = Vv =vheg, (2.56) Divergence of a vector v 12 (fe v’). (2.570) divv =V-v= = Ye 0 Divergence of a tensor T of second order DwE=V- T=t" lg, (25%) Rotation of a vector v (2.38) LAPLACE operator AG = V'S = divgradS = (fee ,);- (239) Bipotential upezator AAG = VG = V'(V'S) = Bl} = 1 1 xd = og liee| (vee Palade GAUSSIAN theorem I]. VE ae ae? ae? = ffv's, dA. (241) (240)16. 2 Tensor algebra and analysis Bxample: Application of the previous formulas to cylindrical coordinates Single-valued relations between Cartesian coordinates x! and cylindrical coordinates | read as follows (see Fig. 2): tates? , x =ésinf . x Position vector r(€) =@cosfe, + fsinge, + Fe, (242) Covariant base vectors according to (2.25) (242a) g, = cose, + sinfe,, g =-fsinge, +fcsle,, (243) Covariant metric components according to (24a) 8 = Bi 8}. For example’ g,,=g,° 8, =(¢) sim? + (Pas? =(¢). Covariant metric tensor 1 0 0 (e)=[0 (ey of. (244) 0 0 1 According to (25c), because of g;=g,- 8, =0 fori Fj ——+ Orthogonal base Contravariant componente from (25¢) Gi) _ 1 5 Bi) >) &iiy =1 Fig. 2.5: Presentation of a position vector in cylindxical coordinates2.5 Curvilinear coordinates a7 Contravariant metric tensor 1 0 0 (s")=]0 (€y" 0 (245) 0 0 1 Determinant g of the covariant metric tensor 1 0 0 ls|=lej)-|o (8 of=cey. (246) 0 0 1 CHRISTOFFEL symbols of the first kind according to (2.90) 1 1 For example: Tag) — 7 (8a,2 + 82,2 82a) ~ yO + 0-28) = CHRISTOFFEL symbols in matrix notation 0 0 0 (Py)=] 0 -€ of, 0 Oo 0 ra) oe ¢ (Mp=]@ 0 of, Oo 0 0 (Ty) = CHRISTOFFEL symbols of the second kind according to (2.1) For example: Ty, = ¢ "Py, =1-(-@)+0-040-0= CHRISTOFFEL symbols in matrix notation o 0 0 (T)=]e -€ of, 0 0 0 (248) 0 (éy* 0 (m=[(2yt oo of, 0 0 018 3 State of stress LAPLACE operator according to (2.39) 1 ik 1 in, 52 i3 _ 6 = + (fee ,),- | fale, + 8°, +8°6,)/ = Fe lRO tay Te [Rag #08, +089), 1 =F (Vee dat YEE sat (fee"#,.)a|= RCO. + ba)o+ CFO] = mall en 4O,4 gen +€o.,5| — AG= 84 58+ Cay tint Ea (849) 3 State of stress 3.1 Stress vector ‘The essential objective of structural analysis ie the calculation of stresses and deformations of bodies. As shown in Fig. 3.1 we make a cut through the body, which is in equilibrium under external loads in the form of volume forces f,, surface tractions p, and concentrated forces F,. A resulting force AF is transmitted at every clement AA of the cut. Fig. 3.1: Cut through a body Fig. 3.2: Resolution of the stress vector31_ Stress vector 19 According to NEWTON's principle of "actio = reactio”, reaction of the resulting force AF is found on the same plane of the opposite part of the body, in the form of an opposite directed force of the same magnitude. We assume that the relation AF/A A in the limit of AA —> 0 tends to a finite value, and we call this limiting value AF _ dF oAA~ dA* (34) stress vector t= lim BA Here it is assumed that only forces (and no moments) are transmitted at any point of the cut. The stress vector t can be resolved into a part perpendicular to the surface of the cut, the so-called normal stress of the value c, and into a part tan- gential to the surface, the shear stress of the value 7 ( Fig. 3.2). Stresces on cut planes with outward normale pointing in the positive (respective negative) coordinate direc- tions, are taken positive in the positive (respective negative ) coordinate directions (Fig. 33). Sign convention: Stress vectors on the positive cut planes of the cubic element in Cartesian coordinates: xx y Tyx =x me > =| on] > a (38) Tay Tye Cun In this context, o (i = x,y,) are normel stresses and 74 (i,j = =, 9¥52) are shear stresses, "positive" "negative" cut planes cut planes a e a. stresses stresses Fig, 3.3: Sign convention for the stresses20 3 State of stress 3.2 Stress tensor The stress vectors can be assembled in matrix notation as the so-called stress tensor S, In Cartesian coordinates it reads xx | Tox | Tax s Tay | yy | Tay (33) Ta! Ty 1 Oe ‘The superscript T indicates transpose of a matrix. ‘The important CAUCHY's formula in arbitrary coordinates is written Sn (34a) or (34b) In words: This formula gives the stress vector t at a given surface or cut plane in terms of the stress tensor S and the unit outward nor mial vector n for the surface or cut, The stress vector t acts on the infinitesimal area dA of the inclined cut plane characterized by the unit. ontward normal vector n (Fig. 24): ny] [cosa 7 nyJ=|cos6} , t=/ ty]. (85a,b) n, cosy t ‘The remaining infinitesimal surfaces of the tetrahedron result from the pro- jection of dA which can be written as follows in index notation with x = x', 22 = x3 y=x,2=x dA, = dAcosa = dAn, dA, = dAcosf=dAn, | dA, =dAn,. (36) dA, — dAcos) — dAn, Fig. 8.4: Stress vector ¢ at a tetrahedron in Cartesian coordinates3.2_Coordinate transformation — principal axes a Relationships between Cartesian and other coordinates will be given later. In Cartesian coordinates there is no difference between a covariant and a contravariant base, and for this reason the indices can always be lowered. Component notation in Cartesian coordinates ty = yg My + Ty By + Tax My ty = Tay By + yy By + Try By (37) = TaN + Tygy + Oy, - Nete: Shear stresses are pairwise equal to one another, i.e, the stress tensor Sis symmetric —> (38) Tay = Tx 8 Te = Tey Tax The symmetry reflects satisfaction of moment equilibrium conditions. 3.3 Coordinate transformation — principal axes We consider a Cartesian coordinate system x! and a rotated system x’ (see Fig. 3.5). Stresses in a rotated system according to (2.14) i pi =f, Bit - Symbolic notation S'=B-S-B (39) 7 Fig. 3.5: Stresses in a rotated coordinate system22 3. State of stress Arrangement of transformation coefficients in a rotation matrix B AO B=|A BB |. (340) Am Principal stresses, principal axes Principal stresses (see (222) and (223)) (4-08) nt =0 (Sit) Characteristic equation o- Lo +ho- (218) with the invariants for any direction and for the principal stresses 9, (i=1,0,m): Ty = Og, + yy +04, = 0 + Oy + Oy sum of normal stresses, (3.180) xx Tay ny Tye oz Tax | xy Oy ye Cee ax Oxx (3.13b) Oxx Tay Tae Tay Sy Tye | = ON (3.18c) Te Tye oF Note: It can be shown that the three roots of (3.12) comprise the maximum and the minimum normal stress appearing on all possible cut planes through a given point. That is where the name principal stresses is coming from. For the symmetrical stress tensor the principal stresses are always teal. ‘The directions of principal stresses of different mag- nitudes are always unique and mutually orthogonal. State of plane stress in Cartesian coordinates Definition: — ¢, Ty = 0. (314) co ye Stress tensor s-[ "= ml: (3.15)3.3. Coordinate transformation - principal axes 23 Fig. 3.6: Coordinate transformation Transformation coefficients according to Fig. 36 AY = cosa , By = c00(F ~a) = sine, (316) , . . , =cos(F+a)=-sina , fh; = cosa. Formulas of transformation for any rotation a of the coordinate system x 7 yy COS ZO + Ty SIN ZA, 1 . (04, ~ oy eos 2a ~ 7, 8in2a, (317) Principal stresses 7 1 xx ~ yy YP 2 3. I =F Cat Oy dey (CMG) aay - (318) The directions of the principal stresses follow from the extremal condition to be (319) and from this the principal directions 2a* and 2a* + 7 or a* and a* + $ The principal directions are orthogonal to each other. Maximum shear stress (320) Direction at = at +7.24 3_ State of stress Fig. 3.7: MOHR’s stress circle MOHR's circle The formulas for transformation of the plane state of stress lead to the MOUR's circle (Oy: 20 Tye 27) xx + 4%, ” 2 2 2 (o - "x ) +7 =(0-0,) +7 (321) with the distance of the centre M on the c-axis 1 0 = Fax + yy) (3220) and the radius of the circle r= (322b) 3.4 Stress devialor Definition: "T= 7) Oy 55 (323) with the mean normal stress 044 1 1 Om = 3 xe Oy + Oe) = FCO + Oy + Oy) = (3.24) Physical interpretation : The stress deviator '7} expresses the deviation of the state of stress from the mean normal stress. Since "I = 0, the principal values of the stress deviator follow in analogy to (342) from 'o + 'I,'o "1, =0 (325) ap 2 with "lh = 1, - 30%, 3 } (3.26) =1,- Loy + 20m -3.5 Equilibrium conditions 25 3.5 Equilibrium conditions The conditions of force equilibrium are stated with regard to the unde- formed configuration of the body in this section [ A.LL, A.15, A.16, A.l/]. at, + dz dz ye 2, + ody oy + 20x. dx Fig. 3.8: Equilibrium for an infinitesimal volume element in Cartesian coordinates 1) Cartesian coordinates (Fig. 3.8) 09, oT, eT, ae my Ta yg = oe ay t +f, oT, oo, a7, Sy a ve + By +a +h (32%) or, eT, dg, 2 tn a Ix t +aR+ f, f, (i = x,y,z) are the components of the vector of volume forces. Abbreviated notation ieco, (32%) 2) Curvilinear coordinates P+ to (3.260) or DivS+f=0. (3.28) 3) Cylindrical coordinates e 21,029,622) (Fig 25) 0, 127, 1 ar trap r oe) +f, = 0, ar, 1 8a, er, 2 OTe 5 120 4 2Tae 4 2 of roe te titathy (3292) OTs or, 2%, 1 1 ote y 1s le 1 _ or tr oy ton tem th=26 4 State of strain Fig, 3.9; Spherical coordinates 2.w Two-dimensional case: Polar coordinates 1, ¢ ° (3.296) OTe, 125 2 wr 12% 2, =o Sr tr op Tria t 4) Spherical coordinates r, 9, y (Fig. 29) 2% 1 OT 1 ot or +r oo Boye Ty WHI + L 7 2a ~ 99 ~ Foo) + A, = at 129, 1 OTe. 3 wy d Lote 4 3 8.80, or tr ed trand op tr "wt (880) 1 + COLD (O55 — ay) + fy = 05 a7, 1 87, 1 ey, 3 2 or tr ad tran’ do tr tT octet 4 State of strain 4.1 Kinematics of a deformable body Description of the deformation of a body with LAGRANGE's notation: The displacement of a material point of a body B is observed as a function of the initial state. ‘We distinguish between the initial state t = to (without ~) and the defor med state of the body t = t (with ~).41 Kinematics of a deformable body 27 Fig. 4.1: Kinematics of a deformable body Position vector # of the material point P of the deformed body B ( Fig. 4.1) F(€) = r(€) + v(E) (41) with the position vector r(é’) and the displacement vector v(é’) of the same material point P of the undeformed body B. Differential increase dv of the displacement vector v dv = df - dr = Var, (42) where V is the tensor of the displacement derivatives. According to (2.25) the base vectors g; and , result from the total differen- tial of the respective position vectors dr = Sia = nde! = gl, (48a) dF = cee =F dei = g, ae. (4.38) ‘These infinitesimal changes of the position vectors lead to the points Q and Q adjacent to P and P (see Fig. 411). In accordance with the rules for the transformation behaviour of tensors (Sections 2.3, 24) the base vectors of the deformed body can be expressed by those of the undeformed body and vice versa: & = Alg;, (44a) 6: = Fig,- (448)28 4 State of strain The displacement vector can be written as follows (45a) (4.5) 18) — Fk 8 = Fa - Since the base of the undeformed body or the base of the deformed body can be used alternately in order to illustrate the displacement vector v, we have two different covariant derivatives of the displacement components as a result (according to (4.5b)). Here, one line stands for the covariant deri- vative applied to the base of the undeformed body, a double line stands for the covariant derivative related to the base of the deformed body. With the KRONECKER cymbol we obtain the relation between the base vectors of the deformed body and the undeformed body [ 4.7]: &=(8 +h e, (4.60) & = (4 - o\)g,. (40) The elements of the transformation tensor then read pladsw, (4) (47) i = GF, Corresponding transformation relations are valid for the line elements dr and di in analogy with the base vectors. If we define a mixed tensor of or- der two according to (2.12) F=(+v)ee =I+Vv, (48a) F=(6-9))e@ -1-¥, (488) for the line elements this results in df = Far, (49a) dr = Fdé (4.9b) By means of (4.2) and due to (4.5a) we obtain the following total differen- tial of the displacement vector dv: dy = df ~ dr = (F-I)de = Var. (4.100) According to (4.2), V is called the tensor of the displacement derivatives or the deformation gradient.4.2. Strain tensor 29 Due to (4.10a), the total differential dv can be written in other notation by means of (2.866) dv — Gradv+ de. (4.106) In Cartesian coordinates the relation (4.106) for a time independent dis- placement vector u(x,y,2) v(x,y,2) =| v(x,y,2) w(x,¥52) reads as follows in matrix notation Qu au ou du ox ey oz dx = =| 2e 4m aw w/e]. ax Oy oa [@]. (4.10¢) awl | ow aw aw | Lda ox dy 2 4.2 Strain tensor The state of strain of an elastic body is obtained by subtraction of the squared line elements of a deformed and an undeformed body. Thus, we ob- tain a measure of how the distances of single points have changed due to a load [A.7, A.8]. We write dé « df - dr - dr = de? — ds? = (8, - & - - @) de de! = = (By ~ By) deh de! = 2a deiag?, (4.ta) where 74 are the components of the strain tensor. Accordingly, they can be determined as follows 1 Vg = (Gj ~ Bij) (4-116) Expressing the metric components of the deformed body by those of the undeformed body, we obtain the GREEN-LAGRANGE’s components of strain [ A.6] I). (4.120) ny Fr + vil. + vy Linearized components of strain by neglecting the quadratic terms in (4.1820): [a= 4 vd +40 | . (4.128)30 4 State of strain 4.3 Strain-displacement relations = Cartesian coordinates From the tensor of the displacement derivatives V follows as the symmetric part the linear strain tensor. According to the rules (2.18), it becomes in Cartesian coordinates: 1 1 Exx 9 Vey 2 Tan 1 T 1 =g(VtV )=| © ey ae fo (413) : en where due to (4.12b) ey =e = 24,2 OK? Yay = By FO? ov = 20, ow Wy? We = Sa 3K? (414) _ ow _ ov | ow fm ~ an? Wa on bay 44 (i,j = X,y,z) are the so-called technical shears, and ¢ (i = x,y,z) are the normal strains. Special cases: ~ Cylindrical coordinates ( Fig. 25) (1,0; 9,v; 2) or “ew (415) é ~ Spherical a. < “° (4.16)45 Compatibility conditions 31 4.4 Transformation of principal axis The principal strains are determined in analogy to the principal stresses. Characteristic equation according to (2.28c): MTN 4LA-l,=0. (4170) The first invariant corresponds to the so-called volume dilatation e: (41%) l,=e=divv= 4.5 Compatibility conditions The linear strain-displacement relations (4.12b) form a system of six coupled, partial differential equations for the three components v, of the displacements for given values of the strain tensor. Thus, the system is kinematically redundant. In order that there will exist a displacement vector v; subject to given values of the six mutually independent compo- nents of the strain tensor, it is necessary that the three components of the displacement vector satisfy the following compatibility conditions (DE SAINT VENANT ): tha + aki ~ Aales ~ tela = Yeh = 0. (4.18) Mechanical interpretation: The interior coherence of the body has to be preserved after the defor- mation, ie. material gaps or overlaps must not occur. For a two-dimensional state of stress or strain the compatibility condition in Cartesian coordinates reads as follows: (419) 5 Constitutive laws of linearly elastic bodies 5.1 Basic concepts In the following we are going to deal with bodies for which there exist re- versibly unique relations between the components of the stain tenser and the stress tensor, and we furthermore assume that these relations are time independent. The behaviour of the bodies is denoted as clastic, ie. there are no permanent strains ¢,; after removing the load of the body (Fig. 5.1). The bodies considered shall furthermore, as it is usual in the classical elasticity theory, be made of a linearly elastic material such that their constitutive law expresses linear relationship between the components of the stress ten- sor and the strain tensor (range 0- A ~ in Fig. 5.1). Such bodies are usually called HOOKEAN bodies.32 5 Constitutive laws of linearly elastic bodies stress, ,elastic Fig. 5.1: 0,¢- diagram of a real partially plastic material with a linear ~ elastic range A & limit of proportionality B 2 elastic limit C = upper yield point D = lower yield point D-E © elastic plastic state F 2 ultimate stress limit 0 — i : Oey strain & 2 plastic strain For a great number of problems in practice this assumption is feasible, even if we have to consider non-linear strain - displacement relations (eg. geo- metrical non - linearities for the poot— buckling of plates and chelle ). 5.2 Generalized HOOKE-DUHAMEL’s law for thermo— elastic, isotropic materials — Cartesian coordinates Op ¥(Oyy + Oy + OPO ey = Ey? wy (Suet yO ye = Eee (51) VO + H+ Or . WHS with EB YOUNG!s modulus , y POISSON's ratio , G E shear modulus 21t+v , op one~ dimensional thermal expansion coefficient , O=T,-T, difference between final and initial temperature Symbolic notation (52) with = sum of normal stresses52 Generalized HOOKE -DUHAMEL's law 33 Solving (51) with respect to stresses yields mm rr (4 E v wy rare + are) ~ = (53) Cam ey (tet a " Symbolic notation (54) with e — volume dilatation . ~ Curvilinear coordinates in index notation According to (5.1) it follows that l+uv vy ii a = SE Ta BAT] + BO = Dia + orb — (55a) with the flexibility tensor of fourth order lev y Din =" gE (8x8 + &n8x) ~ Bi ea - (55b) Solving (55a) with regard to strecees leads to 1a 5 Pacha, - ple (56a) with the elasticity tensor of fourth order oF ack) gle® 1 ay eet’) (560) and the thermo - elastic tensor of second order Ea I 0 TO BO =Be = 75,8 - (56c) Other notation of (56a): ‘ikl Th CON (ty ~ On BO) (8770) with CARTS agg a (ail 4 gl git) (57%)34 5 Constitutive laws of linearly elastic bodies and the LAME constants av 1~2v- The relations between the different specific elasticity constants can be drawn from the following table: d= u=Ge E= vs IESE) x roe > # “Ken | D+) dv Xd AQl_ 2v l1+v)(lL-2v)r v Qu v H(E - 2p E-2p az) HR) “ E i Ev Ev E E y (1+vj)(1-2v 2(1+v) Table 5.1: Elasticity properties The linearly elastic constitutive equations shall be augmented by another system of equations which allows a physical interpretation, and which is applied in elastoplastic structures. Therefore, we split both the strain and the stress tensor in a spherical-symmetrical and a deviatoric part accord- ing to the following relations : m 1 1, me pee tle m_ 1pm, am (58) TH! = 78h + HE - In (58) the known expressions for the sum of strains e = 9 or the sum of stresses s = 7, occur which are the first invariants of the strain tensor or the ctrece tensor according to (4.2%). Substituting (58) in the goncralised HOOKE's law leads to the following two equations (59a) m 1 m mek Pagar, 1 m=k — e=yoy+3a,0 (59) with k =-_® compression modulus 3(1-2v) ’ Oy = 5/3 mean value of the normal stresses , 3p volumetric thermal expansion coefficient