Draft: Finite Element I

Draft
DRAFT
Lecture Notes in:
FINITE ELEMENT I
Framed Structures
Victor E. Saouma
Dept. of Civil Environmental and Architectural Engineering University of Colorado, Boulder, CO 80309-0428
Draft
Contents
1 INTRODUCTION 1.1 Why Matrix Structural Analysis? 1.2 Overview of Structural Analysis . 1.3 Structural Idealization . . . . . . 1.3.1 Structural Discretization . 1.3.2 Coordinate Systems . . . 1.3.3 Sign Convention . . . . . 1.4 Degrees of Freedom . . . . . . . . 1.5 Course Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 17 18 19 20 21 21 22 25
Matrix Structural Analysis of Framed Structures

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
29 29 29 30 31 32 32 32 35 36 39 39 40 40 40 42 42 43 43 45 47 47 47 49 49 51 54 56
2 ELEMENT STIFFNESS MATRIX 2.1 Introduction . . . . . . . . . . . . . . . . 2.2 Inuence Coecients . . . . . . . . . . . 2.3 Flexibility Matrix (Review) . . . . . . . 2.4 Stiness Coecients . . . . . . . . . . . 2.5 Force-Displacement Relations . . . . . . 2.5.1 Axial Deformations . . . . . . . . 2.5.2 Flexural Deformation . . . . . . 2.5.3 Torsional Deformations . . . . . 2.5.4 Shear Deformation . . . . . . . . 2.6 Putting it All Together, [k] . . . . . . . 2.6.1 Truss Element . . . . . . . . . . 2.6.2 Beam Element . . . . . . . . . . 2.6.2.1 Euler-Bernoulli . . . . . 2.6.2.2 Timoshenko Beam . . . 2.6.3 2D Frame Element . . . . . . . . 2.6.4 Grid Element . . . . . . . . . . . 2.6.5 3D Frame Element . . . . . . . . 2.7 Remarks on Element Stiness Matrices . 2.8 Homework . . . . . . . . . . . . . . . . . 3 STIFFNESS METHOD; 3.1 Introduction . . . . . . 3.2 The Stiness Method . 3.3 Examples . . . . . . . E 3-1 Beam . . . . . E 3-2 Frame . . . . . E 3-3 Grid . . . . . . 3.4 Observations . . . . .
Part I: ORTHOGONAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
STRUCTURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Draft
CONTENTS 5.7
5 Nodal Displacements Reactions . . . . . . . Internal Forces . . . . Sample Output File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 111 112 113 115 117 117 117 117 119 120 120 122 124 124 125 125 126 126 127 127 128 130 133 133 133 134 135 135 135 137 137 137 138 139 140 141 142 143 143 145 145 145 145 145 145
5.6.2.11 5.6.2.12 5.6.2.13 5.6.2.14 Homework . . .
6 EQUATIONS OF STATICS and KINEMATICS 6.1 Statics Matrix [B] . . . . . . . . . . . . . . . . . . . . . 6.1.1 Statically Determinate . . . . . . . . . . . . . . E 6-1 Statically Determinate Truss Statics Matrix . . E 6-2 Beam Statics Matrix . . . . . . . . . . . . . . . 6.1.2 Statically Indeterminate . . . . . . . . . . . . . E 6-3 Statically Indeterminate Truss Statics Matrix . E 6-4 Selection of Redundant Forces . . . . . . . . . 6.1.3 Kinematic Instability . . . . . . . . . . . . . . . 6.2 Kinematics Matrix [A] . . . . . . . . . . . . . . . . . . E 6-5 Kinematics Matrix of a Truss . . . . . . . . . . 6.3 Statics-Kinematics Matrix Relationship . . . . . . . . 6.3.1 Statically Determinate . . . . . . . . . . . . . . 6.3.2 Statically Indeterminate . . . . . . . . . . . . 6.4 Kinematic Relations through Inverse of Statics Matrix 6.5 Congruent Transformation Approach to [K] . . . . . E 6-6 Congruent Transformation . . . . . . . . . . . . E 6-7 Congruent Transformation of a Frame . . . . .
7 FLEXIBILITY METHOD 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Flexibility Matrix . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 Solution of Redundant Forces . . . . . . . . . . . . 7.2.2 Solution of Internal Forces and Reactions . . . . . 7.2.3 Solution of Joint Displacements . . . . . . . . . . . E 7-1 Flexibility Method . . . . . . . . . . . . . . . . . . 7.3 Stiness Flexibility Relations . . . . . . . . . . . . . . . . 7.3.1 From Stiness to Flexibility . . . . . . . . . . . . . E 7-2 Flexibility Matrix . . . . . . . . . . . . . . . . . . 7.3.2 From Flexibility to Stiness . . . . . . . . . . . . . E 7-3 Flexibility to Stiness . . . . . . . . . . . . . . . . 7.4 Stiness Matrix of a Curved Element . . . . . . . . . . . . 7.5 Duality between the Flexibility and the Stiness Methods 7.6 Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 SPECIAL ANALYSIS PROCEDURES 8.1 Semi-Rigid Beams . . . . . . . . . . . . 8.2 Nonuniform Torsion . . . . . . . . . . . 8.3 Inclined Supports . . . . . . . . . . . . . 8.4 Condensation . . . . . . . . . . . . . . . 8.5 Substructuring . . . . . . . . . . . . . . 8.6 Reanalysis . . . . . . . . . . . . . . . . . 8.7 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Victor Saouma
Finite Element I; Framed Structures
Draft
CONTENTS
7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . second Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 188 188 188 189 189 189 190 190 191 191 194 197 198 199 201 203 203 203 204 205 205 207 209 209 210 211 211 212 213 213 213 214 214 214 216 217 217 217 218 218 218 219 220 220 220 221
E 10-8 Tapered Beam; Fourrier Series . . . . . . . . . . . . . . . 10.4 Complementary Potential Energy . . . . . . . . . . . . . . . . . 10.4.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4.2 Castiglianos Second Theorem . . . . . . . . . . . . . . . . E 10-9 Cantilivered beam . . . . . . . . . . . . . . . . . . . . . . 10.4.2.1 Distributed Loads . . . . . . . . . . . . . . . . . E 10-10Deection of a Uniformly loaded Beam using Castiglianos 10.5 Comparison of Alternate Approximate Solutions . . . . . . . . . E 10-11Comparison of MPE Solutions . . . . . . . . . . . . . . . 10.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.7 Variational Calculus; Preliminaries . . . . . . . . . . . . . . . . . 10.7.1 Euler Equation . . . . . . . . . . . . . . . . . . . . . . . . 10.7.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . . E 10-12Extension of a Bar . . . . . . . . . . . . . . . . . . . . . . E 10-13Flexure of a Beam . . . . . . . . . . . . . . . . . . . . . . 10.8 Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11 INTERPOLATION FUNCTIONS 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Shape Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2.1 Axial/Torsional . . . . . . . . . . . . . . . . . . . . . . . . 11.2.2 Generalization . . . . . . . . . . . . . . . . . . . . . . . . 11.2.3 Flexural . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2.4 Constant Strain Triangle Element . . . . . . . . . . . . . 11.3 Interpolation Functions . . . . . . . . . . . . . . . . . . . . . . . 11.3.1 C 0 : Lagrangian Interpolation Functions . . . . . . . . . . 11.3.1.1 Constant Strain Quadrilateral Element . . . . . 11.3.1.2 Solid Rectangular Trilinear Element . . . . . . . 11.3.2 C 1 : Hermitian Interpolation Functions . . . . . . . . . . . 11.4 Interpretation of Shape Functions in Terms of Polynomial Series 12 FINITE ELEMENT FORMULATION 12.1 Strain Displacement Relations . . . . . 12.1.1 Axial Members . . . . . . . . . 12.1.2 Flexural Members . . . . . . . 12.2 Virtual Displacement and Strains . . . 12.3 Element Stiness Matrix Formulation 12.3.1 Stress Recovery . . . . . . . . . 13 SOME FINITE ELEMENTS 13.1 Introduction . . . . . . . . . . . . . . . 13.2 Truss Element . . . . . . . . . . . . . . 13.3 Flexural Element . . . . . . . . . . . . 13.4 Triangular Element . . . . . . . . . . . 13.4.1 Strain-Displacement Relations 13.4.2 Stiness Matrix . . . . . . . . . 13.4.3 Internal Stresses . . . . . . . . 13.4.4 Observations . . . . . . . . . . 13.5 Quadrilateral Element . . . . . . . . . 13.6 Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Victor Saouma
Draft
CONTENTS
9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 270 270 271 271 272 272 272 273 273
E B-3 Example . . . . . . . . . . . . . . . . B.2.3 Choleskys Decomposition . . . . . . E B-4 Choleskys Decomposition . . . . . . B.2.4 Pivoting . . . . . . . . . . . . . . . . B.3 Indirect Methods . . . . . . . . . . . . . . . B.3.1 Gauss Seidel . . . . . . . . . . . . . B.4 Ill Conditioning . . . . . . . . . . . . . . . . B.4.1 Condition Number . . . . . . . . . . B.4.2 Pre Conditioning . . . . . . . . . . . B.4.3 Residual and Iterative Improvements
C TENSOR NOTATION 275 C.1 Engineering Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 C.2 Dyadic/Vector Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 C.3 Indicial/Tensorial Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 D INTEGRAL THEOREMS 279 D.1 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 D.2 Green-Gradient Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 D.3 Gauss-Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Victor Saouma
Draft
List of Figures
1.1 1.2 1.3 1.4 1.5 1.6 1.7 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3.1 3.2 3.3 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 Global Coordinate System . . . . . . . . . . . . . . . . . Local Coordinate Systems . . . . . . . . . . . . . . . . . Sign Convention, Design and Analysis . . . . . . . . . . Total Degrees of Freedom for various Type of Elements Independent Displacements . . . . . . . . . . . . . . . . Examples of Global Degrees of Freedom . . . . . . . . . Organization of the Course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 21 22 22 23 24 26 30 33 34 34 35 37 38 39 40
Example for Flexibility Method . . . . . . . . . . . . . . . . . . . . . Denition of Element Stiness Coecients . . . . . . . . . . . . . . . Stiness Coecients for One Dimensional Elements . . . . . . . . . . Flexural Problem Formulation . . . . . . . . . . . . . . . . . . . . . . Torsion Rotation Relations . . . . . . . . . . . . . . . . . . . . . . . Deformation of an Innitesimal Element Due to Shear . . . . . . . . Eect of Flexure and Shear Deformation on Translation at One End Eect of Flexure and Shear Deformation on Rotation at One End . . Coordinate System for Element Stiness Matrices . . . . . . . . . . .
Problem with 2 Global d.o.f. 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Frame Example (correct K23 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Grid Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Arbitrary 3D Vector Transformation . . . . . . . . . . . 3D Vector Transformation . . . . . . . . . . . . . . . . . 2D Frame Element Rotation with respect to Z (or z) . . Grid Element Rotation . . . . . . . . . . . . . . . . . . . 2D Truss Rotation . . . . . . . . . . . . . . . . . . . . . Reduced 3D Rotation . . . . . . . . . . . . . . . . . . . Special Case of 3D Transformation for Vertical Members Complex 3D Rotation . . . . . . . . . . . . . . . . . . . Frame Example . . . . . . . . . . . . . . . . . . . . Example for [ID] Matrix Determination . . . . . . Simple Frame Analyzed with the MATLAB Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Simple Frame Analyzed with the MATLAB Code . Stiness Analysis of one Element Structure . . . . Example of Bandwidth . . . . . . . . . . . . . . . . Numbering Schemes for Simple Structure . . . . . Program Flowchart . . . . . . . . . . . . . . . . . . Programs Tree Structure . . . . . . . . . . . . . . Flowchart for the Skyline Height Determination . . Flowchart for the Global Stiness Matrix Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 61 62 63 64 65 66 67 69 72 73 75 80 82 86 87 91 92 94 96
Draft
List of Tables
1.1 1.2 1.3 1.4 2.1 4.1 6.1 6.2 10.1 10.2 10.3 10.4 Example of Nodal Denition . Example of Element Denition Example of Group Number . . Degrees of Freedom of Dierent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structure Types Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 20 21 23
Examples of Inuence Coecients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3D Transformations of Linear Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Internal Element Force Denition for the Statics Matrix . . . . . . . . . . . . . . . . . . . 118 Conditions for Static Determinacy, and Kinematic Instability . . . . . . . . . . . . . . . . 124 Possible Combinations of Real and Hypothetical Formulations . . Comparison of 2 Alternative Approximate Solutions . . . . . . . Summary of Variational Terms Associated with One Dimensional Essential and Natural Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 191 194 198
11.1 Characteristics of Beam Element Shape Functions . . . . . . . . . . . . . . . . . . . . . . 208 11.2 Interpretation of Shape Functions in Terms of Polynomial Series (1D & 2D) . . . . . . . . 212 11.3 Polynomial Terms in Various Element Formulations (1D & 2D) . . . . . . . . . . . . . . . 212
Draft
a A A b B [B ] [B] C [C1|C2] {d} {dc } [D] E [E] {F} {F0 } {Fx } {Fe } 0 {F } e {F } {F} FEA G I [L] [I] [ID] J [k] [p] [kg ] [kr ] [K] [Kg ] L L lij {LM } {N} {p} {P} P, V, M, T R S t t u u u(x) u, v, w U
LIST OF TABLES
15
NOTATION
Vector of coeccients in assumed displacement eld Area Kinematics Matrix Body force vector Statics Matrix, relating external nodal forces to internal forces Statics Matrix relating nodal load to internal forces p = [B ]P Matrix relating assumed displacement elds parameters to joint displacements Cosine Matrices derived from the statics matrix Element exibility matrix (lc) Structure exibility matrix (GC) Elastic Modulus Matrix of elastic constants (Constitutive Matrix) Unknown element forces and unknown support reactions Nonredundant element forces (lc) Redundant element forces (lc) Element forces (lc) Nodal initial forces Nodal energy equivalent forces Externally applied nodal forces Fixed end actions of a restrained member Shear modulus Moment of inertia Matrix relating the assumed displacement eld parameters to joint displacements Idendity matrix Matrix relating nodal dof to structure dof St Venants torsional constant Element stiness matrix (lc) Matrix of coecients of a polynomial series Geometric element stiness matrix (lc) Rotational stiness matrix ( [d] inverse ) Structure stiness matrix (GC) Structures geometric stiness matrix (GC) Length Linear dierential operator relating displacement to strains Direction cosine of rotated axis i with respect to original axis j structure dof of nodes connected to a given element Shape functions Element nodal forces = F (lc) Structure nodal forces (GC) Internal forces acting on a beam column (axial, shear, moment, torsion) Structure reactions (GC) Sine Traction vector Specied tractions along t Displacement vector Neighbour function to u(x) Specied displacements along u Translational displacements along the x, y, and z directions Strain energy Finite Element I; Framed Structures
Victor Saouma
Draft
Chapter 1
INTRODUCTION
1.1
1
Why Matrix Structural Analysis?
In most Civil engineering curriculum, students are required to take courses in: Statics, Strength of Materials, Basic Structural Analysis. This last course is a fundamental one which introduces basic structural analysis (determination of reactions, deections, and internal forces) of both statically determinate and indeterminate structures.
2
Also Energy methods are introduced, and most if not all examples are two dimensional. Since the emphasis is on hand solution, very seldom are three dimensional structures analyzed. The methods covered, for the most part lend themselves for back of the envelope solutions and not necessarily for computer implementation.
3 Those students who want to pursue a specialization in structural engineering/mechanics, do take more advanced courses such as Matrix Structural Analysis and/or Finite Element Analysis.
Matrix Structural Analysis, or Advanced Structural Analysis, or Introduction to Structural Engineering Finite Element, builds on the introductory analysis course to focus on those methods which lend themselves to computer implementation. In doing so, we will place equal emphasis on both two and three dimensional structures, and develop a thorough understanding of computer aided analysis of structures.
4 5
This is essential, as in practice most, if not all, structural analysis are done by the computer and it is imperative that as structural engineers you understand what is inside those black boxes, develop enough self assurance to be capable of opening them and modify them to perform certain specic tasks, and most importantly to understand their limitations.
6 With the recently placed emphasis on the nite element method in most graduate schools, many students have been tempted to skip a course such as this one and rush into a nite element one. Hence it is important that you understand the connection and role of those two courses. The Finite Element Method addresses the analysis of two or three dimensional continuum. As such, the primary unknowns is u the nodal displacements, and internal forces are usually restricted to stress . The only analogous one dimensional structure is the truss. 7 Whereas two and three dimensional continuum are essential in civil engineering to model structures such as dams, shells, and foundation, the majority of Civil engineering structures are constituted by rod one-dimensional elements such as beams, girders, or columns. For those elements, displacements and internal forces are somehow more complex than those encountered in continuum nite elements. 8 Hence, contrarily to continuum nite element where displacement is mostly synonymous with translation, in one dimensional elements, and depending on the type of structure, generalized displacements may include translation, and/or exural and/or torsional rotation. Similarly, internal forces are not stresses, but rather axial and shear forces, and/or exural or torsional moments. Those concepts are far
Draft
13
1.3 Structural Idealization 3. Type of solution: (a) Continuum, analytical, Partial Dierential Equation (b) Discrete, numerical, Finite ELement, Finite Dierence, Boundary Element Structural design must satisfy: 1. Strength ( < f ) 2. Stiness (small deformations) 3. Stability (buckling, cracking)
19
14
Structural analysis must satisfy 1. Statics (equilibrium) 2. Mechanics (stress-strain or force displacement relations) 3. Kinematics (compatibility of displacement)
1.3
15
Structural Idealization
Prior to analysis, a structure must be idealized for a suitable mathematical representation. Since it is practically impossible (and most often unnecessary) to model every single detail, assumptions must be made. Hence, structural idealization is as much an art as a science. Some of the questions confronting the analyst include: 1. Two dimensional versus three dimensional; Should we model a single bay of a building, or the entire structure? 2. Frame or truss, can we neglect exural stiness? 3. Rigid or semi-rigid connections (most important in steel structures) 4. Rigid supports or elastic foundations (are the foundations over solid rock, or over clay which may consolidate over time) 5. Include or not secondary members (such as diagonal braces in a three dimensional analysis). 6. Include or not axial deformation (can we neglect the axial stiness of a beam in a building?) 7. Cross sectional properties (what is the moment of inertia of a reinforced concrete beam?) 8. Neglect or not haunches (those are usually present in zones of high negative moments) 9. Linear or nonlinear analysis (linear analysis can not predict the peak or failure load, and will underestimate the deformations). 10. Small or large deformations (In the analysis of a high rise building subjected to wind load, the moments should be amplied by the product of the axial load times the lateral deformation, P eects). 11. Time dependent eects (such as creep, which is extremely important in prestressed concrete, or cable stayed concrete bridges). 12. Partial collapse or local yielding (would the failure of a single element trigger the failure of the entire structure?). Victor Saouma Finite Element I; Framed Structures
Draft
1.3 Structural Idealization Group No. 1 2 3 Element Type 1 2 1 Material Group 1 1 2
21
Table 1.3: Example of Group Number
Y X X Z
Y X 3D TRUSS GRID & FRAME
BEAM
2D TRUSS FRAME
Figure 1.1: Global Coordinate System
1.3.2
22
Coordinate Systems
We should dierentiate between 2 coordinate systems:
Global: to describe the structure nodal coordinates. This system can be arbitrarily selected provided it is a Right Hand Side (RHS) one, and we will associate with it upper case axis labels, X, Y, Z, Fig. 1.1 or 1,2,3 (running indeces within a computer program). Local: system is associated with each element and is used to describe the element internal forces. We will associate with it lower case axis labels, x, y, z (or 1,2,3), Fig. 1.2.
23 The x-axis is assumed to be along the member, and the direction is chosen such that it points from the 1st node to the 2nd node, Fig. 1.2.
24
Two dimensional structures will be dened in the X-Y plane.
1.3.3
Sign Convention
25 The sign convention in structural analysis is completely dierent than the one previously adopted in structural analysis/design, Fig. 1.3 (where we focused mostly on exure and dened a positive moment as one causing tension below. This would be awkward to program!).
y, 2 x, 1 x, 1 z, 3
BEAM, TRUSS
GRID, FRAME
Figure 1.2: Local Coordinate Systems Victor Saouma Finite Element I; Framed Structures
Draft
1.4 Degrees of Freedom
23
Figure 1.5: Independent Displacements various types of structures made up of one dimensional rod elements, Table 1.4.
34
This table shows the degree of freedoms and the corresponding generalized forces. Type Node 1 Node 2 1 Dimensional Fy3 , Mz4 v3 , 4 2 Dimensional Fx2 u2 Fx4 , Fy5 , Mz6 u 4 , v5 , 6 Tx4 , Fy5 , Mz6 44 44 [k] (Local) [K] (Global)
Beam
{p} {}
Fy1 , Mz2 v1 , 2 Fx1
Truss
{p} {} {p} {} {p} {}
u1 Fx1 , Fy2 , Mz3 u 1 , v2 , 3 Tx1 , Fy2 , Mz3 1 , v2 , 3 Fx1 ,
22 66 66
44 66 66
Frame
Grid
Truss
{p} {} {p}
4 , v5 , 6 3 Dimensional Fx2 u2 Fx7 , Fy8 , Fy9 , Tx10 My11 , Mz12 u 7 , v8 , w 9 , 10 , 11 12
u1 , Fx1 , Fy2 , Fy3 , Tx4 My5 , Mz6 u 1 , v2 , w 3 , 4 , 5 6
22
66
Frame {}
12 12
12 12
Table 1.4: Degrees of Freedom of Dierent Structure Types Systems

35
We should distinguish between local and global d.o.f.s. The numbering scheme follows the following simple rules: Local: d.o.f. for a given element: Start with the rst node, number the local d.o.f. in the same order as the subscripts of the relevant local coordinate system, and repeat for the second node.
Global: d.o.f. for the entire structure: Starting with the 1st node, number all the unrestrained global d.o.f.s, and then move to the next one until all global d.o.f have been numbered, Fig. 1.6. Victor Saouma Finite Element I; Framed Structures
Draft
1.5
1.5 Course Organization
25
Course Organization
Victor Saouma
Draft
Part I
Matrix Structural Analysis of Framed Structures
Draft
Chapter 2
ELEMENT STIFFNESS MATRIX

2.1 Introduction
1 In this chapter, we shall derive the element stiness matrix [k] of various one dimensional elements. Only after this important step is well understood, we could expand the theory and introduce the structure stiness matrix [K] in its global coordinate system. 2 As will be seen later, there are two fundamentally dierent approaches to derive the stiness matrix of one dimensional element. The rst one, which will be used in this chapter, is based on classical methods of structural analysis (such as moment area or virtual force method). Thus, in deriving the element stiness matrix, we will be reviewing concepts earlier seen.
The other approach, based on energy consideration through the use of assumed shape functions, will be examined in chapter 12. This second approach, exclusively used in the nite element method, will also be extended to two and three dimensional continuum elements.
3
2.2
Inuence Coecients
4 In structural analysis an inuence coecient C ij can be dened as the eect on d.o.f. i due to a unit action at d.o.f. j for an individual element or a whole structure. Examples of Inuence Coecients are shown in Table 2.1.
Inuence Line Inuence Line Inuence Line Flexibility Coecient Stiness Coecient
Unit Action Load Load Load Load Displacement
Eect on Shear Moment Deection Displacement Load
Table 2.1: Examples of Inuence Coecients

5 It should be recalled that inuence lines are associated with the analysis of structures subjected to moving loads (such as bridges), and that the exibility and stiness coecients are components of matrices used in structural analysis.
Draft
2.4 Stiness Coecients
31
Virtual Force: x x dvol M xy x = I My x U = x = E = EI y 2 dA = I dvol = dAdx W = P U = W U = l M dx M EI 0
M
0
M dx = P EI
(2.3)
Hence: EI 1
M
d11 =
0
1
M M
x L
dx =
L 3
(2.4)
Similarly, we would obtain:

L
EId22 EId12
=
0 L
x L 1
dx =
L 3 = EId21
(2.5-a) (2.5-b)
=
0
x L
x L dx = L 6
Those results can be summarized in a matrix form as: [d] = L 6EIz 2 1 1 2 (2.6)
The exibility method will be covered in more detailed, in chapter 7.
2.4
9
Stiness Coecients
In the exibility method, we have applied a unit force at a time and determined all the induced displacements in the statically determinate structure.
10
In the stiness method, we 1. Constrain all the degrees of freedom 2. Apply a unit displacement at each d.o.f. (while restraining all others to be zero) 3. Determine the reactions associated with all the d.o.f.
{p} = [k]{}
11
(2.7)
Hence kij will correspond to the reaction at dof i due to a unit deformation (translation or rotation) at dof j, Fig. 2.2. Finite Element I; Framed Structures
Victor Saouma
Draft
2.5 Force-Displacement Relations
33
Figure 2.2: Denition of Element Stiness Coecients
Victor Saouma
Draft
2.7 Remarks on Element Stiness Matrices Upon substitution, the grid element stiness matrix is given by 1x T1x GIx L V1y 0 M1z 0 T2x Gix L V2y 0 M2z 0 u1y 0
12EIz L3 6EIz L2
43
1z 0
6EIz L2 4EIz L
[kg ] =
2x GIx L 0 0
GIx L
u2y 0 12EIz L3 6EIz L2 0

12EIz L3 6EIz L2
12EIz L3
6EIz L2
6EIz L2
2EIz L3
0 0
2z 0 6EIz L2 2EIz L 0 6EIz L2 4EIz

L
(2.46)
52
Note that if shear deformations must be accounted for, the entries corresponding to shear and exure must be modied in accordance with Eq. 2.42
2.6.5
3D Frame Element
u1 t Px1 k11 0 Vy1 V z1 0 Tx1 0 My1 0 Mz1 0 t Px2 k21 Vy2 0 V z2 0 Tx2 0 My2 0 Mz2 0
u1 EA l 0
[k3df r ] =
v1 0 b k11 0 0 0 b k21 0 b k13 0 0 0 b k12
w1 0 0 b k11 0 b k32 0 0 0 b k13 0 b k12 0
x1 0 0 0 g k11 0 0 0 0 0 g k12 0 0
y1 0 0 b k12 0 b k22 0 0 0 b k14 0 b k24 0
z1 0 b k12 0 0 0 b k22 0 b k14 0 0 0 b k24
u2 t k21 0 0 0 0 0 t k22 0 0 0 0 0
v2 0 b k13 0 0 0 b k12 0 b k33 0 0 0 b k43
w2 0 0 b k13 0 b k12 0 0 0 b k33 0 b k43 0
x2 0 0 0 g k12 0 0 0 0 0 g k22 0 0
y2 0 0 b k14 0 b k24 0 0 0 b k34 0 b k44 0
For [k3D ] and with we obtain: 11

v1 0 12EIz L3 0 0 0 6EIz L2 0 12EIz L3 0 0 0 6EIz L2 w1 0 0 12EIy L3 0 6EIy L2 0 0 0 12EIy L3 0 6EIy L2 0 x1 0 0 0 GIx L 0 0 0 0 0 GIx L 0 0 y1 0 0 6EIy L2 0 4EIy L 0 0 0 EIy 6 L2 0 2EIy L 0 z1 0 6EIz L2 0 0 0 4EIz L 0 6EIz L2 0 0 0 2EIz L u2 EA L 0 0 0 0 0 EA L 0 0 0 0 0 v2 0 12EIz L3 0 0 0 6EIz L2 0 12EIz L3 0 0 0 6EIz L2 w2 0 0 12EIy L3 0 6EIy L2 0 0 0 12EIy L3 0 6EIy L2 0 x2 0 0 0 GIx L 0 0 0 0 0 GIx L 0 0 y2 0
z2 0 k b 14 0 0 0 b k24 0 b k34 0 0 0 b k44

z2 0 6EIz L2 0 0 0 2EIz L 0 6EIz L2 0 0 0 4EIz L
(2.47)
[k
3d r f
] =
0 Vz1 Tx1 0 0 My1 0 Mz1 Px2 EA L 0 Vy2 0 Vz2 Tx2 0 0 My2

Vy1 Mz2 0
Px1
0 EIy 6 L2 0 2EIy L 0 0 0 6EIy L2 0 4EIy L 0
(2.48)
53
Note that if shear deformations must be accounted for, the entries corresponding to shear and exure must be modied in accordance with Eq. 2.42
2.7
Remarks on Element Stiness Matrices
Singularity: All the derived stiness matrices are singular, that is there is at least one row and one column which is a linear combination of others. For example in the beam element, row 4 = row Victor Saouma Finite Element I; Framed Structures
Draft
2.8 Homework
45
2.8
Homework
Using the virtual force method, derive the exibility matrix of a semi-circular box-girder of radius R and angle in terms of shear, axial force, and moment. The arch is clamped at one end, and free at the other. Note: In a later assignment, you will combine the exibility matrix with equilibrium relations to derive the element stiness matrix.
Y O Z
Victor Saouma

Y X O Z
y z
Draft
Chapter 3
STIFFNESS METHOD; Part I: ORTHOGONAL STRUCTURES

3.1
1
Introduction
In the previous chapter we have rst derived displacement force relations for dierent types of rod elements, and then used those relations to dene element stiness matrices in local coordinates.
2 In this chapter, we seek to perform similar operations, but for an orthogonal structure in global coordinates.
3 In the previous chapter our starting point was basic displacement-force relations resulting in element stiness matrices [k].
In this chapter, our starting point are those same element stiness matrices [k], and our objective is to determine the structure stiness matrix [K], which when inverted, would yield the nodal displacements. The element stiness matrices were derived for fully restrained elements.
This chapter will be restricted to orthogonal structures, and generalization will be discussed later. The stiness matrices will be restricted to the unrestrained degrees of freedom.
7
From these examples, the interrelationships between structure stiness matrix, nodal displacements, and xed end actions will become apparent. Then the method will be generalized in chapter 5 to describe an algorithm which can automate the assembly of the structure global stiness matrix in terms of the one of its individual elements.
3.2
The Stiness Method
8 As a vehicle for the introduction to the stiness method let us consider the problem in Fig 3.1-a, and recognize that there are only two unknown displacements, or more precisely, two global d.o.f: 1 and 2 . 9
If we were to analyse this problem by the force (or exibility) method, then 1. We make the structure statically determinate by removing arbitrarily two reactions (as long as the structure remains stable), and the beam is now statically determinate. 2. Assuming that we remove the two roller supports, then we determine the corresponding deections due to the actual laod (B and C ).
Draft
Chapter 4
TRANSFORMATION MATRICES
4.1
4.1.1
Preliminaries
[ke ] [Ke ] Relation
1 In the previous chapter, in which we focused on orthogonal structures, the assembly of the structures stiness matrix [Ke ] in terms of the element stiness matrices was relatively straight-forward. 2 The determination of the element stiness matrix in global coordinates, from the element stiness matrix in local coordinates requires the introduction of a transformation. 3 This chapter will examine the 2D and 3D transformations required to obtain an element stiness matrix in global coordinate system prior to assembly (as discussed in the next chapter).
Recalling that {p} {P} = [k(e) ]{} = [K(e) ]{} (4.1) (4.2)
Let us dene a transformation matrix [(e) ] such that: {p} {} = = [(e) ]{} [
(e)
(4.3) (4.4)
]{P}
Note that we use the same matrix one).

6
(e)
since both {} and {p} are vector quantities (or tensors of order
Substituting Eqn. 4.3 and Eqn. 4.4 into Eqn. 4.1 we obtain [(e) ]{P} = [k(e) ][(e) ]{} (4.5) (4.6)
premultiplying by [(e) ]1 {P} = [(e) ]1 [k(e) ][(e) ]{}

7
But since the rotation matrix is orthogonal, we have [(e) ]1 = [(e) ]T and {P} = [(e) ]T [k(e) ][(e) ]{}
[K(e) ]
(4.7)
[K(e) ] = [(e) ]T [k(e) ][(e) ] which is the general relationship between element stiness matrix in local and global coordinates.
(4.8)
Draft
11
4.1 Preliminaries
61
where lij is the direction cosine of axis i with respect to axis j, and thus the rows of the matrix correspond to the rotated vectors with respect to the original ones corresponding to the columns. With respect to Fig. 4.2, lxX = cos ; lxY = cos , and lxZ = cos or
y VY
Y x
V
VZ Z z VX
Figure 4.2: 3D Vector Transformation Vx = = VX lxX + VY lxY + VZ lxZ VX cos + VY cos + VZ cos (4.14-a) (4.14-b)
12
Direction cosines are unit orthogonal vectors satisfying the following relations:
3
lij lij = 1
j=1
i = 1, 2, 3
(4.15)
i.e:
2 2 2 l11 + l12 + l13 = 1 or cos2 + cos2 + cos2 = 1 = 11
(4.16)
and
3
lij lkj
j=1
l11 l21 + l12 l22 + l13 l23

T
i = 1, 2, 3 k = 1, 2, 3 = 0 i=k = 0 = 12
T T 1
(4.17-a) (4.17-b)
By direct multiplication of [] and [] it can be shown that: [] [] = [I] [] = [] an orthogonal matrix.

13 14
[] is
The reverse transformation (from local to global) would be {V} = [] {v}

T
(4.18)
or
lxX VX VY = lxY lxZ VZ
lyX lyY lyZ
[ ]1 =[ ]T
lzX Vx Vy lxY Vz lzZ
(4.19)
Victor Saouma
Draft
4.2 Transformation Matrices For Framework Elements
67
Y Y
2
Y Y
1
Z Z
Z
1
X
2
Y Y Z Z X X
Y Y
2
Figure 4.8: Complex 3D Rotation
Victor Saouma
Draft
Chapter 5
STIFFNESS METHOD; Part II

5.1
5.1.1
Direct Stiness Method

Global Stiness Matrix
1 The physical interpretation of the global stiness matrix K is analogous to the one of the element, i.e. If all degrees of freedom are restrained, then Kij corresponds to the force along global degree of freedom i due to a unit positive displacement (or rotation) along global degree of freedom j.
For instance, with reference to Fig. 5.1, we have three global degrees of freedom, 1 , 2 , and 3 . and
EI
K11
Figure 5.1: Frame Example
qr5qr5q 5
5 VW5VW
5 pi 5ip
IPIP
K21
K31
F5F 5 5 EEF5EEF
K12
55 XY5XY5XY
QRQR
QR
IP
5S5 T ST55ST5USU
K22
K 32
55 46546546
t s5 5st
UU
D C5 D5 C5 5CDCD
e e 55e h gh g55 f f 55gh 55f
5 ba 5ab
L/2
L/2
P/2
878
55 9@59@59@5`9`
``
A A5A BB5A
P
A B
HGH G dcd c
w
C
K23 1
K 33 K 13
Draft
5.2 Logistics
71
(e)
at the element level where pint is the six by six array of internal forces, k(e) the element stiness matrix in local coordinate systems, and (e) is the vector of nodal displacements in local coordinate system. Note that this last array is obtained by rst identifying the displacements in global coordinate system, and then premultiplying it by the transformation matrix to obtain the displacements in local coordinate system.
5.2
5.2.1
10
Logistics
Boundary Conditions, [ID] Matrix
Because of the boundary condition restraints, the total structure number of active degrees of freedom (i.e unconstrained) will be less than the number of nodes times the number of degrees of freedom per node.
11
To obtain the global degree of freedom for a given node, we need to dene an [ID] matrix such that: ID has dimensions l k where l is the number of degree of freedom per node, and k is the number of nodes). ID matrix is initialized to zero. 1. At input stage read ID(idof,inod) of each degree of freedom for every node such that: ID(idof, inod) = 0 1 if unrestrained d.o.f. if restrained d.o.f. (5.6)
2. After all the node boundary conditions have been read, assign incrementally equation numbers (a) First to all the active dof (b) Then to the other (restrained) dof, starting with -1. Note that the total number of dof will be equal to the number of nodes times the number of dof/node NEQA. 3. The largest positive global degree of freedom number will be equal to NEQ (Number Of Equations), which is the size of the square matrix which will have to be decomposed.
12
For example, for the frame shown in Fig. 5.2: 1. The input data le may contain: Node No. 1 2 3 4 2. At this stage, the [ID] matrix is equal to: 0 ID = 0 0 1 1 0 0 0 0 1 0 0 (5.7) [ID]T 000 110 000 100
3. After we determined the equation numbers, we would have: 1 10 5 12 8 ID = 2 11 6 3 4 7 9 Victor Saouma
(5.8)
Draft
5.2 Logistics
20
73
[KS ](N EQAN EQA ) to zero, and then loop through all the elements, and then through each entry of
(e) (e)
the respective element stiness matrix Kij . The assignment of the element stiness matrix term Kij (note that e, i, and j are all known since we are looping on e from 1 to the number of elements, and then looping on the rows and columns of the S element stiness matrix i, j) into the global stiness matrix Kkl is made through the LM vector (note that it is k and l which must be determined).
21 Since the global stiness matrix is also symmetric, we would need to only assemble one side of it, usually the upper one.
22
Contrarily to the previous method, we will assemble the full augmented stiness matrix. Example 5-1: Assembly of the Global Stiness Matrix As an example, let us consider the frame shown in Fig. 5.3.
50kN
8m
1
7.416 m
8m
Figure 5.3: Simple Frame Analyzed with the MATLAB Code The ID matrix is initially set to: 1 [ID] = 1 1 0 0 0 1 1 1
Finally the LM vectors for the two elements (assuming that Element 1 is dened from node 1 to node 2, and element 2 from node 2 to node 3): [LM ] = 4 1 5 2 6 3 1 7 2 8 3 9 (5.11)
We then modify it to generate the global degrees of freedom of each node: 4 1 7 [ID] = 5 2 8 6 3 9
Let us simplify the operation by designating the element stiness matrices in global coordinates as follows: 4 4 A11 5 A21 6 A31 1 A41 2 A51 3 A61 5 A12 A22 A32 A42 A52 A62 6 A13 A23 A33 A43 A53 A63 1 A14 A24 A34 A44 A54 A64 2 A15 A25 A35 A45 A55 A65 3 A16 A26 A36 A46 A56 A66
K (1)
Victor Saouma
yyyxy xxx
4 kN/m
3m
vuv vwv uw w vuv vwvuwwv vwv vuvuw vuvuv vwvuwvu vuvuwv wuuwuuww vvvuuuu
(5.9)
(5.10)
(5.12-a)
Draft
5.2 Logistics
24
75
5. Backsubstitute and obtain nodal displacements in global coordinate system.
6. Solve for the reactions, Eq. 5.3. 7. For each element, transform its nodal displacement from global to local coordinates {} = [(e) ]{}, and determine the internal forces [p] = [k]{}. Some of the prescribed steps are further discussed in the next sections. Example 5-2: Direct Stiness Analysis of a Truss Using the direct stiness method, analyze the truss shown in Fig. 5.4.
4 4 5 50k
5 1 3
6 8 12
2 100k
16
16
Figure 5.4: Solution: 1. Determine the structure ID matrix Node # 1 2 3 4 5 0 1 0 0 Bound. X 0 0 1 0 0 1 1 0 0 2 2 3 Cond. Y 1 0 1 0 0 0 0 3 9 10 4 5 4 6 5 7
ID
(5.14-a)
N ode 1 1 8
(5.14-b)
2. The LM vector of each element is evaluated next LM LM LM LM Victor Saouma

1 2 3 4
= = = =
1 8 2 3 2 3 4 5 4 5 6 7
(5.15-a) (5.15-b) (5.15-c) (5.15-d)
Draft
5.2 Logistics
81
% FEA_Rest for all the restrained nodes FEA_Rest=[0,0,0,FEA(4:6,2)]; % Assemble the load vector for the unrestrained node P(1)=50*3/8;P(2)=-50*7.416/8-FEA(2,2);P(3)=-FEA(3,2); % Solve for the Displacements in meters and radians Displacements=inv(Ktt)*P % Extract Kut Kut=Kaug(4:9,1:3); % Compute the Reactions and do not forget to add fixed end actions Reactions=Kut*Displacements+FEA_Rest % Solve for the internal forces and do not forget to include the fixed end actions dis_global(:,:,1)=[0,0,0,Displacements(1:3)]; dis_global(:,:,2)=[Displacements(1:3),0,0,0]; for elem=1:2 dis_local=Gamma(:,:,elem)*dis_global(:,:,elem); int_forces=k(:,:,elem)*dis_local+fea(1:6,elem) end
function [k,K,Gamma]=stiff(EE,II,A,i,j) % Determine the length L=sqrt((j(2)-i(2))^2+(j(1)-i(1))^2); % Compute the angle theta (careful with vertical members!) if(j(1)-i(1))~=0 alpha=atan((j(2)-i(2))/(j(1)-i(1))); else alpha=-pi/2; end % form rotation matrix Gamma Gamma=[ cos(alpha) sin(alpha) 0 0 0 0; -sin(alpha) cos(alpha) 0 0 0 0; 0 0 1 0 0; 0 0 0 cos(alpha) sin(alpha) 0; 0 0 0 -sin(alpha) cos(alpha) 0; 0 0 0 0 0 1]; % form element stiffness matrix in local coordinate system EI=EE*II; EA=EE*A; k=[EA/L, 0, 0, -EA/L, 0, 0, 12*EI/L^3, 6*EI/L^2, 0, -12*EI/L^3, 6*EI/L^2; 0, 6*EI/L^2, 4*EI/L, 0, -6*EI/L^2, 2*EI/L; -EA/L, 0, 0, EA/L, 0, 0; 0, -12*EI/L^3, -6*EI/L^2, 0, 12*EI/L^3, -6*EI/L^2; 0, 6*EI/L^2, 2*EI/L, 0, -6*EI/L^2, 4*EI/L]; % Element stiffness matrix in global coordinate system K=Gamma*k*Gamma; This simple proigram will produce the following results: Displacements = 0.0010 -0.0050 -0.0005 Reactions = 130.4973 55.6766 13.3742 Victor Saouma
0;
Draft
react3 intern3 % end % % draw
5.6 Computer Implementation with MATLAB
105
% CALCULATE THE INTERNAL FORCES FOR EACH ELEMENT
END LOOP FOR EACH LOAD CASE
DRAW THE STRUCTURE, IF USER HAS REQUESTED (DRAWFLAG=1) CALL SCRIPTFILE DRAW.M
st=fclose(all); % END OF MAIN PROGRAM (CASAP.M) disp(Program completed! - See "casap.out" for complete output);
5.6.2.2
Assembly of ID Matrix
%************************************************************************************************ %SCRIPTFILE NAME: IDRASMBL.M % %MAIN FILE : CASAP % %Description : This file re-assambles the ID matrix such that the restrained % degrees of freedom are given negative values and the unrestrained % degrees of freedom are given incremental values beginning with one % and ending with the total number of unrestrained degrees of freedom. % % %************************************************************************************************ % TAKE CARE OF SOME INITIAL BUSINESS: TRANSPOSE THE PNODS ARRAY
Pnods=Pnods.; % SET THE COUNTER TO ZERO
count=1; negcount=-1; % REASSEMBLE THE ID MATRIX
if istrtp==3 ndofpn=3; nterm=6; else error(Incorrect structure type specified) end % SET THE ORIGINAL ID MATRIX TO TEMP MATRIX orig_ID=ID; % % REASSEMBLE THE ID MATRIX, SUBSTITUTING RESTRAINED DEGREES OF FREEDOM WITH NEGATIVES, AND NUMBERING GLOBAL DEGREES OF FREEDOM
for inode=1:npoin XXXXXXXXXXXXXXXXXXXXXX COMPLETE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX % END OF IDRASMBL.M SCRIPTFILE
5.6.2.3
Element Nodal Coordinates
%**********************************************************************************************
Victor Saouma
Draft
6.1 Statics Matrix [B] the unknown forces and reactions 0 0 F1 F2 0 0 F3 0 0 0 0 F4 = Rx1 1 0 Ry1 0 1 Rx4 0 0 Ry4 0 0
{F}
119 can be determined through inversion of [B]: Py2 0 1 0 0 0 0 0 Px2 1 0 0 0 0 0 0 1 1 Px2 0 0 0 0 Px2 C C C S S Py2 S Px2 C 0 C 1 0 0 C = 1 0 1 0 0 0 0 Px2 S S S 1 0 0 0 0 C Px2 + Py2 C C 0 0 0 0 1 0 0 0 S S S 0 C 0 C 1 0 1 C Px2
[B]1 {P}
(6.3)
We observe that the matrix [B] is totally independent of the external load, and once inverted can be used for multiple load cases with minimal computational eorts.
Example 6-2: Beam Statics Matrix Considering the beam shown in Fig. 6.2, we have 3 elements, each with 2 internalunknowns (v and m) plus two unknown reactions, for a total of 8 unknowns. To solve for those unknowns we have 2 equations of equilibrium at each of the 4 nodes. Note that in this problem we have selected as primary unknowns the shear and moment at the right end of each element. The left components can be recovered from equilibrium. From equilibrium we thus have:
Figure 6.2: Example of [B] Matrix for a Statically Determinate Beam P1 M1 P2 M2 P3 M3 P4 M4 Victor Saouma 1 8 1 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 1 2 1 0 0 0 0 0 0 1 0 1 0 0
[B]
{P}
0 0 0 0 1 3 1 0
0 0 0 0 0 1 0 1
1 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0
v1 m1 v2 m2 v3 m3 R1 R2
{F}
(6.4)
Draft
H
6.1 Statics Matrix [B]

Y P y2 2 1 2 3 4 5 1 L 4 X Rx1 Px2 3 2 f1 Py1 f1 1
121
P y2 f2 f5 f3 Px1
Py3 f2 f3 f5 f 4 Rx4 4 R y4 3 f4 Py4 Px4 Px3
Px2
R y1
Figure 6.3: Example of [B] Matrix for a Statically Indeterminate Truss
We can solve for the internal forces in terms of the (still unknown) redundant force {F0 } F1 F2 F3 F4 Rx1 Ry1 Rx4 Ry4 Or using the = [B0 ]1 {P} [B0 ]1 [Bx ] {Fx } = 0 0 0 0 1 0 0 0 0 0 0 1 1 0 C S 0 C 0 1 S 1 C 0 0 S 0 C 1 0 0 0 0 1 0 0 0 0
1 C S C
0 S C
S C
0 0 0 1 0 0 0 1
0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 1
{F0 }
[B0 ]1 [C1 ]
0 0 Px2 Py2 0 0 0 0
{P}
0 0 C S {F5 } (6.10-b) 0 0 {Fx } C S

[Bx ]
(6.10-a)
{F0 }
following relations 0 0 F1 F2 0 0 F3 0 0 0 0 F4 = Rx1 1 0 Ry1 0 1 Rx4 0 0 Ry4 0 0
[B0 ]1 [C1 ] and [B0 ]1 [Bx ] [C2 ] we obtain 0 1 0 0 0 0 0 S 1 0 0 0 0 0 0 C 1 1 Px2 1 0 0 0 0 C C S S Py2 S C 0 C 1 0 0 + {F } 1 0 1 0 0 0 0 C 5 S S 0 0 {Fx } 1 0 0 0 C C 0 0 0 0 1 0 0 C S S 0 0 C 0 C 1 0 1
[C1 ] {P} [C2 ]
(6.11)
Note, that this equation is not sucient to solve for the unknown forces, as {F x } must be obtained through force displacement relations ([D] or [K]).
10 ddagWhereas the identication of redundant forces was done by mere inspection of the structure in hand based analysis of structure, this identication process can be automated.
11
Starting with [B]2n(2n+r) {F}2n+r1 [I]2n2n {P}2n1 = {0} B I

2n(4n+r)
{P}2n1 = [B]2n(2n+r) {F}(2n+r)1 F P
= {0}
(6.12-a)
Victor Saouma
Draft
6.1 Statics Matrix [B] 2. Interchange columns A B C D E F G H Rx1 1 0 0 0 0 0 0 0 Ry1 0 1 0 0 0 0 0 0 F2 0 0 1 0 1 0 0 0 F1 0 1 0 1 0 0 0 0 F3 C S 0 0 C S 0 0 F4 F5 0 0 0 0 0 C 0 S 0 0 1 0 0 C 1 S Rx4 0 0 0 0 0 0 1 0 Ry4 0 0 0 0 0 0 0 1 Px2 0 0 1 0 0 0 0 0 Py2 0 0 0 1 0 0 0 0 Py2 0 1 0 1 0 0 0 0 Py2 0 1 0 1 0 0 0 0
123
(6.17)
3. Operate as indicates A B = B + DB C D E = E+C C F G H 4. Operate as indicated = A + CE = B + SE = C = F SE =H +F Rx1 1 0 0 0 0 0 0 0 Ry1 0 1 0 0 0 0 0 0 F2 0 0 1 0 0 0 0 0 F1 0 0 0 1 0 0 0 0 F3 0 0 0 0 1 0 0 0 F4 0 0 0 0 0 1 0 0 F5 C 0 C S 1 S C 0 Rx4 0 0 0 0 0 0 1 0 Ry4 0 0 0 0 0 0 0 1 Px2 1 S/C 1 0 1/C S/C 0 S/C Rx1 1 0 0 0 0 0 0 0 Ry1 0 1 0 0 0 0 0 0 F2 0 0 1 0 1 0 0 0 F1 0 0 0 1 0 0 0 0 F3 C S 0 0 1 S 0 0 F4 F5 0 0 0 S 0 C 0 S 0 1 1 0 0 C 1 S Rx4 0 0 0 0 0 0 1 0 Ry4 0 0 0 0 0 0 0 1 Px2 0 0 1 0 1/C 0 0 0
(6.18)
A B C D E F G H
(6.19)
5. Interchange columns and observe that F5 is the selected redundant. A B C D E F G H Rx1 1 0 0 0 0 0 0 0 Ry1 0 1 0 0 0 0 0 0 F2 0 0 1 0 0 0 0 0 F1 0 0 0 1 0 0 0 0 F3 0 0 0 0 1 0 0 0 F4 0 0 0 0 0 1 0 0 Rx4 0 0 0 0 0 0 1 0 Ry4 0 0 0 0 0 0 0 1 F5 C 0 C S 1 S C 0 Px2 1 S/C 1 0 1/C S/C 0 S/C Py2 0 1 0 1 0 0 0 0
(6.20)
Victor Saouma
Draft
17 18
6.3 Statics-Kinematics Matrix Relationship
125
[A] is a rectangular matrix which number of rows is equal to the number of the element internal displacements, and the number of columns is equal to the number of nodal displacements. Contrarily to the rotation matrix introduced earlier and which transforms the displacements from global to local coordinate for one single element, the kinematics matrix applies to the entire structure. It can be easily shown that for trusses (which corresponds to shortening or elongation of the member): e = (u2 u1 ) cos + (v2 v1 ) sin (6.23)
where is the angle between the element and the X axis. whereas for exural members: v21 z21 = = v2 v1 z1 L z2 z1 (6.24-a) (6.24-b)
Example 6-5: Kinematics Matrix of a Truss Considering again the statically indeterminate truss of will be given by: e 0 1 0 1 0 1 e 2 0 1 0 1 0 e 3 C S 0 0 C e 4 0 0 0 0 0 e 0 C S 0 = 0 5 u1 1 0 0 0 0 v1 1 0 0 0 0 u4 0 0 0 0 0 v4 0 0 0 0 0
[A]
the previous example, the kinematic matrix 0 0 S 1 0 0 0 0 0 0 0 0 0 C 0 0 1 0 0 0 0 1 S 0 0 0 1
u1 v1 u2 v2 u3 v3 u4 v4
(6.25)
Applying the constraints: u1 = 0; v1 = 0; u4 = 0; and v4 = 0 we obtain: e 0 1 0 1 0 0 0 0 1 e 2 0 0 1 0 1 0 0 0 e 0 0 C S 0 0 3 C S e 0 0 0 0 1 0 1 4 0 e 0 C S 0 0 C S 5 = 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1

[A]
u1 v1 u2 v2 u3 v3 u4 v4
(6.26)
We should observe that [A] is indeed the transpose of the [B] matrix in Eq. 6.15
6.3
19
Statics-Kinematics Matrix Relationship
Having dened both the statics [B] and kinematics [A] matrices, it is intuitive that those two matrices must be related. In this section we seek to determine this relationship for both the statically determinate and statically indeterminate cases. Finite Element I; Framed Structures
Victor Saouma
Draft
6.4
29
6.4 Kinematic Relations through Inverse of Statics Matrix
127
Kinematic Relations through Inverse of Statics Matrix
28 We now seek to derive some additional relations between the displacements through the inverse of the statics matrix. Those relations will be used later in the exibility methods, and have no immediate applications.
Rewriting Eq. 6.33 as {} = [A0 ]1 {0 } = [B0 ]

t 1
{0 } = [B0 ]
1
{0 }
(6.36)
we can solve for {F0 } from Eq. 6.7 {F0 } = [B0 ]
{P} [B0 ]
[Bx ] {Fx }
(6.37)
[C1 ]
[C2 ]
30
Combining this equation with [B0 ]1 = [C1 ] from Eq. 6.37, and with Eq. 6.36 we obtain {} = [C1 ]t {0 } (6.38)
31
Similarly, we can revisit Eq. 6.33 and write {x } = [Ax ] {} (6.39)
When the previous equation is combined with the rightmost side of Eq. 6.36 and 6.35 we obtain {x } = [Bx ] [B0 ]
32
{0 }
(6.40)
Thus, with [B0 ]1 [Bx ] = [C2 ] from Eq. 6.37 {x } = [C2 ]t {0 } (6.41)
This equation relates the unknown relative displacements to the relative known ones.
6.5
Congruent Transformation Approach to [K]
Note: This section is largely based on section 3.3 of Gallagher, Finite Element Analysys, Prentice Hall.
33 For an arbitrary structure composed of n elements, we can dene the unconnected nodal load and displacement vectors in global coordinate as
where {Pi } and {i } are the nodal load and displacements arrays of element i. The size of each submatrix (or more precisely of each subarray) is equal to the total number of d.o.f. in global coordinate for element i.
34
{ }
{Pe }
e
= =
P1
1
P2
2
... ...
Pn
n
T T
(6.42-a) (6.42-b)
Similarly, we can dene the unconnected (or unassembled) global stiness matrix of the structure as [Ke ]: {F}
e
[K ]
[Ke ] {} [K1 ] [K2 ] =
[K ] .. . [Kn ]
(6.43-a)
(6.43-b)
Victor Saouma
Draft
6.5 Congruent Transformation Approach to [K]
129
Figure 6.4: Example 1, Congruent Transfer We shall determine the global stiness matrix using the two approaches: Direct Stiness [ID] {LM1 }
2
= = =
0 0 0 0 1
1 2 3 0 2
0 0 0
(6.47-a) 2 0 3 0
T T
1 0
(6.47-b) (6.47-c)
{LM }
[K]
(7.692 + 1 105 ) (0. + 0.) (18.75 + 0.) (0. + 0.) (.4 105 + 14.423) (12. + 0.) (0. + 18.75) (12. + 0.) (.0048 + .00469) 1 105 0. 18.75 0. .4 105 12. 18.75 12. .00949
(6.48-a)
Congruent Transformation 1. The unassembled stiness matrix [Ke ], for node 2, is given by:
{F}
1 Mx 1 My 1 Fz 2 Mx 2 My 2 Fz
{} 0 .4 105 0. 0 12. .0048 1 10 sym

5
7.692 sym
0. 0. 14.423 18.75 0. .00469
Note that the B.C. are implicitely accounted for by ignoring the restrained d.o.f. however the connectivity of the elements is not reected by this matrix. 2. The kinematics matrix is given by: {} = [A] {} (6.50-a)
1 x 1 y 1 Wz 2 x 2 y 2 Wz
(6.49-a)
element 1 (6.49-b) element 2
Victor Saouma
Draft
6.5 Congruent Transformation Approach to [K] []BC = = [I]
131 (6.52-b)
The element stiness matrices in global coordinates will then be given by: [K]AB []T [k]AB []AB AB .645 .259 7.031 .109 17.381 1 105 200 sym .645 .259 7.031 .645 .259 .109 17.381 .259 .109 7.031 17.381 .5 105 7.031 17.381 1 105 (6.53-a)
(6.53-b)
and [K]BC = [k]BC
Direct Stiness: We can readily assemble the global stiness matrix: (.645 + .75) sym 1.395 sym .259 .1137 (.259 + 0.) (.109 + .00469) 7.031 1.37 2 105 (7.031 + 0.) (17.38 + 18.75) (1 + 1) 105
[K]
200 200
(6.54-a)
(6.54-b)
Congruent Transformation, global axis, Boolean [A] 1. We start with the unconnected global stiness matrix in global coordinate system:
{F}
1 PX 1 PY 1 MZ 2 PX 2 PY 2 MZ e
2. Next we determine the kinematics matrix A: {} = [A] {} 1 1 0 u 1 v 0 1 1 0 0 = 2 1 0 u 2 v 0 1 2 0 0

36 66 63
200
{} .645 sym 0 sym .259 .109 7.031 17.381 1 105 0 .75 0. .00469 0. 18.75 1 105
(6.55-a)
1 UX 1 VY 1 Z 2 UX 2 VY 2 Z
(6.55-b)
0 0 1 0 0 1
(6.56-a)
3. Finally, if we take the product [A]T [Ke ] [A] we obtain the structure global stiness matrix [K] in Eq. 6.54-b Congruent Transformation (local axis): 1. Unconnected stiness matrix in local coordinates: Victor Saouma
{p }
1 PX 1 PY 1 MZ 2 PX 2 PY 2 MZ e
u v
(6.56-b)
200
{ } .75 sym 0. .00469 0. 18.75 1 105
.75 sym
0. .00469
0. 18.75 1 105
u1 x 1 vy 1 z u2 x 2 vy 2 z
(6.57-a)
(6.57-b)
Draft
Chapter 7
FLEXIBILITY METHOD
7.1
1
Introduction
Recall the denition of the exibility matrix (7.1)
{} [d]{p}
where {}, [d], and {p} are the element relative displacements, element exibility matrix, and forces at the element degrees of freedom free to displace.
2
As with the congruent approach for the stiness matrix, we dene: {F(e) } = = F(1) (1) F(2) (2) ... ... F(n) (n)
T T
(7.2-a) (7.2-b)
{(e) }
for n elements, and where {Fi } and {i } are the nodal load and displacements vectors for element i. The size of these vectors is equal to the total number of global dof for element i.
3
Denoting by {R} the reaction vector, and by {R } the corresponding displacements, we dene the unassembled structure exibility matrix as: (e) R = d(e) [0] F(e) R (7.3)
where d(e) is the unassembled global exibility matrix.

4
In its present form, Eq. 7.3 is of no help as the element forces F(e) and reactions {R} are not yet known.
7.2
Flexibility Matrix
5 We recall from Sect. 6.1.2 that we can automatically identify the redundant forces [F x ] and rewrite Eq. 7.3 as: (e) d00 0 [0] F0 = (7.4) (e) x Fx [0] dxx
Draft
7.2.2
12
7.2 Flexibility Matrix
135
Solution of Internal Forces and Reactions
The internal forces and reactions can in turn be obtained through Eq. 6.14: {F0 } = [C1 ] {P} + [C2 ] {Fx } (7.12)
which is combined with Eq. 7.11 to yield: {F0 } = [C1 ] [C2 ] [Dxx ]
1
[Dxp ] {P}
(7.13)
7.2.3
13
Solution of Joint Displacements
Joint displacements are in turn obtained by considering the top partition of Eq. 7.9: {p } = [Dpp ] [Dpx ] [Dxx ]
[D] 1
[Dxp ] {P}
(7.14)
This equation should again be compared with Eq. 5.4.
Example 7-1: Flexibility Method Solve for the internal forces and displacements of joint 2 of the truss in example 6.1.2. Let H = 0.75L and assign area A to members 3 and 5, and 0.5A to members 1, 2, and 4. Let f 5 be the redundant force, and use the [C1 ] and [C2 ] matrices previously derived. Solution: H C = L2L 2 = 0.8 and S = L2 +H 2 = 0.6 +H From Eq. 7.4 we obtain 0 x u1 v1 (e) 2 (e) 1 (e) 3 (e) 4 u4 v4 (e) 5 = d00 0
(e)
0 (e) dxx 0 0 2
F0 Fx Rx1 Ry1 f2 f1 f3 f4 Rx4 Ry4 f5
(7.15-a)
L AE
1.5 1.25 1.5 0 0 1.25
(7.15-b)
From Example 6.1.2 we have 1 0 S/C 1 1 0 0 1 [C1 ] = 1/C 0 S/C 0 0 0 S/C 0 Victor Saouma
1 0.75 1 0 1.25 0.75 0 0.75
0 1 0 1 0 0 0 0
C 0 C S [C2 ] = = 1 S C 0
0.8 0 0.8 0.6 1 0.6 0.8 0
(7.16)
Draft
Chapter 8
SPECIAL ANALYSIS PROCEDURES

To be edited 8.1
1
Semi-Rigid Beams
Often times, a beam does not have either a exible or rigid connection, but rather a semi-rigid one (such as in type II stell connection), Fig. 8.1
The stiness relationship for the beam element (with shear deformation) was previously derived in Eq. 2.42
2
M1 M2
= =
(4 + y )EIz 1 + (1 + y )L (2 y )EIz 1 + L(1 + y )
(2 y )EIz 2 L(1 + y ) (4 + y )EIz 2 L(1 + y )
(8.1-a) (8.1-b)
s s If we now account for the springs at both ends, Fig. 8.2, with stinesses k1 and k2 , then s s M1 = k1 (1 1 ) and M2 = k2 (2 2 )
(8.2)
The stiness relationship between M1 and M2 and the joint rotations 1 and 2 is obtained by eliminating 1 and 2 from above, yielding M1 = EI EI (S11 1 + S12 2 ) and M2 = (S21 1 + S22 2 ) L L
Flexible
(8.3)
Figure 8.1: Flexible, Rigid, and Semi-Rigid Beams
Rigid Semi-Rigid
Draft
Part II
Introduction to Finite Elements
Draft
Chapter 9
REVIEW OF ELASTICITY
9.1
1
Stress
A stress, Fig 9.1 is a second order cartesian tensor, ij where the 1st subscript (i) refers to the
X3
33 31 13 11 X2
12 32
23
21
X3
X2
22
X1
X1
Figure 9.1: Stress Components on an Innitesimal Element direction of outward facing normal, and the second one (j) to the direction of component force. 11 12 13 t1 t2 = ij = 21 22 23 = t3 31 32 33
(9.1)
2 In fact the nine rectangular components ij of turn out to be the three sets of three vector components (11 , 12 , 13 ), (21 , 22 , 23 ), (31 , 32 , 33 ) which correspond to the three tractions t1 , t2 and t3 which are acting on the x1 , x2 and x3 faces (It should be noted that those tractions are not necesarily normal to the faces, and they can be decomposed into a normal and shear traction if need be). In other words, stresses are nothing else than the components of tractions (stress vector), Fig. 9.2.
The state of stress at a point cannot be specied entirely by a single vector with three components; it requires the second-order tensor with all nine components.
Draft
9.2 Strain
151
X2
-t * 1 S 1
-t
n
S3
* tn S
O
C
h N
X1
-t 2 S2
X3
b
*
We seek to determine the traction (or stress vector) t passing through P and parallel to the plane ABC where A(4, 0, 0), B(0, 2, 0) and C(0, 0, 6). Solution: The vector normal to the plane can be found by taking the cross products of vectors AB and AC: N = = The unit normal of N is given by n= Hence the stress vector (traction) will be
3 7 6 7 2 7
V
*
Figure 9.3: Stress Traction Relations
ABAC =
12e1 + 24e2 + 8e3 6 2 3 e1 + e2 + e3 7 7 7 0 1 = 2
e1 4 4
e2 2 0
e3 0 6
(9.6-a) (9.6-b)
(9.7)
and thus t = 9 e1 + 5 e2 + 7 7
10 7 e3
7 5 0
5 3 1
9 7
5 7
10 7
(9.8)
9.2
7
Strain
Given the displacement ui of a point, the strain ij is dened as 1 (ui,j + uj,i ) 2 (9.9)
ij =
Victor Saouma
Draft
Chapter 10
VARIATIONAL AND ENERGY METHODS

10.1
10.1.1
1
Work, Energy & Potentials; Denitions

Introduction
Work is dened as the product of a force and displacement W dW

def b
F.ds
a
(10.1-a) (10.1-b)
Fx dx + Fy dy
Energy is a quantity representing the ability or capacity to perform work.
The change in energy is proportional to the amount of work performed. Since only the change of energy is involved, any datum can be used as a basis for measure of energy. Hence energy is neither created nor consumed.
4
The rst law of thermodynamics states The time-rate of change of the total energy (i.e., sum of the kinetic energy and the internal energy) is equal to the sum of the rate of work done by the external forces and the change of heat content per unit time:
d dt (K
+ U ) = We + H
(10.2)
where K is the kinetic energy, U the internal strain energy, W the external work, and H the heat input to the system.
5 For an adiabatic system (no heat exchange) and if loads are applied in a quasi static manner (no kinetic energy), the above relation simplies to:
We = U
(10.3)
Draft
12 13
10.1 Work, Energy & Potentials; Denitions
159
Considering uniaxial stresses, in the absence of initial strains and stresses, and for linear elastic systems, Eq. 10.9 reduces to U= 1 2 E d

(10.10)
When this relation is applied to various one dimensional structural elements it leads to
Axial Members: U = d 2
= P A P = AE d = Adx Torsional Members:
U=
1 2
P2 dx AE
(10.11)
U U xy xy d J Flexural Members:
= =
1 2 1 2 xy G Tr J
= = = rddrdx
r 2
xy Gxy d E d
xy
=
o 0
r2 ddr
U=
1 2
T2 dx GJ
(10.12)
1 2
x = z y = Mzz EI d = dAdx y 2 dA
A
E Mz y I
U=
1 2
M2 dx EIz
(10.13)
Iz
10.1.2.1
14
Internal Work versus Strain Energy
During strain increment, the work done by internal forces in a dierential element will be the negative of that performed by the stresses acting upon it. Wi = d d
(10.14)
15 If the strained elastic solid were permitted to slowly return to their unstrained state, then the solid would return the work performed by the external forces. This is due to the release of strain energy stored in the solid.
Victor Saouma
Draft
137
10.7 Variational Calculus; Preliminaries
197
136 It can be shown that in the principle of virtual displacements, the Euler equations are the equilibrium equations, whereas in the principle of virtual forces, they are the compatibility equations.
Euler equations are dierential equations which can not always be solved by exact methods. An alternative method consists in bypassing the Euler equations and go directly to the variational statement of the problem to the solution of the Euler equations.
138 Finite Element formulation are based on the weak form, whereas the formulation of Finite Dierences are based on the strong form.
139
We still have to dene . The rst variation of a functional expression is F = =

F F u u + u b F dx a
=
a
F F u + u u u
dx
(10.189)
As above, integration by parts of the second term yields

b
=
a
F d F u dx u
dx
(10.190)
140
We have just shown that nding the stationary value of by setting = 0 is equivalent to nding the extremal value of by setting d() equal to zero. d
=0
Similarly, it can be shown that as with second derivatives in calculus, the second variation 2 can be used to characterize the extremum as either a minimum or maximum.
141
10.7.2
142
Boundary Conditions
Revisiting the second part of Eq. 10.184, we had (x) Ess. F u

b
=0
a
(10.191)
Nat.
Boundary Cond. This can be achieved through the following combinations (a) = 0 (a) = 0 F (a) = 0 u F (a) = 0 u and and and and (b) = 0 F (b) = 0 u (b) = 0 F (b) = 0 u (10.192-a) (10.192-b) (10.192-c) (10.192-d)
143 Generalizing, for a problem with, one eld variable, in which the highest derivative in the governing dierential equation is of order 2m (or simply m in the corresponding functional), then we have
Essential (or forced, or geometric) boundary conditions, (because it was essential for the derivation of the Euler equation) if (a) or (b) =0. Essential boundary conditions, involve derivatives of order zero (the eld variable itself) through m-1. Trial displacement functions are explicitely required to satisfy this B.C. Mathematically, this corresponds to Dirichlet boundary-value problems. Victor Saouma Finite Element I; Framed Structures
Draft
10.7 Variational Calculus; Preliminaries Euler Equation: Natural Boundary Condition: EA du P =0 dx EA 2 at x = L

2
199
d dx
EA
du dx
=0
0<x<L
(10.196)
(10.197)
Solution II We have F (x, u, u ) = du dx
(10.198)
(note that since P is an applied load at the end of the member, it does not appear as part of F (x, u, u ) To evaluate the Euler Equation from Eq. 10.186, we evaluate F F =0 & = EAu u u Thus, substituting, we obtain d F F u dx u du EA dx = = 0 d (EAu ) = 0 Euler Equation dx (10.200-a) (10.200-b) (10.199-a)
0 B.C.
Example 10-13: Flexure of a Beam The total potential energy of a beam supporting a uniform load p is given by
L
=
0
1 M pw dx = 2
L 0
1 (EIw )w pw dx 2
F
(10.201)
Derive the rst variational of . Solution: Extending Eq. 10.189, and integrating by part twice
L L
=
0 L
F dx =
0
F F w + w dx w w
(10.202-a) (10.202-b) (10.202-c) (10.202-d)
=
0
(EIw w pw)dx
L L 0
(EIw w )|0
L
[(EIw ) w pw] dx
L L
= (EIw w )|0 [(EIw ) w]|0 + Nat. Or (EIw ) = p Ess. BC Nat. Ess.
[(EIw ) + p] wdx = 0
0
Euler Eq.
for all x
which is the governing dierential equation of beams and Victor Saouma Finite Element I; Framed Structures
Draft
10.8
10.8 Homework
201
Homework
q
subjected to a linearly varying vertical load of magnitude q = q0 (1 x ): l
1. By means of the principle of virtual displacements, reanalyze the previous problem using the following expression for both the virtual and actual systems: v = (1 cos and: (a) (b) (c) (d) Determine the vertical displacement of node 3. On the basis of your results, draw the shear and moment diagrams. Compute the the total potential energy of the system. Compare your results with those of problem 1 (exact solution) and Discuss. x 3x )a1 + (1 cos )a2 2L 2L (10.203)
Note: You should rst t the coecients a1 and a2 to the displacements of points 2 and 3, v2 and v3 . 2. Using the principle of virtual force, analyse the cantilivered beam (using a virtual unit point load), and: (a) Determine the vertical displacement of point 3. (b) Draw the shear and moment diagram. (c) Compute the total potential energy. 3. Analyse using the principle of minimum total potential energy. 4. Analyse this problem by the Rayleigh-Ritz method. 5. Analyse using the principle of minimum complementary total potential energy. 6. Analyse this problem using Castiglianos second theorem 7. Discuss your results You are strongly advised to use Mathematica.
Victor Saouma
t se te sest te sest te sest te sest veeeeee~v eeeveeeeeveeyv z eeveeevevevevevevevevevevevev testst te vv vv vv vv vv vv vv v vv vv vv vv v vv vv v vv v v v vw v v xx v v se vveeeeev eeeveeeeeve eevee veveveeveeveeveeve se ueeeeeue~veeeeueeeeeueeyv z eeueee|v } ueueueveueveueveueveuev est veeeeeveeeeveeeeeve eevee veveveeveeveeveeve t ueeeeeueeeeueeeeeue eeuee ueueueeueeueeueeue se eeeeeeeeeeeeeeeeyezuvuv eeeee|e}uv eeeuveeuveeuveeuveeuv est uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv{ uvuv uvuv uvuv ww xx eeeeeueeeeueeeeeueeeeuee ueueueeueeueeueeue uveeeeeveeeeveeeeeveeyezeeeveee|e}uv veveveuveveuveveuveveuveveuv eeeeeueeeeueeeeeueeeeuee ueueueeueeueeueeue uveeeeeveeeeveeeeeveeeeveee|e}uvuvuv veveveuveveuveveuveveuveveuv uvuv uvuv uvuv uvuv v uvuv uvuv uvuv v uvuv uvuv uvuv uvuv{{ v uvuv uvuv uvuv v uvuv v v vww v v xxx v v eeeeeueeeeueeeeeueeeeueeeueueueeueeueeueeue uveeeeeveeeeveeeeeveeeeveee|e}eveveveuveveuveveuveveuveveuv eeeeeueeeeueeeeeueeeeueeeueueueuveueuveueuveueuveueuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uvuv uv uuveeeeeueeeeueeeeeueeeeueeeueueueuveueuveueuveueuveueuv ueeeeeueeeeueeeeeueeeeueeeueueueeueeueeueeue eeeeeeeeeeeeeeeeeeeeeeeeuueeuueeuueeuueeuu uu uu uu uu uu uu uu uu uu uu uu uu uu uu uu uu
0
For the cantilivered beam shown below:
L/2
L/2
Draft
Chapter 11
INTERPOLATION FUNCTIONS
11.1 Introduction
1 Application of the Principle of Virtual Displacement requires an assumed displacement eld. This displacement eld can be approximated by interpolation functions written in terms of:
1. Unknown polynomial coecients, most appropriate for continuous systems, and the Rayleigh-Ritz method y = a 1 + a 2 x + a 3 x2 + a 4 x3 (11.1) A major drawback of this approach, is that the coecients have no physical meaning. 2. Unknown nodal deformations, most appropriate for discrete systems and Potential Energy based formulations (11.2) y = = N 1 1 + N2 2 + . . . + N n n
2
For simple problems both Eqn. 11.1 and Eqn. 11.2 can readily provide the exact solutions of the d4 q governing dierential equation (such as dxy = EI for exure), but for more complex ones, one must use 4 an approximate one.
11.2
Shape Functions
3 For an element (nite or otherwise), we can write an expression for the generalized displacement (translation/rotation), at any point in terms of all its known nodal ones, .
=
i=1
Ni (x)i = N(x) {}
(11.3)
where: 1. i is the (generalized) nodal displacement corresponding to d.o.f i 2. Ni is an interpolation function, or shape function which has the following characteristics: (a) Ni = 1 at node i (b) Ni = 0 at node j where i = j. 3. Summation of N at any point is equal to unity N = 1. 4. N can be derived on the bases of: (a) Assumed deformation state dened in terms of polynomial series. (b) Interpolation function (Lagrangian or Hermitian).
Draft
12
11.2 Shape Functions Substituting and rearranging those expressions into Eq. 11.5 we obtain u = = or: N1 = 1 x N2 = L
x L
205
u2 u1 )x + u1 L L x x (1 ) u1 + u2 L L
N1 N2
(11.8-a) (11.8-b)
(11.9)
11.2.2
13
Generalization
The previous derivation can be generalized by writing: u = a1 x + a2 = x 1

[p]
a1 a2
{a}
(11.10)
where [p] corresponds to the polynomial approximation, and {a} is the coecient vector.
14
We next apply the boundary conditions: u1 u2

{}
0 1 L 1
[L]
a1 a2
{a}
(11.11)
following inversion of [L], this leads to a1 a2

{a}
1 L
1 L
[L]1
1 0
u1 u2
{}
(11.12)
15
Substituting this last equation into Eq. 11.10, we obtain: u= (1

x L) x L
u1 u2
{}
(11.13)
[p][L]1
[N]
16
Hence, the shape functions [N] can be directly obtained from [N] = [p][L]1 (11.14)
11.2.3
17
Flexural
With reference to Fig. 11.2. We have 4 d.o.f.s, {}41 : and hence will need 4 shape functions, N1 to N4 , and those will be obtained through 4 boundary conditions. Therefore we need to assume a polynomial approximation for displacements of degree 3. v = = a 1 x3 + a 2 x2 + a 3 x + a 4 dv = 3a1 x2 + 2a2 x + a3 dx (11.15-a) (11.15-b)
Victor Saouma
Draft
Chapter 12
FINITE ELEMENT FORMULATION

1 Having introduced the virtual displacement method in chapter 10, the shape functions in chapter 11, and nally having reviewed the basic equations of elasticity in chapter 9, we shall present a general energy based formulation of the element stiness matrix in this chapter. 2 Whereas chapter 2 derived the stiness matrices of one dimensional rod elements, the approach used could not be generalized to general nite element. Alternatively, the derivation of this chapter will be applicable to both one dimensional rod elements or contnuum (2D or 3D) elements.
It is important to note that whereas the previously presented method to derive the stiness matrix yielded an exact solution, it can not be generalized to continuum (2D/3D elements). On the other hands, the method presented here is an approximate method, which happens to result in an exact stiness matrix for exural one dimensional elements. Despite its approximation, this so-called nite element method will yield excellent results if enough elements are used.
12.1
4
Strain Displacement Relations
The displacement at any point inside an element can be written in terms of the shape functions N and the nodal displacements {} (x) = N(x) {} (12.1) (12.2)
The strain is then dened as:
(x) = [B(x)]{}
def
where [B] is the matrix which relates nodal displacements to strain eld and is clearly expressed in terms of derivatives of N.
12.1.1
Axial Members
x ) L
N
u(x)
(1
N1
x L
N2
u1 u2
{}
(12.3-a)
Draft
10
12.3 Element Stiness Matrix Formulation Let us now apply the principle of virtual displacement and restate some known relations: U U = =
215
W {}d
(12.9-a) (12.9-b) (12.9-c) (12.9-d) (12.9-e) (12.9-f)
{}
{} = [D]{} [D]{0 } {} = [B]{} = [B]{} = [B]T
11 Combining Eqns. 12.9-a, 12.9-b, 12.9-c, 12.9-f, and 12.9-d, the internal virtual strain energy is given by:
[B]T [D][B]{} d
{ } { } T
[B]T [D]{0 } d
{ } { 0 } 0
(12.10)
=
12
[B] [D][B] d{}
[B] [D]{ }d
The virtual external work in turn is given by: W = {F} +

l
q(x)dx
(12.11)
Virt. Nodal Displ. Nodal Force
13
combining this equation with: {} = [N]{} (12.12)
yields:
l
W = {F} +
14
[N]T q(x) dx
0
(12.13)
Equating the internal strain energy Eqn. 12.10 with the external work Eqn. 12.13, we obtain:
[B]T [D][B] d{}

[k] U
[B]T [D]{0 }d =
{F }
0
(12.14)
{F} +
[N] q(x) dx
0 {F }
e
15 Cancelling out the term, this is the same equation of equilibrium as the one written earlier on. It relates the (unknown) nodal displacement , the structure stiness matrix [k], the external nodal
force vector F , the distributed element force

16
, and the vector of initial displacement.
From this relation we dene: [k] =
The element stiness matrix: [B]T [D][B]d (12.15)
Victor Saouma
Draft
Chapter 13
SOME FINITE ELEMENTS

13.1 Introduction
1 Having rst introduced the method of virtual displacements in Chapter 10, than the shape functions [N] (Chapter 11) which relate internal to external nodal displacements, than the basic equations of elasticity (Chapter 9) which dened the [D] matrix, and nally having applied the virtual displacement method to nite element in chapter 12, we now revisit some one dimensional element whose stiness matrix was earlier derived, and derive the stiness matrices of additional two dimensional nite elements.
13.2
2
Truss Element
The shape functions of the truss element were derived in Eq. 11.9: x N1 = 1 L x N2 = L The corresponding strain displacement relation [B] is given by: xx = = = du dx [ dN1 dx [
1 L [B]
dN2 dx 1 L ]
] (13.2-a)
For the truss element, the constitutive matrix [D] reduces to the scalar E; Hence, substituting into [B]T [D][B]d and with d = Adx for element with constant cross sectional area we
L
Eq. 12.15 [k] = obtain:
[k] = A
0
1 L 1 L L 0
E 1 1 1 1
1 L
1 L
dx
(13.3)
[k] = =
AE L
AE L2 1 1
1 1
dx (13.4)
Draft
Chapter 14
DYNAMIC ANALYSIS
14.1
1
Introduction
When the frequency of excitation of a structure is less than about a third of its lowest natural frequency of vibration, then we can neglect inertia eects and treat the problem as a quasi-static one.
2 If the structure is subjected to an impact load, than one must be primarily concerned with (stress) wave propagation. In such a problem, we often have high frequencies and the duration of the dynamic analysis is about the time it takes for the wave to travel across the structure.
If inertia forces are present, then we are confronted with a dynamic problem and can analyse it through any one of the following solution procedures: 1. Natural frequencies and mode shapes (only linear elastic systems) 2. Time history analysis through modal analysis (again linear elastic), or direct integration.
4 Methods of structural dynamics are essentially independent of nite element analysis as these methods presume that we already have the stiness, mass, and damping matrices. Those matrices may be obtained from a single degree of freedom system, from an idealization/simplication of a frame structure, or from a very complex nite element mesh. The time history analysis procedure remains the same.
14.2
Variational Formulation
5 In a general three-dimensional continuum, the equations of motion of an elementary volume without damping is LT + b = m u (14.1)
where m is the mass density matrix equal to I, and b operator L is 0 x 0 y 0 0 L= y x z 0 0 z

6
is the vector of body forces. The Dierential 0 0 z 0 x

y
(14.2)
For linear elastic material = De (14.3)
Draft
14.2 Variational Formulation Since this identity must hold for any admissible u, we conclude that
t Mu = fe fit t
225
(14.12)
Which represents the semi-discrete linear equation of motion in the explicit time integration in terms of the mass matrix M=
NT mNd
(14.13)
the vector of external forces

t fe =
NT bt d +

NT tt d
(14.14)
and the vector of internal forces fit = But since = D = DB, then fit = BT DBdu
BT t d
(14.15)
(14.16)
It should be noted that Eq. 14.13 denes the consistent mass matrix, which is a fully populated matrix. Alternatively, a lumped mass matrix can be dened, resulting only in a diagonal matrix. For an axial element, the lumped mass matrix is obtained by placing half of the total element mass as a particle at each node.
10 11
For linear elastic systems, and if damping was considered

t t t Mu + Cu + Ku = fe
(14.17)
the damping matrix matrix is often expressed in terms of the mass and stiness matrix, called Rayleigh damping, as C = M + K (14.18)
14.2.2
14.2.2.1
Implicit Time Integration

Linear Case
12 For implicit time integrate we consider the virtual work equation at time t + t (instead of t in the explicit method), thus Eq. 14.8 transforms into
uT md t+t + u

T t+t d =
uT tt+t d +
uT bt+t d
(14.19)
13
For linear problems t+t = De But+t (14.20)
Victor Saouma
Draft
20
14.3 Time Discretization
227
In practical nite element analysis, we are therefore mainly interested in a few eective methods, namely direct integration and mode superposition. These two techniques are closely related and the choice for one method or the other is determined only by their numerical eectiveness. In the present implementation the direct integration method is employed.
21 In the direct integration method Eq. ?? are integrated using a numerical step by step procedure, the term direct meaning that prior to the numerical integration, no transformation of the equations into a dierent form is carried out.
14.3.1
22
Explicit Time Integration
For the explicit time integration, we adopt at starting point a slight variation of the dierential equation previously derived, one which includes the eects of damping. Mu + Cu + Ku = fe
FI (t) FD (t) Fi (t) fe (t)
(14.29)
where M, C and K are the mass, damping and stiness matrices, fe is the external load vector and u, u and u are the acceleration, velocity and displacement vectors of the nite element assemblage.
23 An alternative way to consider this equation, is to examine the equation of statics at time t, where FI (t), FD (t), Fi (t), Fe (t) are the inertia, damping, internal and external elastic forces respectively.
14.3.1.1
24
Linear Systems
For the semi-discretized equation of motion, we adopt the central dierence method. This method is based on a nite dierence approximation of the time derivatives of displacement (velocity and acceleration). Assuming a linear change in displacement over each time step, for the velocity and the acceleration at time t: t u t u where t is the time step.
25
= =
ut+t utt 2t utt 2ut + ut+t t2
(14.30-a) (14.30-b)
t t t Substituting into Eq. 14.17 (Mu + Cu + Ke u = fe ) we obtain a0 a1 a2
1 1 1 1 2 t M+ C ut+t = fe K M u t 2 M C utt t2 t 2t t2 2t
a0 a1 M
t fe
(14.31)
from where we can solve for ut+t .

26
It should be noted that: 1. Solution is based on the equilibrium condition at time t 2. No factorization of the stiness matrix K is necessary (i.e. no K1 term appears in the equation). 3. To initiate the method, we should determine the displacements, velocity and acceleration at time t.
27
Algorithm: Finite Element I; Framed Structures
Victor Saouma
Draft
32
14.3 Time Discretization
229
31 Since no structural stiness matrix needs to be assembled, very large problems can be eectively solved. This includes quasi-static problems (linear and non-linear) where the load is applied incrementally in time. Codes duch as DYNA or ABAQUS/EXPLICIT exploit this feature to perform complex analyses such as metal forming, penetration, crash worthiness, etc...
A major disadvantage of the explicit method is that it converges only if t < tcr = 2 max (14.41)
where max is the highest natural frequency (or smallest wave length) of the structure. Recall that = 2 T (14.42)
If this condition is violated, numerical instability occurs.

33
Hence, an eigenvalue analysis of the complete system must rst be undertaken. But since the global structural stiness matrix is not necessarily assembled, an upper bound estimate of max is obtained from (e) max < max (14.43)
(e)
where max is the maximum frequency of a single nite element (usually the smallest one). It can in turn be determined from e (14.44) det(K(e) (max )2 Me ) = 0 in nonlinear problems, K is the tangent stiness. Note that the determination of the eigenvalues (which correspond to 2 ) of the simple problem without damping K 2 M can be solved by either expanding the determinant of the resulting matrix, or simply determining the eigenvalues of M1 K.
34 35
Alternatively,
e max =
2c L
(14.45)
where c is the acoustic wave speed c = E/ and L is a representative length of element (e). Essentially, this means that t must be small enough so that information does not propagate across more than one element per time step. Example 14-1: MATLAB Code for Explicit Time Integration %================================================================ %I N I T I A L I Z A T I O N %================================================================ % Initialize the matrices M=[2 0;0 1]; K=[6 -2;-2 4];C=[0 0;0 0];P=[0 ; 10]; % determine the eigenvalues, minimum frequency, and delta t critical omega2=eig(inv(M)*K); Tcrit=min(2*pi./sqrt(omega2)); Delta_t=Tcrit/10; % Initialize the displacement and velocity vectors at time = 0; u_t(1:2,:)=0; du_t(1:2,:)=0; Phat_t(1:2,:)=0; % solve for the initial acceleration at time 0 ddu_t(1:2,:)=inv(M)*(P- K*u_t); Victor Saouma Finite Element I; Framed Structures
Draft
Chapter 15
GEOMETRIC NONLINEARITY
15.1
1
Introduction
First order elastic analysis
With refernce to Fig. 15.1, we distinguish dierent levels of analysis:

Stiffening
Load
Bifurcation
Bifurcation Softening
Elastic Critical Load Elastic Stability Limit

Second order elastic analysis
Bifurcation
Inelastic Critical Load Plastic Limit Load Inelastic Stability Limit
First order inelastic analysis Second order inelastic analysis
Displacement
Figure 15.1: Level of Analysis First Level elastic which excludes anly nonlinearities. This is usually acceptable for service loads. Elastic Critical load is usually determined from an eigenvalue analysis resulting in the buckling load. Second-order elastic accounts for the eects of nite deformation and displacements, equilibrium equations are written in terms of the geometry of the deformed shape, does not account for material non-linearilties, may be able to detect bifurcation and or increased stiness (when a member is subjected to a tensile axial load). First-order inelastic equilibrium equations written in terms of the geometry of the undeformed structure, accounts for material non-linearity. Second-order inelastic equations of equilibrium written in terms of the geometry of the deformed shape, can account for both geometric and material nonlinearities. Most suitable to determine failure or ultimate loads.
Draft
15.1 Introduction Thus buckling will occur if

P EI
237 =
n 2 L
or P = n2 2 EI L2
10 The fundamental buckling mode, i.e. a single curvature deection, will occur for n = 1; Thus Euler critical load for a pinned column is
Pcr =
2 EI L2
(15.7)
11
The corresponding critical stress is cr = 2 E

L 2 r
(15.8)
where I = Ar 2 .
12
Note that buckling will take place with respect to the weakest of the two axis. Higher Order Dierential Equation; Essential and Natural B.C.
15.1.1.2
13 In the preceding approach, the buckling loads were obtained for a column with specied boundary conditons. A second order dierential equation, valid specically for the member being analyzed was used.
14
In the next approach, we derive a single fourth order equation which will be applicable to any column regardelss of the boundary conditions.
15
Considering a beam-column subjected to axial and shear forces as well as a moment, Fig. 15.3, taking the moment about i for the beam segment and assuming the angle dv between the axis of the beam and dx
the horizontal axis is small, leads to M M+

16
Neglecting the terms in dx2 which are small, and then dierentiating each term with respect to x, we obtain d2 M dV d2 v P 2 =0 (15.10) dx2 dx dx However, considering equilibrium in the y direction gives dV = w dx (15.11) Finite Element I; Framed Structures
17
Victor Saouma
y, u
dx
Figure 15.3: Simply Supported Beam Column; Dierential Segment; Eect of Axial Force P
w(x)
w P M x P
v x
i
V+ V dx x dx P
P M+ M dx x
i j
dM (dx)2 dV dx + w + V + dx 2 dx
dx P
dv dx
dx = 0
(15.9)
Draft
obtain and
15.3 Elastic Instability; Bifurcation Analysis
245
Ke =
EI L3
AL2 I AL2 I
0 0 0
4 2 AL I 2 2 AL I 0 0 0 4 0 0 0 0 0
3 0 0 4L2 6L 2L2 5 0 0
L 10 12 5
5 0 0 6L 24 0 6 0 0
6 0 0 2L2 0 8L2
(15.54)
4. Noting that in this case K = Kg for P = 1, the determinant |Ke + K | = 0 leads to g g 1 2 1 AL I 2 4 AL I 3 0 5 0 6 0 5. Introducing =
AL2 I
1 0 0 P 0 Kg = L 0 0
3 0 0 2 2 15 L
L 10 L2 30
L2 30
(15.55)
0 4 2 15 L
4 2 AL I 2 2 AL I 0 0 0
3 0 0 4 2 4L2 15 L EI 3 1 6L + 10 L EI 4 1 2L2 + 30 L EI
5 0 0 3 1 6L + 10 L EI 2 24 12 L 5 EI 0
6 0 0 4 1 2L2 + 30 L = 0 EI 0 4 4 2 8L 15 L EI
(15.56)
and =
L2 EI ,
the determinant becomes 3 0 0 2 2 15 6L + 10 2 + 30 5 0 0 6L + 12 2 0 6 0 0 2 + 30 0 4 2 15
1 4 1 4 2 3 0 0 5 0 0 6 0 0
10 5
=0
(15.57)
6. Expanding the determinant, we obtain the cubic equation in 33 2202 + 3, 840 14, 400 = 0 and the lowest root of this equation is = 5.1772 . 7. We note that from Eq. 15.21, the exact solution for a column of length L was Pcr = (4.4934)2 (4.4934)2 EI = EI = 5.0477 EI L2 l2 (2L)2 (15.59) (15.58)
and thus, the numerical value is about 2.6 percent higher than the exact one.
51
The mathematica code for this operation is:

*) -e a/l 0 0 e a/l 0 0 , , , , , , 0 -12 e i/l^3 -6 e i/l^2 0 12 e i/l^3 -6 e i/l^2 , , , , , , 0 6 e i/l^2 2 e i/l 0 -6 e i/l^2 4 e i/l }, }, }, }, }, }
(* Define elastic stiffness matrices ke[e_,a_,l_,i_]:={ {e a/l , 0 , 0 , {0 , 12 e i/l^3 , 6 e i/l^2 , {0 , 6 e i/l^2 , 4 e i/l , {-e a/l , 0 , 0 , { 0 , -12 e i/l^3 , -6 e i/l^2 , { 0 , 6e i/l^2 , 2 e i/l ,
Victor Saouma
Draft
15.3 Elastic Instability; Bifurcation Analysis The element stiness matrices are given by u1 20 1, 208 2 1, 208 96, 667 0 0 0
247
k1 e
(15.60-a)
k1 g
u1 0.01 0.10 P 0
k2 e
2 128, 890 64, 440
2 0.10 16.00
k3 e
k3 g
The global equilibrium relation can now be written as
u1 0.01667 0.1 P
u1 47 1, 678
3 1, 678 80, 556
0 3 64, 440 128, 890 0 0
(15.60-b)
(15.60-c)
(15.60-d)
3 0 0 0.1 9.6
(15.60-e)
(Ke P Kg ) = 0 u1 (66.75) P (0.026666) (1, 208.33) P (0.1) (1, 678.24) P (0.1) 2 (1, 208.33) P (0.1) (225, 556.) P (16.) (64, 444.) P (0) 3 (1, 678.24) P (0.1) (64, 444.4) P (0) = 0 (209, 444.) P (9.6)
(15.61)
(15.62)
The smallest buckling load amplication factor is thus equal to 2, 017 kips.
(* Initialize constants *) a1=0 a2=0 a3=0 i1=100 i2=200 i3=50 l1=10 12 l2=15 12 l3=6 12 e1=29000 e2=e1 e3=e1 (* Define elastic stiffness matrices *) ke[e_,a_,l_,i_]:={ {e a/l , 0 , 0 , -e a/l {0 , 12 e i/l^3 , 6 e i/l^2 , 0 {0 , 6 e i/l^2 , 4 e i/l , 0 {-e a/l , 0 , 0 , e a/l
, , , ,
0 -12 e i/l^3 -6 e i/l^2 0
, , , ,
0 6 e i/l^2 2 e i/l 0
}, }, }, },
Victor Saouma
Draft
15.5 Summary
255
STRONG FORM
WEAK FORM
? 2nd Order D.E. 2 B.C.
? 4th Order D.E. 4 B.C.
? du y d2 v + x = dx dx2
1 2
dv dx
? d2 y P = 0 EI dx2 v = A sin kx B cos kx ? d4 v P d2 v = w EI 4 dx dx2 v = C1 sin kx + C2 cos kx + C3 x + C4
? U=
1 2
E2 d
? K= Ke Kg
? P = (Ke + Kg )u
? |Ke + Kg | = 0
Figure 15.5: Summary of Stability Solutions
Victor Saouma
Draft
Chapter 16
REFERENCES
Basic Structural Analysis : 1. Arbabi, F., Structural Analysis and Behavior, McGraw-Hill, Inc., 1991 2. Beaufait, F.W., Basic Concepts of Structural Analysis, Prentice-Hall Inc., Englewood Clis, N.J., 1977 3. Chajes, A., Structural Analysis, Prentice-Hall, Inc., Englewood Clis, N.J., 1983 4. Gerstle, K.H., Basic Structural Analysis, (Local Reprint 1984. 5. Ghali, A., and Neville, A.M., Structural Analysis, Chapan and Hall, London, 1978 6. Gutowski, R.M., Structures: Fundamental Theory and Behavior, Van Nostrand Reinhold Co., N.Y., 1984 7. Hsieh, Y.Y., Elementary Theory of Structures, 2nd Ed., Prentice Hall, Inc., Englewood Clis, N.J., 1982 8. Laursen, H.I., Structural Analysis, 2nd Ed., McGraw-Hill, N.Y., 1978 9. Morris, J.C., Wilbur, S., and Utku, S., Elementary Structural Analysis, McGraw-Hill, N.Y., 1976 10. Wang, C.K., Intermediate Structural Analysis, McGraw-Hill, N.Y., 1983 Matrix Analysis : 1. Argyris, J.H., Recent Advances in Matrix Methods of Structural Analysis, Pergamon Press, Oxford, 1964 2. Beaufait, F.W., Rowan Jr., W.H., Hoadley, P.G., and Hackett, R.M., Computer Methods of Structural Analysis, 4th Edition, 1982 3. Bhatt, P., Programming the Matrix Analysis of Skeletal Structures, Halsted Press, 1986 4. Elias, Z.M., Theory and Methods of Structural Analysis, John Wiley & Sons, 1986 5. Holzer, S.M., Computer Analysis of Structures. Elsevier, 1985 6. Livesley, R., Matrix Methods of Structural Analysis, Pergamon Press, Oxford, 964 7. Martin, H.C., Introduction to Matrix Methods of Structural Analysis, McGraw-Hill, N.Y., 1966 8. McGuire, W., and Gallagher, R.H., Matrix Structural Analysis, John Wiley and Sons Inc., N.Y., 1979 9. Meek, J.L., Matrix Structural Analysis, McGraw-Hill, N.Y., 1971 10. Meyers, V.J., Matrix Analysis of Structures, Harper and Row, Publ., N.Y., 1983 11. Przemieniecki, J.S., Theory of Matrix Structural Analaysis, McGraw-Hill, N.Y., 1968 12. Weaver Jr, W., and Gere, J.M., Matrix Analysis of Framed Structures, 2nd Ed., Van Nostrand Co., N.Y., 1980 Introduction to Finite Element and Programming : 1. Bathe, K.J., Finite Element Procedures in Engineering Analysis, Prentice-Hall, Inc., Englewood Clis, N.J., 1982
Draft
Appendix A
REVIEW of MATRIX ALGEBRA

Because of the discretization of the structure into a nite number of nodes, its solution will always lead to a matrix formulation. This matrix representation will be exploited by the computer ability to operate on vectors and matrices. Hence, it is essential that we do get a thorough understanding of basic concepts of matrix algebra.
A.1
Matrix:
Denitions
[A] = Ai1 . . . Am1 A11 A21 . . . A12 A22 . . . Ai2 . . . Am2 ... ... .. . ... .. . ... A1j A2j . . . Aij . . . Amj ... ... .. . ... .. . ... A1n A2n . . . Ain . . . Amn (A.1)
We would indicate the size of the matrix as [A]mn , and refer to an individual term of the matrix as Aij . Note that matrices, and vectors are usually boldfaced when typeset, or with a tilde when handwritten A. Vectors: are one column matrices: B1 B2 . . . {X} = Bi . . . Bm B2 ...
(A.2)
A row vector would be C =
B1
Bi
...
Bm
(A.3)
Note that scalars, vectors, and matrices are tensors of order 0, 1, and 2 respectively. Square matrix: are matrices with equal number of rows and columns. [A]mm Symmetry: Aij = Aji Identity matrix: is a square matrix with all its entries equal to zero except the diagonal terms which are equal to one. It is often denoted as [I], and Iij = 0, 1, if i = j if i = j (A.4)
Draft
A.3 Determinants Scalar Multiplication: [B] Bij = k [A] = kAij
261
(1.11-a)
Matrix Multiplication: of two matrices is possible if the number of columns of the rst one is equal to the number of rows of the second. [A]mn Aij = = [B]mp [C]pn Bi
p 1p 11
(1.12-a)
{Cj }p1 (1.12-b) (1.12-c)
=
r=1
Bir Crj
Some important properties of matrix products include: Associative: [A]([B][C]) = ([A][B])[C] Distributive: [A]([B] + [C]) = [A][B] + [A][C] Non-Commutativity: [A][B] = [B][A]
A.3
Determinants
n
The Determinant of a matrix [A]nn , denoted as det A or |A|, is recursively dened as det A =
j=1
(1)1+j a1j det A1j
(A.13)
Where A1j is the (n 1)x(n 1) matrix obtained by eliminating the ith row and the jth column of matrix A. For a 2 2 matrix a11 a12 = a11 a22 a12 a21 (A.14) a21 a22 For a 3 3 matrix a11 a21 a31 a12 a22 a32 a13 a23 a33 = a11 a22 a32 a23 a33 a12 a21 a31 a23 a33 + a13 a21 a31 a22 a32 (1.15-a) (1.15-b) (1.15-c) (1.15-d) (1.15-e)
= a11 (a22 a33 a32 a23 ) a12 (a21 a33 a31 a23 ) +a13 (a21 a32 a31 a22 ) = a11 a22 a33 a11 a32 a23 a12 a21 a33 + a12 a31 a23 +a13 a21 a32 a13 a31 a22
Can you write a computer program to compute the determinant of an n n matrix? We note that an n n matrix would have a determinant which contains n! terms each one involving n multiplications. Hence if n = 10 there would be 10! = 3, 628, 800 terms, each one involving 9 multiplications hence over 30 million oating operations should be performed in order to evaluate the determinant. This is why it is impractical to use Cramers rule to solve a system of linear equations. Some important properties of deteminants: Victor Saouma Finite Element I; Framed Structures
Draft
A.6
A.6 Eigenvalues and Eigenvectors
263
Eigenvalues and Eigenvectors

A1n A2n . . . Ann x1 x2 . . . xn B1 B2 = . . . Bn
A special form of the system of linear equation A11 A12 . . . A21 A22 . . . [A] = . . .. . . . . . Ai1 Ai2 . . .
(1.21)
is one in which the right hand side is a multiple of the solution: [A] {x} = {x}
(1.22) (1.23)
which can be rewritten as [A I] {x} = 0 A nontrivial solution to this system of equations is possible if and only if [A I] is singular or |A I| = 0 or [A] = A11 A21 . . . Ai1 A12 ... A22 . . . . .. . . . Ai2 ... A1n A2n . . . Ann (1.24)
=0
(1.25)
When the determinant is expanded, we obtain an nth order polynomial in terms of which is known as the characteristic equation of [A]. The n solutions (which can be real or complex) are the eigenvalues of [A], and each one of them i satises [A] {xi } = i {xi } where {xi } is a corresponding eigenvector. It can be shown that: 1. The n eigenvalues of real symmetric matrices of rank n are all real. 2. The eigenvectors are orthogonal and form an orthogonal basis in En . Eigenvalues and eigenvectors are used in stability (buckling) analysis, dynamic analysis, and to assess the performance of nite element formulations. (1.26)
Victor Saouma
Draft
Appendix B
SOLUTIONS OF LINEAR EQUATIONS

Note this chapter is incomplete
B.1
Introduction
1 Given a system of linear equations [A] nn {x} = {b} (which may result from the direct stiness method), we seek to solve for {x}. Symbolically this operation is represented by: {x} = [A] 1 {b} 2
There are two approaches for this operation:
Direct inversion using Cramers rule where [A]1 = [adjA] . However, this approach is computationally [A] very inecient for n 3 as it requires evaluation of n high order determinants. Decomposition: where in the most general case we seek to decompose [A] into [A] = [L][D][U] and where: [L] lower triangle matrix [D] diagonal matrix [U] upper triangle matrix There are two classes of solutions
Direct Method: characterized by known, nite number of operations required to achieve the decomposition yielding exact results. Indirect methods: or iterative decomposition technique, with no a-priori knowledge of the number of operations required yielding an aapproximate solution with user dened level of accuracy.
B.2
B.2.1
3
Direct Methods
Gauss, and Gaus-Jordan Elimination
Given [A]{x} = {b}, we seek to transform this equation into
2. Gauss-Jordan Elimination: is similar to the Gaus Elimination, however tather than transforming the [A] matrix into an upper diagonal one, we transform [A|I] into [I|A1 ]. Thus no backsubstitution is needed and the matrix inverse can be explicitely obtained.
1. Gaus Elimination: [U]{x} = {y} where [U]is an upper triangle, and then backsubstitute from the bottom up to solve for the unknowns. Note that in this case we operate on both [A] & {b}, yielding {x}.
Draft
B.2 Direct Methods 2. Elimination of the rst column: (a) row 1=0.1(row 1) (b) row 2=(row2)+20(new row 1) (c) row 3=(row 3) -5(new row 1) 1 0.1 0 5 0 2.5 0.5 10 7.5 0.1 2 0.5 0 0 0.1 1 0 4 0 1 5.5
267
(2.6-a)
3. Elimination of second column (a) row 2=0.2(row 2) (b) row 1=(row 1)-0.1(new row 2) (c) row 3=(row 3) -2.5(new row 2) 1 0 0 1 0 0 0.7 2 2.5 0.06 0.4 1.5
0.02 0.2 0.5
4. Elimination of the third column (a) row 3=0.4(row 3) (b) row 1=(row 1)+0.7(new row 3) (c) row 2=(row 2)-2(new row 3) 1 0 0 1 0 0 This last equation is [I|A1 ] B.2.1.1
4
0 0.02 0.8 0 1 3.5
(2.7-a)
0 0.36 0 1.6 1 0.6
0.16 0.6 0.2
0.28 1 2 0.8 1.4 0.4

{x}
(2.8-a)
Algorithm
Based on the preceding numerical examples, we dene a two step algorithm for the Gaussian ellimination.
5
Dening ak to be the coecient of the ith row & j th column at the k th reduction step with i k & ij j k: Reduction: ak+1 ik ak+1 ij bk+1 ij Backsubstitution: xij =
bi ij
=0 = = ak ij bk ij
ak ak ik kj ak kk k k aik bkj ak kk
k<in
k < i n; k < j n k < i n; 1 < j m

n k=i+1 ai ii
(2.9)
ai xkj ik
(2.10)
Note that Gauss-Jordan produces both the solution of the equations as well as the inverse of the original matrix. However, if the inverse is not desired it requires three times (N 3 ) more operations than Gauss 3 or LU decomposition ( N ). 3 Victor Saouma Finite Element I; Framed Structures
Draft
Note:
B.2 Direct Methods 4. Take row by row or column by column

aij
269
lij =
uij = aij lii = 1
lik ukj k=1 ujj i1 lik ukj k=1
j1
i>j ij
(2.16)
1. Computed elements lij or uij may always overwrite corresponding element aij 2. If [A] is symmetric [L]T = [U], symmetry is destroyed in [A]F For symmetric matrices, LU decomposition reduces to: uij = aij lii = 1 uji lij = ujj
i1 k=1
lik ukj
ij
(2.17)
Example B-3: Example Given: 7 4 A= 1 3 Solution: Following the above procedure, it can be decomposed into: Row 2: Row 1: u11 = a11 = 7; u12 = a12 = 9; u13 = a13 = 1; u14 = a14 = 2 l21 u22 u23 u24 Row 3: l31 l32 u33 u34 Row 4: l41 l42 l43 u44
a = u41 11 41 = a42 l22 u12 u a43 l41 u13 l42 u23 = u33 = a44 l41 u14 l42 u24 l43 u34 3 =7 = 2(0.4286)(9) 10.1429 = 1(0.4286)(1)(0.5775)(2.5714) 1.6622 = 5 (0.4286)(2) (0.5775)(8.1429) (1.2371)(8.0698) a = u31 11 31 = a32 l22 u12 u = a33 l31 u13 l32 u23 = a34 l31 u14 l32 u24 a = u21 11 = a22 l21 u12 = a23 l21 u13 = a24 l21 u14 4 =7 9 = 5 4 7 1 = 2 + 47 2 = 7 4 7
9 5 6 2
1 2 3 1
2 7 4 5
(2.18-a)
= 10.1429 = 2.5714 = 10.1429
=1 7 = 6(0.1429)(9) 10.1429 = 3 (0.1429)(1) (0.4647)(2.5714) = 4 (0.1429)(2) (0.4647)(8.1429)
= 0.4647 = 1.6622 = 8.0698
= 0.5775 = 1.2371 = 8.8285
Victor Saouma
Draft
Given: Solution: Column 1:
B.3 Indirect Methods
271
4 6 A= 10 4
6 13 13 6
10 13 27 2
4 6 2 72
(2.28-a)
l11 l21 l31 l41 Column 2: l22 l32 l42 Column 3: l33 l43 Column 4: l44 = = = = = =
= = = =
a21 l11 a31 l11 a41 l11
a11
= = = =
6 2 10 2 4 2
=2 =3 =5 =2
a32 l31 l21 l22 a42 l41 l21 l22
2 a22 l21
= = =
13 32
13(5)(3) 2 6(2)(3) 2
=2 = 1 =0
2 2 a33 l31 l32 a43 l41 l31 l42 l32 l33
= =
2(2)(5)(0)(1) 1
27 52 (1)2
=1 = 8
2 2 2 a44 l41 l42 l43
72 (2)2 (0)2 (8)2
=2
or 2 3 5 2 0 2 1 0
[L]
0 0 1 8
0 2 0 0 0 0 2 0
[A]
3 2 0 0
5 1 1 0
[U]
2 0 8 2
(2.33-a)
B.2.4
Pivoting
B.3
16
Indirect Methods
Iterative methods are most suited for 1. Very large systems of equation n > 10, or 100,000 2. systems with a known guess of the solution
17
The most popular method is the Gauss Seidel. Finite Element I; Framed Structures
Victor Saouma
Draft
B.4.2
22
B.4 Ill Conditioning
273
Pre Conditioning
If a matrix [K] has an unacceptably high condition number, it can be preconditionedthrough a congruent operation: [K ] = [D1 ][K][D2 ] (2.41)
However there are no general rules for selecting [D1 ] and [D2 ].
B.4.3
Residual and Iterative Improvements
Victor Saouma
Draft
Appendix C
TENSOR NOTATION
NEEDS SOME EDITING
1
Equations of elasticity are expressed in terms of tensors, where A tensor is a physical quantity, independent of any particular coordinate system yet specied most conveniently by referring to an appropriate system of coordinates. A tensor is classied by the rank or order A Tensor of order zero is specied in any coordinate system by one coordinate and is a scalar.
A tensor of order one has three coordinate components in space, hence it is a vector. For example, force and a stress are tensors of order 1 and 2 respectively.
In general 3-D space the number of components of a tensor is 3n where n is the order of the tensor.
3 To express tensors, there are three distinct notations which can be used: 1) Engineering; 2) indicial; or 3) Dyadic. 4 Whereas the Engineering notation may be the simplest and most intuitive one, it often leads to long and repetitive equations. Alternatively, the tensor and the dyadic form will lead to shorter and more compact forms.
C.1
Engineering Notation
In the engineering notation, we carry on the various subscript(s) associated with each coordinate axis, for example xx , xy .
C.2
5
Dyadic/Vector Notation
Uses bold face characters for tensors of order one and higher, , . This notation is independent of coordinate systems. Since scalar operations are in general not applicable to vectors, we dene A+B AB A AB = B+A (3.1-a) (3.1-b) (3.1-c) (3.1-d)
= BA = Ax i + A y j + Az k = |A||B| cos(A, B) = A x Bx + A y By + A z Bz
Draft
Similarly:
C.3 Indicial/Tensorial Notation
277
this simple compacted equation (expressed as x = cz in dyadic notation), when expanded would yield: x1 x2 x3 = = = c11 z1 + c12 z2 + c13 z3 c21 z1 + c22 z2 + c23 z3 c31 z1 + c32 z2 + c33 z3 (3.9)
(3.8-a)
Aij = Bip Cjq Dpq
A11 A12 A21 A22
= B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22 = B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22 = B21 C11 D11 + B21 C12 D12 + B22 C11 D21 + B22 C12 D22 = B21 C21 D11 + B21 C22 D12 + B22 C21 D21 + B22 C22 D22 (3.10-a)
Victor Saouma
Draft
Appendix D
INTEGRAL THEOREMS
1
Some useful integral theorems are presented here without proofs. Scheys textbook div grad curl and all that provides an excellent informal presentation of related material.
D.1
Integration by Parts
The integration by part formula is

b a b b
u(x)v (x)dx = u(x)v(x)|a
v(x)u (x)dx
a
(4.1)
or
b a
udv = uv|a
vdu
a
(4.2)
D.2
Green-Gradient Theorem
(Rdx + Sdy) =
Greens theorem is R S x y dxdy (4.3)
D.3
2
Gauss-Divergence Theorem
The general form of the Gauss integral theorem is (4.4)
v.nd =

divvd
or vi ni d =

vi,i d
(4.5)

Draft: Finite Element I

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Draft: Finite Element I

Uploaded by

Copyright:

Available Formats

Draft

Lecture Notes in:

Matrix Structural Analysis of Framed Structures

5.6.2.11 5.6.2.12 5.6.2.13 5.6.2.14 Homework . . .

Finite Element I; Framed Structures

Finite Element I; Framed Structures

Finite Element I; Framed Structures

Why Matrix Structural Analysis?

1.3 Structural Idealization Group No. 1 2 3 Element Type 1 2 1 Material Group 1 1 2

Table 1.3: Example of Group Number

Y X 3D TRUSS GRID & FRAME

Figure 1.1: Global Coordinate System

We should dierentiate between 2 coordinate systems:

Two dimensional structures will be dened in the X-Y plane.

1.4 Degrees of Freedom

Fy1 , Mz2 v1 , 2 Fx1

{p} {} {p} {} {p} {}

u1 Fx1 , Fy2 , Mz3 u 1 , v2 , 3 Tx1 , Fy2 , Mz3 1 , v2 , 3 Fx1 ,

4 , v5 , 6 3 Dimensional Fx2 u2 Fx7 , Fy8 , Fy9 , Tx10 My11 , Mz12 u 7 , v8 , w 9 , 10 , 11 12

u1 , Fx1 , Fy2 , Fy3 , Tx4 My5 , Mz6 u 1 , v2 , w 3 , 4 , 5 6

Table 1.4: Degrees of Freedom of Dierent Structure Types Systems

1.5 Course Organization

Finite Element I; Framed Structures

Matrix Structural Analysis of Framed Structures

ELEMENT STIFFNESS MATRIX

Unit Action Load Load Load Load Displacement

Eect on Shear Moment Deection Displacement Load

Table 2.1: Examples of Inuence Coecients

2.4 Stiness Coecients

Virtual Force: x x dvol M xy x = I My x U = x = E = EI y 2 dA = I dvol = dAdx W = P U = W U = l M dx M EI 0

Similarly, we would obtain:

The exibility method will be covered in more detailed, in chapter 7.

2.5 Force-Displacement Relations

Figure 2.2: Denition of Element Stiness Coecients

Finite Element I; Framed Structures

u2y 0 12EIz L3 6EIz L2 0

2z 0 6EIz L2 2EIz L 0 6EIz L2 4EIz

v1 0 b k11 0 0 0 b k21 0 b k13 0 0 0 b k12

w1 0 0 b k11 0 b k32 0 0 0 b k13 0 b k12 0

y1 0 0 b k12 0 b k22 0 0 0 b k14 0 b k24 0

z1 0 b k12 0 0 0 b k22 0 b k14 0 0 0 b k24

v2 0 b k13 0 0 0 b k12 0 b k33 0 0 0 b k43

w2 0 0 b k13 0 b k12 0 0 0 b k33 0 b k43 0

y2 0 0 b k14 0 b k24 0 0 0 b k34 0 b k44 0

For [k3D ] and with we obtain: 11

z2 0 k b 14 0 0 0 b k24 0 b k34 0 0 0 b k44

0 Vz1 Tx1 0 0 My1 0 Mz1 Px2 EA L 0 Vy2 0 Vz2 Tx2 0 0 My2

0 EIy 6 L2 0 2EIy L 0 0 0 6EIy L2 0 4EIy L 0

Remarks on Element Stiness Matrices

Finite Element I; Framed Structures

STIFFNESS METHOD; Part I: ORTHOGONAL STRUCTURES

The Stiness Method

Note that we use the same matrix one).

premultiplying by [(e) ]1 {P} = [(e) ]1 [k(e) ][(e) ]{}

l11 l21 + l12 l22 + l13 l23

By direct multiplication of [] and [] it can be shown that: [] [] = [I] [] = [] an orthogonal matrix.

The reverse transformation (from local to global) would be {V} = [] {v}

lxX VX VY = lxY lxZ VZ

lyX lyY lyZ

lzX Vx Vy lxY Vz lzZ

Finite Element I; Framed Structures

4.2 Transformation Matrices For Framework Elements