V, Saouma FEM I PDF

Draft
DRAFT
Lecture Notes in:
FINITE ELEMENT I
Framed Structures
CVEN4525/5525
c VICTOR
E. SAOUMA,
Fall 2002
Dept. of Civil Environmental and Architectural Engineering

University of Colorado, Boulder, CO 80309-0428
Draft
02
Blank page
Victor Saouma
Finite Element I; Framed Structures
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 . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Matrix Structural Analysis of Framed Structures
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 . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
11
12
13
14
15
16
16
19
111
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
. 21
. 21
. 22
. 23
. 24
. 24
. 24
. 27
. 29
. 211
. 212
. 212
. 213
. 213
. 215
. 215
. 216
. 217
. 218
3 STIFFNESS METHOD; Part I: ORTHOGONAL STRUCTURES

31
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 The Stiness Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Draft
02
3.3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 33
. 33
. 35
. 38
. 311
. 312
. . . . . .
. . . . . .
. . . . . .
Elements
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
. 41
. 41
. 42
. 44
. 45
. 45
. 45
. 46
. 47
. 411
5 STIFFNESS METHOD; Part II

5.1 Direct Stiness Method . . . . . . . . . . . . . . . . . . . . .
5.1.1 Global Stiness Matrix . . . . . . . . . . . . . . . . .
5.1.1.1 Structural Stiness Matrix . . . . . . . . . .
5.1.1.2 Augmented Stiness Matrix . . . . . . . . .
5.2 Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 Boundary Conditions, [ID] Matrix . . . . . . . . . . .
5.2.2 LM Vector . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.3 Assembly of Global Stiness Matrix . . . . . . . . . .
E 5-1 Assembly of the Global Stiness Matrix . . . . . . . .
5.2.4 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . .
E 5-2 Direct Stiness Analysis of a Truss . . . . . . . . . . .
E 5-3 Analysis of a Frame with MATLAB . . . . . . . . . .
E 5-4 Analysis of a simple Beam with Initial Displacements
5.3 Skyline Storage of Global Stiness Matrix, MAXA Vector . .
5.4 Computer Program Organization . . . . . . . . . . . . . . . .
5.5 Computer Program Flow Charts (FORTRAN) . . . . . . .
5.5.1 Input . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5.2 Element Stiness Matrices . . . . . . . . . . . . . . . .
5.5.3 Assembly . . . . . . . . . . . . . . . . . . . . . . . . .
5.5.4 Decomposition . . . . . . . . . . . . . . . . . . . . . .
5.5.5 Load . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5.6 Backsubstitution . . . . . . . . . . . . . . . . . . . . .
5.5.7 Internal Forces and Reactions . . . . . . . . . . . . . .
5.6 Computer Implementation with MATLAB . . . . . . . . . . .
5.6.1 Program Input . . . . . . . . . . . . . . . . . . . . . .
5.6.1.1 Input Variable Descriptions . . . . . . . . . .
5.6.1.2 Sample Input Data File . . . . . . . . . . . .
5.6.1.3 Program Implementation . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
51
. 51
. 51
. 52
. 52
. 53
. 53
. 54
. 54
. 55
. 57
. 57
. 512
. 515
. 519
. 522
. 524
. 525
. 528
. 528
. 531
. 531
. 531
. 531
. 535
. 535
. 536
. 537
. 538
3.4
3.5
Examples . .
E 3-1 Beam
E 3-2 Frame
E 3-3 Grid .
Observations
Homework . .
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4 TRANSFORMATION MATRICES
4.1 Preliminaries . . . . . . . . . . . . . . .
4.1.1 [ke ] [Ke ] Relation . . . . . . . .
4.1.2 Direction Cosines . . . . . . . . .
4.2 Transformation Matrices For Framework
4.2.1 2 D cases . . . . . . . . . . . . .
4.2.1.1 2D Frame . . . . . . . .
4.2.1.2 Grid . . . . . . . . . . .
4.2.1.3 2D Truss . . . . . . . .
4.2.2 3D Frame . . . . . . . . . . . . .
4.2.3 3D Truss . . . . . . . . . . . . .
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
CONTENTS
5.6.2
5.7
Program
5.6.2.1
5.6.2.2
5.6.2.3
5.6.2.4
5.6.2.5
5.6.2.6
5.6.2.7
5.6.2.8
5.6.2.9
5.6.2.10
5.6.2.11
5.6.2.12
5.6.2.13
5.6.2.14
Homework . . .
03
Listing . . . . . . . . . . . . . . . . . . . . .
Main Program . . . . . . . . . . . . . . . . .
Assembly of ID Matrix . . . . . . . . . . . .
Element Nodal Coordinates . . . . . . . . . .
Element Lengths . . . . . . . . . . . . . . . .
Element Stiness Matrices . . . . . . . . . .
Transformation Matrices . . . . . . . . . . .
Assembly of the Augmented Stiness Matrix
Print General Information . . . . . . . . . .
Print Load . . . . . . . . . . . . . . . . . . .
Load Vector . . . . . . . . . . . . . . . . . .
Nodal Displacements . . . . . . . . . . . . .
Reactions . . . . . . . . . . . . . . . . . . . .
Internal Forces . . . . . . . . . . . . . . . . .
Sample Output File . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 538
. 538
. 540
. 541
. 542
. 542
. 543
. 544
. 545
. 545
. 546
. 547
. 548
. 549
. 550
. 552
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 . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
61
. 61
. 61
. 62
. 63
. 64
. 65
. 67
. 69
. 69
. 610
. 610
. 611
. 611
. 612
. 612
. 613
. 615
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 . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
71
71
72
73
73
73
75
75
76
76
77
Draft
04
7.4
7.5
7.6
CONTENTS
Stiness Matrix of a Curved Element . . . . . . . . . . . . . . . . . . . . . . . . . 79

Duality between the Flexibility and the Stiness Methods . . . . . . . . . . . . . 710
Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
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 . . . . . . . . . . . . . . . .
II
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Introduction to Finite Elements
9 REVIEW OF ELASTICITY
9.1 Stress . . . . . . . . . . . . . . . . .
9.1.1 Stress Traction Relation . . .
E 9-1 Stress Vectors . . . . . . . . .
9.2 Strain . . . . . . . . . . . . . . . . .
9.3 Fundamental Relations in Elasticity
9.3.1 Equation of Equilibrium . . .
E 9-2 Equilibrium Equation . . . .
9.3.2 Compatibility Equation . . .
9.4 Stress-Strain Relations in Elasticity .
9.5 Strain Energy Density . . . . . . . .
9.6 Summary . . . . . . . . . . . . . . .
81
81
83
83
83
83
83
83
85
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10 VARIATIONAL AND ENERGY METHODS

10.1 Work, Energy & Potentials; Denitions . . . . . . . . . . . . . .
10.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .
10.1.2 Internal Strain Energy . . . . . . . . . . . . . . . . . . .
10.1.2.1 Internal Work versus Strain Energy . . . . . .
10.1.3 External Work . . . . . . . . . . . . . . . . . . . . . . .
10.1.3.1 Path Independence of External Work . . . .
10.1.4 Virtual Work . . . . . . . . . . . . . . . . . . . . . . . .
10.1.5 Complementary Virtual Work . . . . . . . . . . . . . . .
10.1.5.1 Internal Virtual Work . . . . . . . . . . . . . .
10.1.5.1.1 Elastic Systems . . . . . . . . . . . .
10.1.5.1.2 Linear Elastic Systems . . . . . . . .
10.1.5.2 External Virtual Work W . . . . . . . . . . .
10.1.6 Complementary Virtual Work . . . . . . . . . . . . . . .
10.1.6.1 Internal Complementary Virtual Strain Energy
10.1.6.1.1 Arbitrary System . . . . . . . . . . .
10.1.6.1.2 Linear Elastic Systems . . . . . . . .
10.1.6.2 External Complementary Virtual Work W .
10.1.7 Potential Energy . . . . . . . . . . . . . . . . . . . . . .
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
91
. 91
. 92
. 92
. 94
. 95
. 95
. 96
. 96
. 97
. 98
. 98
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
U
. . .
. . .
. . .
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
101
. 101
. 101
. 102
. 103
. 104
. 105
. 106
. 106
. 106
. 106
. 108
. 109
. 109
. 109
. 109
. 1010
. 1010
. 1011
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
CONTENTS
10.2
10.3
10.4
10.5
10.6
10.7
10.8
05
10.1.7.1 Potential Functions . . . . . . . . . . . . . . . . . . . . . . . . . 1011

10.1.7.2 Potential of External Work . . . . . . . . . . . . . . . . . . . . . 1011
10.1.7.3 Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . 1011
Principle of Virtual Work and Complementary Virtual Work . . . . . . . . . . . 1011
10.2.1 Principle of Virtual Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 1012
10.2.1.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1012
E 10-1 Tapered Cantiliver Beam, Virtual Displacement . . . . . . . . . . . . . . . 1015
10.2.2 Principle of Complementary Virtual Work . . . . . . . . . . . . . . . . . . 1017
10.2.2.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017
E 10-2 Tapered Cantilivered Beam; Virtual Force . . . . . . . . . . . . . . . . . . 1018
E 10-3 Three Hinged Semi-Circular Arch . . . . . . . . . . . . . . . . . . . . . . . 1020
E 10-4 Cantilivered Semi-Circular Bow Girder . . . . . . . . . . . . . . . . . . . . 1021
Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1023
10.3.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1023
10.3.2 Euler Equations of the Potential Energy . . . . . . . . . . . . . . . . . . 1025
10.3.3 Castiglianos First Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 1027
E 10-5 Fixed End Beam, Variable I . . . . . . . . . . . . . . . . . . . . . . . . . . 1028
10.3.4 Rayleigh-Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1029
E 10-6 Uniformly Loaded Simply Supported Beam; Polynomial Approximation . 1031
E 10-7 Uniformly Loaded Simply Supported Beam; Fourrier Series . . . . . . . . 1032
E 10-8 Tapered Beam; Fourrier Series . . . . . . . . . . . . . . . . . . . . . . . . 1033
Complementary Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . 1035
10.4.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035
10.4.2 Castiglianos Second Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 1035
E 10-9 Cantilivered beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
10.4.2.1 Distributed Loads . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
E 10-10Deection of a Uniformly loaded Beam using Castiglianos second Theorem1036
Comparison of Alternate Approximate Solutions . . . . . . . . . . . . . . . . . . 1037
E 10-11Comparison of MPE Solutions . . . . . . . . . . . . . . . . . . . . . . . . 1037
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1038
Variational Calculus; Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
10.7.1 Euler Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
10.7.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045
E 10-12Extension of a Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045
E 10-13Flexure of a Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1047
Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048
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 . .
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
111
. 111
. 111
. 112
. 113
. 114
. 115
. 117
. 118
. 118
. 119
Draft
06
CONTENTS
11.3.2 C 1 : Hermitian Interpolation Functions . . . . . . . . . . . . . . . . . . . . 1110

11.4 Interpretation of Shape Functions in Terms of Polynomial Series . . . . . . . . . 1110
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 . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
121
. 121
. 121
. 122
. 122
. 122
. 124
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 . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
131
. 131
. 131
. 132
. 132
. 133
. 133
. 134
. 134
. 135
. 136
14 DYNAMIC ANALYSIS
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .
14.2 Variational Formulation . . . . . . . . . . . . . . . . .
14.2.1 Explicit Time Integration; Central Dierence .
14.2.2 Implicit Time Integration . . . . . . . . . . . .
14.2.2.1 Linear Case . . . . . . . . . . . . . . .
14.2.2.2 NonLinear Case . . . . . . . . . . . .
14.3 Time Discretization . . . . . . . . . . . . . . . . . . .
14.3.1 Explicit Time Integration . . . . . . . . . . . .
14.3.1.1 Linear Systems . . . . . . . . . . . . .
E 14-1 MATLAB Code for Explicit Time Integration .
14.3.1.2 NonLinear Systems . . . . . . . . . .
14.3.2 Implicit Case . . . . . . . . . . . . . . . . . . .
14.3.2.1 Newmark Method; Forward dierence
14.3.2.2 Linear Case . . . . . . . . . . . . . . .
14.3.2.3 NonLinear Case . . . . . . . . . . . .
14.4 Free Vibration . . . . . . . . . . . . . . . . . . . . . .
14.5 Homework . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
141
. 141
. 141
. 143
. 144
. 144
. 144
. 145
. 145
. 145
. 148
. 149
. 149
. 149
. 1411
. 1411
. 1411
. 1413
15 GEOMETRIC NONLINEARITY
151
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
15.1.1 Strong Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
15.1.1.1 Lower Order Dierential Equation; Essential B.C. . . . . . . . . 152
15.1.1.2 Higher Order Dierential Equation; Essential and Natural B.C. 153
15.1.2 Weak Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Victor Saouma
Draft
CONTENTS
15.2
15.3
15.4
15.5
07
15.1.2.1 Strain Energy . . . . . . . . . . . . . .

15.1.2.2 Euler Equation . . . . . . . . . . . . . .
Finite Element Discretization . . . . . . . . . . . . . . .
Elastic Instability; Bifurcation Analysis . . . . . . . . .
E 15-1 Column Stability . . . . . . . . . . . . . . . . . .
E 15-2 Frame Stability . . . . . . . . . . . . . . . . . . .
Second-Order Elastic Analysis; Geometric Non-Linearity
E 15-3 Eect of Axial Load on Flexural Deformation . .
E 15-4 Bifurcation . . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
16 REFERENCES
A REVIEW of MATRIX ALGEBRA
A.1 Denitions . . . . . . . . . . . . . .
A.2 Elementary Matrix Operations . .
A.3 Determinants . . . . . . . . . . . .
A.4 Singularity and Rank . . . . . . . .
A.5 Inversion . . . . . . . . . . . . . . .
A.6 Eigenvalues and Eigenvectors . . .
. 156
. 157
. 158
. 159
. 1510
. 1513
. 1515
. 1516
. 1519
. 1522
161
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
B SOLUTIONS OF LINEAR EQUATIONS

B.1 Introduction . . . . . . . . . . . . . . . . . .
B.2 Direct Methods . . . . . . . . . . . . . . . .
B.2.1 Gauss, and Gaus-Jordan Elimination
E B-1 Gauss Elimination . . . . . . . . . .
E B-2 Gauss-Jordan Elimination . . . . . .
B.2.1.1 Algorithm . . . . . . . . .
B.2.2 LU Decomposition . . . . . . . . . .
B.2.2.1 Algorithm . . . . . . . . .
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
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A1
. A1
. A3
. A3
. A4
. A5
. A5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
B1
. B1
. B1
. B1
. B2
. B3
. B4
. B4
. B5
. B5
. B6
. B7
. B8
. B8
. B8
. B9
. B9
. B9
. B10
C TENSOR NOTATION
C1
C.1 Engineering Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C1
C.2 Dyadic/Vector Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C1
C.3 Indicial/Tensorial Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C2
Victor Saouma
Draft
08
CONTENTS
D INTEGRAL THEOREMS
D1
D.1 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D1
D.2 Green-Gradient Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D1
D.3 Gauss-Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D1
Victor Saouma
Draft
List of Figures
1.1
1.2
1.3
1.4
1.5
1.6
1.7
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 . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 16
. 16
. 16
. 17
. 17
. 19
. 110
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
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 . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 22
. 25
. 26
. 26
. 28
. 29
. 211
. 212
. 212
3.1
3.2
3.3
Problem with 2 Global d.o.f. 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . 32

Frame Example (correct K23 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Grid Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
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 . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 43
. 43
. 45
. 46
. 46
. 48
. 49
. 410
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
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 . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 51
. 54
. 55
. 58
. 513
. 516
. 520
. 521
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
02
LIST OF FIGURES
5.9
5.10
5.11
5.12
5.13
5.14
5.15
5.16
Program Flowchart . . . . . . . . . . . . . . . . . .
Programs Tree Structure . . . . . . . . . . . . . .
Flowchart for the Skyline Height Determination . .
Flowchart for the Global Stiness Matrix Assembly
Flowchart for the Load Vector Assembly . . . . . .
Flowchart for the Internal Forces . . . . . . . . . .
Flowchart for the Reactions . . . . . . . . . . . . .
Structure Plotted with CASAP . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 526
. 527
. 529
. 530
. 532
. 533
. 534
. 550
6.1
6.2
6.3
6.4
6.5
Example
Example
Example
Example
Example
of [B] Matrix for a Statically Determinate Truss .

of [B] Matrix for a Statically Determinate Beam .
of [B] Matrix for a Statically Indeterminate Truss
1, Congruent Transfer . . . . . . . . . . . . . . . .
2. . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 62
. 64
. 65
. 614
. 616
7.1
Stable and Statically Determinate Element . . . . . . . . . . . . . . . . . . . . . 75
8.1
8.2
8.3
Flexible, Rigid, and Semi-Rigid Beams . . . . . . . . . . . . . . . . . . . . . . . . 81

Deformations of a Semi-Rigid Beams . . . . . . . . . . . . . . . . . . . . . . . . . 82
Fixed End Forces in a Semi-Rigid Beams . . . . . . . . . . . . . . . . . . . . . . 82
9.1
9.2
9.3
9.4
9.5
Stress Components on an Innitesimal Element

Stresses as Tensor Components . . . . . . . . .
Stress Traction Relations . . . . . . . . . . . .
Equilibrium of Stresses, Cartesian Coordinates
Fundamental Equations in Solid Mechanics . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
91
92
93
95
99
10.1 Strain Energy and Complementary Strain Energy . . . . . . . . . . . . . . . . . . 102

10.2 Eects of Load Histories on U and Wi . . . . . . . . . . . . . . . . . . . . . . . . 104
10.3 Torsion Rotation Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10.4 Flexural Member . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10.5 Tapered Cantilivered Beam Analysed by the Vitual Displacement Method . . . . 1015
10.6 Tapered Cantilevered Beam Analysed by the Virtual Force Method . . . . . . . . 1019
10.7 Three Hinge Semi-Circular Arch . . . . . . . . . . . . . . . . . . . . . . . . . . . 1021
10.8 Semi-Circular Cantilevered Box Girder . . . . . . . . . . . . . . . . . . . . . . . . 1022
10.9 Single DOF Example for Potential Energy . . . . . . . . . . . . . . . . . . . . . . 1024
10.10Graphical Representation of the Potential Energy . . . . . . . . . . . . . . . . . . 1025
10.11Variable Cross Section Fixed Beam . . . . . . . . . . . . . . . . . . . . . . . . . . 1028
10.12Uniformly Loaded Simply Supported Beam Analysed by the Rayleigh-Ritz Method1031
10.13Example xx: External Virtual Work . . . . . . . . . . . . . . . . . . . . . . . . . 1034
10.14Summary of Variational Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 1039
10.15Duality of Variational Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . 1040
10.16Variational and Dierential Operators . . . . . . . . . . . . . . . . . . . . . . . . 1042
11.1
11.2
11.3
11.4
11.5
Axial Finite Element . . . . . . . . . . . . . . . . . . . .

Flexural Finite Element . . . . . . . . . . . . . . . . . .
Shape Functions for Flexure of Uniform Beam Element.
Constant Strain Triangle Element . . . . . . . . . . . . .
Constant Strain Quadrilateral Element . . . . . . . . . .
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 112
. 114
. 116
. 116
. 119
Draft
LIST OF FIGURES
03
11.6 Solid Trilinear Rectangular Element . . . . . . . . . . . . . . . . . . . . . . . . . 1110

15.1
15.2
15.3
15.4
Level of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Euler Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Simply Supported Beam Column; Dierential Segment; Eect of Axial Force P
Solution of the Tanscendental Equation for the Buckling Load of a Fixed-Hinged
Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.5 Summary of Stability Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . .
Victor Saouma
. 151
. 152
. 154
. 155
. 1523
Draft
04
Victor Saouma
LIST OF FIGURES
Draft
List of Tables
1.1
1.2
1.3
1.4
Example of Nodal Denition .

Example of Element Denition
Example of Group Number . .
Degrees of Freedom of Dierent
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
Structure Types Systems
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2.1
Examples of Inuence Coecients . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.1
3D Transformations of Linear Elements . . . . . . . . . . . . . . . . . . . . . . . 47
6.1
6.2
Internal Element Force Denition for the Statics Matrix . . . . . . . . . . . . . . 62

Conditions for Static Determinacy, and Kinematic Instability . . . . . . . . . . . 69
10.1
10.2
10.3
10.4
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
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
14
15
15
18
. 1012
. 1038
. 1041
. 1046
11.1 Characteristics of Beam Element Shape Functions . . . . . . . . . . . . . . . . . 115

11.2 Interpretation of Shape Functions in Terms of Polynomial Series (1D & 2D) . . . 1111
11.3 Polynomial Terms in Various Element Formulations (1D & 2D) . . . . . . . . . . 1111
Draft
02
Victor Saouma
LIST OF TABLES
Draft
LIST OF TABLES
03
NOTATION
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
Victor Saouma
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
Draft
04
t
u
u
u(x)
u, v, w
U
U
x, y
X, Y
W
[]
{}
{}
{}
0
{}
{0 }
{x }
M
P
U
W
t
u
LIST OF TABLES
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
Complementary strain energy
loacal coordinate system (lc)
Global coordinate system (GC)
Work
Coecient of thermal expansion
Transformation matrix
Element nodal displacements (lc)
Nodal displacements in a continuous system
Structure nodal displacements (GC)
Strain vector
Initial strain vector
Element relative displacement (lc)
Nonredundant element relative displacement (lc)
Redundant element relative displacement (lc)
rotational displacement with respect to z direction (for 2D structures)
Variational operator
Virtual moment
Virtual force
Virtual rotation
Virtual displacement
Virtual curvature
Virtual internal strain energy
Virtual external work
Virtual strain vector
Virtual stress vector
Surface
Surface subjected to surface tractions
Surface associated with known displacements
Stress vector
Initial stress vector
Volume of body
lc: Local Coordinate system

GC: Global Coordinate System
Victor Saouma
Draft
Chapter 1
INTRODUCTION
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.
1
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.
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.
3
4 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.
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.
5
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
structures such
are constituted
those elements,
and three dimensional continuum are essential in civil engineering to model

as dams, shells, and foundation, the majority of Civil engineering structures
by rod one-dimensional elements such as beams, girders, or columns. For
displacements and internal forces are somehow more complex than those
Draft
12
INTRODUCTION
encountered in continuum nite elements.

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 more relevant in the analysis/design of most civil engineering
structures.
8
Hence, Matrix Structural Analysis, is truly a bridge course between introductory analysis
and nite element courses. The element stiness matrix [k] will rst be derived using methods
introduced in basic structural analysis, and later using energy based concepts. This later
approach is the one exclusively used in the nite element method.
9
10 An important component of this course is computer programing. Once the theory and the
algorithms are thoroughly explained, you will be expected to program them in either Fortran
(preferably 90) or C (sorry, but no Basic) on the computer of your choice. The program
(typically about 3,500 lines) will perform the analysis of 2 and 3 dimensional truss and frame
structures, and many students have subsequently used it in their professional activities.
There will be one computer assignment in which you will be expected to perform simple symbolic manipulations using Mathematica. For those of you unfamiliar with the Bechtel
Laboratory, there will be a special session to introduce you to the operation of Unix on Sun
workstations.
11
1.2
Overview of Structural Analysis
12 To put things into perspective, it may be helpful to consider classes of Structural Analysis
which are distinguished by:
1. Excitation model
(a) Static
(b) Dynamic
2. Structure model
(a) Global geometry
small deformation ( =
u
x )
du
dv
large deformation (x = dx + 1 dx
2
(b) Structural elements element types:
, P- eects), chapter 15
1D framework (truss, beam, columns)

2D nite element (plane stress, plane strain, axisymmetric, plate or shell elements), chapter 13
3D nite element (solid elements)
(c) Material Properties:
Linear
Nonlinear
Victor Saouma
Draft
1.3 Structural Idealization
13
(d) Sectional properties:

Constant
Variable
(e) Structural connections:
Rigid
Semi-exible (linear and non-linear)
(f) Structural supports:
Rigid
Elastic
3. Type of solution:
(a) Continuum, analytical, Partial Dierential Equation
(b) Discrete, numerical, Finite ELement, Finite Dierence, Boundary Element
13
Structural design must satisfy:

1. Strength ( < f )
2. Stiness (small deformations)
3. Stability (buckling, cracking)
14
Structural analysis must satisfy

1. Statics (equilibrium)
2. Mechanics (stress-strain or force displacement relations)
3. Kinematics (compatibility of displacement)
1.3
Structural Idealization
15 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).
Victor Saouma
Draft
14
INTRODUCTION
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?).
13. Load static or dynamic (when should a dynamic analysis be performed?).
14. Wind load (the lateral drift of a high rise building subjected to wind load, is often the
major limitation to higher structures).
15. Thermal load (can induce large displacements, specially when a thermal gradient is
present.).
16. Secondary stresses (caused by welding. Present in most statically indeterminate structures).
1.3.1
Structural Discretization
16 Once a structure has been idealized, it must be discretized to lend itself for a mathematical
representation which will be analyzed by a computer program. This discretization should
uniquely dene each node, and member.
17 The node is characterized by its nodal id (node number), coordinates, boundary conditions,
and load (this one is often dened separately), Table 1.1. Note that in this case we have two
Node No.
1
2
3
4
Coor.
X
Y
0.
0.
5.
5.
20. 5.
25. 2.5
X
1
0
0
1
B. C.
Y
1
0
0
1
Z
0
0
0
1
Table 1.1: Example of Nodal Denition

nodal coordinates, and three degrees of freedom (to be dened later) per node. Furthermore,
Victor Saouma
Draft
1.3 Structural Idealization
15
a 0 and a 1 indicate unknown or known displacement. Known displacements can be zero

(restrained) or non-zero (as caused by foundation settlement).
18 The element is characterized by the nodes which it connects, and its group number, Table
1.2.
Element
No.
1
2
3
From
Node
1
3
3
To
Node
2
2
4
Group
Number
1
2
2
Table 1.2: Example of Element Denition

Group number will then dene both element type, and elastic/geometric properties. The
last one is a pointer to a separate array, Table 1.3. In this example element 1 has element code
1 (such as beam element), while element 2 has a code 2 (such as a truss element). Material
group 1 would have dierent elastic/geometric properties than material group 2.
19
Group
No.
1
2
3
Element
Type
1
2
1
Material
Group
1
1
2
Table 1.3: Example of Group Number

20 From the analysis, we rst obtain the nodal displacements, and then the element internal
forces. Those internal forces vary according to the element type. For a two dimensional frame,
those are the axial and shear forces, and moment at each node.
21 Hence, the need to dene two coordinate systems (one for the entire structure, and one for
each element), and a sign convention become apparent.
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.
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.
23
24
Two dimensional structures will be dened in the X-Y plane.
Victor Saouma
Draft
16
INTRODUCTION
X
Z
2D TRUSS
FRAME
BEAM
3D TRUSS
GRID & FRAME
Figure 1.1: Global Coordinate System

y, 2
x, 1
BEAM, TRUSS
x, 1
z, 3
GRID, FRAME
Figure 1.2: Local Coordinate Systems
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!).
In matrix structural analysis the sign convention adopted is consistent with the prevailing
coordinate system. Hence, we dene a positive moment as one which is counter-clockwise, Fig.
1.3
26
27
Fig. 1.4 illustrates the sign convention associated with each type of element.
28 Fig. 1.4 also shows the geometric (upper left) and elastic material (upper right) properties
associated with each type of element.
1.4
Degrees of Freedom
29
A degree of freedom (d.o.f.) is an independent generalized nodal displacement of a node.
30
The displacements must be linearly independent and thus not related to each other. For
+ve M
+
+ve M
Design Sign Convention (US)
Analysis Sign COnvention
Figure 1.3: Sign Convention, Design and Analysis

Victor Saouma
Draft
1.4 Degrees of Freedom
17
Beam
I x, L
v1
E A, L
v3
2D Truss
4
u1
u2
v2
u1
v5
3
I , I x, L y
v2
v2
4
u1
u2
4
u1
6
v3
u4
E A, I ,x I y, I z L
,
3D Truss
v5
1
3
E A, L
Grid
2D Frame
E A, I x L
,
3D Frame
v8 11
10
12
u7
v9
Figure 1.4: Total Degrees of Freedom for various Type of Elements
Figure 1.5: Independent Displacements

example, a roller support on an inclined plane would have three displacements (rotation , and
two translations u and v), however since the two displacements are kinematically constrained,
we only have two independent displacements, Fig. 1.5.
31 We note that we have been referring to generalized displacements, because we want this term
to include translations as well as rotations. Depending on the type of structure, there may
be none, one or more than one such displacement. It is unfortunate that in most introductory
courses in structural analysis, too much emphasis has been placed on two dimensional structures,
and not enough on either three dimensional ones, or two dimensional ones with torsion.
32 In most cases, there is the same number of d.o.f in local coordinates as in the global coordinate
system. One notable exception is the truss element. In local coordinate we can only have one
axial deformation, whereas in global coordinates there are two or three translations in 2D and
3D respectively for each node.
33 Hence, it is essential that we understand the degrees of freedom which can be associated
with the 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.
35
We should distinguish between local and global d.o.f.s. The numbering scheme follows the
Victor Saouma
Draft
18
INTRODUCTION
Type
Node 1
Node 2
1 Dimensional
Fy3 , Mz4
{p}
Fy1 , Mz2
{}
v1 , 2
{p}
Fx1
{}
{p}
u1
Fx1 , Fy2 , Mz3
u2
Fx4 , Fy5 , Mz6
{}
{p}
u1 , v2 , 3
Tx1 , Fy2 , Mz3
u4 , v5 , 6
Tx4 , Fy5 , Mz6
{}
1 , v2 , 3
{p}
Fx1 ,
{}
{p}
u1 ,
Fx1 , Fy2 , Fy3 ,
Tx4 My5 , Mz6
u2
Fx7 , Fy8 , Fy9 ,
Tx10 My11 , Mz12
u1 , v2 , w3 ,
4 , 5 6
[k]
(Local)
[K]
(Global)
44
44
22
44
66
66
66
66
22
66
12 12
12 12
u7 , v8 , w9 ,
10 , 11 12
Beam
v3 , 4
2 Dimensional
Fx2
Truss
Frame
Grid
4 , v5 , 6
3 Dimensional
Fx2
Truss
Frame
{}
Table 1.4: Degrees of Freedom of Dierent Structure Types Systems
Victor Saouma
Draft
1.5 Course Organization
19
Figure 1.6: Examples of Global Degrees of Freedom

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.
1.5
Course Organization
Victor Saouma
Draft
110
INTRODUCTION
Figure 1.7: Organization of the Course
Victor Saouma
Draft
Part I
Matrix Structural Analysis of

Framed Structures
Draft
Draft
Chapter 2
ELEMENT STIFFNESS MATRIX

2.1
Introduction
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.
1
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.
2
3 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.
2.2
Inuence Coecients
In structural analysis an inuence coecient Cij 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
Draft
22
Figure 2.1: Example for Flexibility Method

components of matrices used in structural analysis.
2.3
Flexibility Matrix (Review)
6 Considering the simply supported beam shown in Fig. 2.1, and using the local coordinate
system, we have
1
d11 d12
p1
=
(2.1)
2
d21 d22
p2
Using the virtual work, or more specically, the virtual force method to analyze this problem,
(more about energy methods in Chapter 10), we have:
l
M
dx = P + M
EIz
0
External
Internal
M
(2.2)
M
where M , EIz , P and are the virtual internal force, real internal displacement, virtual
external load, and real external displacement respectively. Here, both the external virtual force
and moment are usualy taken as unity.
Victor Saouma
Draft
2.4 Stiness Coecients
23
Virtual Force:
U
x =
x =
x x dvol
M xy
I
x
E
My
EI
y dA = I
dvol = dAdx
W = P
U = W
U =
EI
Hence:
EI 1 d11 =
M
dx
x
L
dx =
M
0
M
dx = P
EI
L
3
(2.3)
(2.4)
M M
Similarly, we would obtain:

L
EId22 =
EId12 =
0
L
0
x
L
1
dx =
x
L
L
3
L
x
dx =
L
6
(2.5-a)
= EId21
(2.5-b)
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
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.
9
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.
Victor Saouma
Draft
24
{p} = [k]{}
(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.
11
12 The actual stiness coecients are shown in Fig. 2.3 for truss, beam, and grid elements in
terms of elastic and geometric properties.
13
In the next sections, we shall derive those stiness coecients.
2.5
2.5.1
14
Force-Displacement Relations
Axial Deformations
From strength of materials, the force/displacement relation in axial members is

= E A =
P
AE
L
1
Hence, for a unit displacement, the applied force should be equal to

at the other end must be equal and opposite.
2.5.2
(2.8)
AE
L .
From statics, the force
Flexural Deformation
Our objective is to seek a relation for the shear and moments at each end of a beam, in terms
of known displacements and rotations at each end.
15
V1 = V1 (v1 , 1 , v2 , 2 )
(2.9-a)
M1 = M1 (v1 , 1 , v2 , 2 )
(2.9-b)
V2 = V2 (v1 , 1 , v2 , 2 )
(2.9-c)
M2 = M2 (v1 , 1 , v2 , 2 )
(2.9-d)
16 We start from the dierential equation of a beam, Fig. 2.4 in which we have all positive
known displacements, we have from strength of materials
M = EI
17
d2 v
= M 1 V1 x
dx2
(2.10)
Integrating twice
1
= M1 x V1 x2 + C1
2
1
1
M1 x2 V1 x3 + C1 x + C2
EIv =
2
6
EIv
Victor Saouma
(2.11-a)
(2.11-b)
Draft
2.5 Force-Displacement Relations
25
Figure 2.2: Denition of Element Stiness Coecients
Victor Saouma
Draft
26
Figure 2.3: Stiness Coecients for One Dimensional Elements
M2
M
V1
2
v
V
2
Figure 2.4: Flexural Problem Formulation
Victor Saouma
Draft

18
27
Applying the boundary conditions at x = 0

v = 1
v = v1
19
v = 2
v = v2
EI2 = M1 L 1 V1 L2 EI1
2
EIv2 = 1 M1 L2 1 V1 L3 EI1 L EIv1
2
6
or
V1 =
M1 V1 L + M2 = 0
(M1 + M2 )
L
V2 = V1
(2.15)
2EIz
L 1
6EIz
L 1
2EIz
L 2
6EIz
v
L2 1
6EIz
v
L2 2
(2.16)
Solving those two equations, we obtain:

M1 =
M2 =
23
(2.14)
Substituting V1 into the expressions for 2 and v2 in Eq. 2.13 and rearranging
M 1 M2 =
2M1 M2 =
22
(2.13)
Since equilibrium of forces and moments must be satised, we have:

V 1 + V2 = 0
21
(2.12)
Applying the boundary conditions at x = L and combining with the expressions for C1 and
C2
20
C1 = EI1
C2 = EIv1
2EIz
(21 + 2 ) +
L
2EIz
(1 + 22 ) +
L
6EIz
(v1 v2 )
L2
6EIz
(v1 v2 )
L2
(2.17)
(2.18)
Finally, we can substitute those expressions in Eq. 2.15

6EIz
12EIz
(1 + 2 ) +
(v1 v2 )
L2
L3
6EIz
12EIz
= 2 (1 + 2 )
(v1 v2 )
L
L3
V1 =
V2
2.5.3
(2.19)
(2.20)
Torsional Deformations
24 From Fig. 2.2-d. Since torsional eects are seldom covered in basic structural analysis, and
students may have forgotten the derivation of the basic equations from the Strength of Material
course, we shall briey review them.
25 Assuming a linear elastic material, and a linear strain (and thus stress) distribution along
the radius of a circular cross section subjected to torsional load, Fig. 2.5 we have:
=
A
max dA
c
stress
(2.21-a)
area arm
F orce
Victor Saouma
torque
Finite
Element I; Framed Structures
Draft
28
Figure 2.5: Torsion Rotation Relations

max
c
2 dA
(2.21-b)
A
J
Tc
J
max =
(2.21-c)
Note the analogy of this last equation with =
Mc
Iz .
2 dA is the polar moment of inertia J. It is also referred to as the St. Venants torsion
26
constant. For circular cross sections

J
2 dA =
=
A
c4
2 (2d)
d4
2
32
For rectangular sections b d, and b < d, an approximate expression is given by
=
= kb3 d
0.3
k =
b
1+ d
(2.22-a)
(2.23-a)
(2.23-b)
For other sections, J is often tabulated.

27
Note that J corresponds to Ixx where x is the axis along the element.
28 Having developed a relation between torsion and shear stress, we now seek a relation between
torsion and torsional rotation. In Fig. 2.5, we consider the arc length BD
max dx = dc
d
d
dx = max
c
dx =
max
max =
max =
G
29
Finally, we can rewrite this last equation as
T =
max
Gc
TC
J
T dx =
d
T
=
dx
GJ
(2.24)
Gjd and obtain:
GJ
(2.25)
Note the similarity between this equation and Equation 2.8.

Victor Saouma
Draft
29
dy
dv
dx
Figure 2.6: Deformation of an Innitesimal Element Due to Shear
2.5.4
Shear Deformation
In general, shear deformations are quite small. However, for beams with low span to depth
ratio, those deformations can not be neglected.
30
31 Considering an innitesimal element subjected to shear, Fig. 2.6 and for linear elastic material, the shear strain (assuming small displacement, i.e. tan ) is given by
dvs
dx
tan =
Kinematics
(2.26)
M aterial
s
where dvx is the slope of the beam neutral axis from the horizontal while the vertical sections
d
remain undeformed, G is the shear modulus, the shear stress, and vs the shear induced
displacement.
32 Along a beam cross section, the shear stress is not constant. For example for rectangular
sections, it varies parabolically, and in I sections, the ange shear components can be neglected.
VQ
Ib
(2.27)
where V is the shear force, Q is the rst moment (or static moment) about the neutral axis of
the portion of the cross-sectional area which is outside of the section where the shear stress is
to be determined, I is the moment of inertia of the cross sectional area about the neutral axis,
and b is the width of the rectangular beam.
33
The preceding equation can be simplied as

=
V
As
(2.28)
where As is the eective cross section for shear (which is the ratio of the cross sectional area
to the area shear factor)
Victor Saouma
Draft
210
34 Let us derive the expression of As for rectangular sections. The exact expression for the
shear stress is
VQ
=
(2.29)
Ib
where Q is the moment of the area from the external bers to y with respect to the neutral
axis; For a rectangular section, this yields
VQ
Ib
V
Ib
=
=
(2.30-a)
h/2
by dy =
y
(2.30-b)
h2
y2
4
6V
bh3
h2
y2
4
V
2I
(2.30-c)
and we observe that the shear stress is zero for y = h/2 and maximum at the neutral axis
V
where it is equal to 1.5 bh .
35
To determine the form factor of a rectangular section such that As = A (clarify)
Q
= VIb
= kV
A
h/2
Q =
y
b
by dy =
2
h2
y2
4
k =
bh
QA
Ib
A
2I
h2
4
y2
k 2 dydz
=
A
= 1.2
(2.31)
Thus, the form factor may be taken as 1.2 for rectangular beams of ordinary proportions,
and As = 1.2A
For I beams, k can be also approximated by 1.2, provided A is the area of the web.
36
Combining Eq. 2.26 and 2.28 we obtain

dvs
V
=
dx
GAs
(2.32)
Assuming V to be constant, we integrate

vs =
V
x + C1
GAs
(2.33)
If the displacement vs is zero at the opposite end of the beam, then we solve for C1 and
obtain
V
vs =
(x L)
(2.34)
GAs
37
38
We dene
Hence for small slenderness ratio

39
def
12EI
GAs L2
24(1 + )
r
L
(2.35)
A
As
r
L
(2.36)
compared to unity, we can neglect .
Next, we shall consider the eect of shear deformations on both translations and rotations
Victor Saouma
Draft
2.6 Putting it All Together, [k]
6EI
L2
211
12EI
L3
1
6EI
L2
12EI
L3
Figure 2.7: Eect of Flexure and Shear Deformation on Translation at One End
Eect on Translation Due to a unit vertical translation, the end shear force is obtained from
Eq. 2.19 and setting v1 = 1 and 1 = 2 = v2 = 0, or V = 12EIz . At x = 0 we have, Fig.
L3
2.7, and Eq. 2.34
VL
vs = GAs = V
v =
(2.37)
V = 12EIz
3
L
s
12EI
= GAs L2
Hence, the shear deformation has increased the total translation from 1 to 1 + . Similar
arguments apply to the translation at the other end.
Eect on Rotation Considering the beam shown in Fig. 2.8, even when a rotation 1 is
applied, an internal shear force is induced, and this in turn is going to give rise to shear
deformations (translation) which must be accounted for. The shear force is obtained from
Eq. 2.19 and setting 1 = 1 and 2 = v1 = v2 = 0, or V = 6EIz . At x = 0,
L2
vs =
V =
=
VL
GAs
6EIz
L2
12EI
GAs L2
vs = 0.5L
(2.38)
in other words, the shear deformation has moved the end of the beam (which was supposed
to have zero translation) by 0.5L.
2.6
Putting it All Together, [k]
40 Using basic structural analysis methods we have derived various force displacement relations
for axial, exural, torsional and shear imposed displacements. At this point, and keeping in
mind the denition of degrees of freedom, we seek to assemble the individual element stiness
matrices [k]. We shall start with the simplest one, the truss element, then consider the beam,
2D frame, grid, and nally the 3D frame element.
41 In each case, a table will cross-reference the force displacement relations, and then the element
stiness matrix will be accordingly dened.
Victor Saouma
Draft
212

6EI
L2
4EI
2EI
L
6EI
L2
1=1
0.5L1
0.5L1
Figure 2.8: Eect of Flexure and Shear Deformation on Rotation at One End
Y
Beam Element
Grid Element
Z
Frame Element
Z
Figure 2.9: Coordinate System for Element Stiness Matrices

42 Fig. 2.9 illustrates the coordinate syatems used for the element stiness matrix denitions
in this section.
2.6.1
Truss Element
43 The truss element (whether in 2D or 3D) has only one degree of freedom associated with
each node. Hence, from Eq. 2.8, we have
AE
[kt ] =
L
2.6.2
44
u1 u2
p1 1 1
p2 1 1
(2.39)
Beam Element
There are two major beam theories:
Victor Saouma
Draft
213
Euler-Bernoulli which is the classical formulation for beams.

Timoshenko which accounts for transverse shear deformation eects.
2.6.2.1
Euler-Bernoulli
45 Using Equations 2.17, 2.18, 2.19 and 2.20 we can determine the forces associated with each
unit displacement.
[kb ] =
46
V1 Eq.
M1 Eq.
V2 Eq.
M2 Eq.
v1
2.19(v1
2.17(v1
2.20(v1
2.18(v1
= 1)
= 1)
= 1)
= 1)
Eq.
Eq.
Eq.
Eq.
1
2.19(1
2.17(1
2.20(1
2.18(1
= 1)
= 1)
= 1)
= 1)
Eq.
Eq.
Eq.
Eq.
v2
2.19(v2
2.17(v2
2.20(v2
2.18(v2
= 1)
= 1)
= 1)
= 1)
Eq.
Eq.
Eq.
Eq.
2
2.19(2
2.17(2
2.20(2
2.18(2
= 1)
= 1)
= 1)
= 1)
(2.40)
The stiness matrix of the beam element (neglecting shear and axial deformation) will thus
be
[kb ] =
2.6.2.2
v1
V1 12EIz
L3
M1 6EIz
L2
V2 12EIz
L3
L2
M2 6EIz
6EIz
L2
4EIz
L
6EIz
L2
2EIz
L
v2
12EIz
L3
6EIz
L2
12EIz
L3
6EIz
L2
6EIz
L2
2EIz
L
6EIz
L2
4EIz
L
(2.41)
Timoshenko Beam
47 If shear deformations are present, we need to alter the stiness matrix given in Eq. 2.41 in
the following manner
1. Due to translation, we must divide (or normalize) the coecients of the rst and third
columns of the stiness matrix by 1 + so that the net translation at both ends is unity.
2. Due to rotation and the eect of shear deformation
(a) The forces induced at the ends due to a unit rotation at end 1 (second column)
neglecting shear deformations are
V1 = V2 =
M1 =
M2 =
6EI
L2
4EI
L
2EI
L
(2.42-a)
(2.42-b)
(2.42-c)
(b) There is a net positive translation of 0.5L at end 1 when we applied a unit rotation
(this parasitic translation is caused by the shear deformation) but no additional
forces are induced.
Victor Saouma
Draft
214
(c) When we apply a unit rotation, all other displacements should be zero. Hence,
we should counteract this parasitic shear deformation by an equal and opposite
one. Hence, we apply an additional vertical displacement of 0.5L and the forces
induced at the ends (rst column) are given by
V1 = V2 =
12EI 1
(0.5L)
L3 1 +
bt
k11
M 1 = M2 =
(2.43-a)
vs
6EI 1
(0.5L)
L2 1 +
(2.43-b)
vs
bt
k21
Note that the denominators have already been divided by 1 + in kbt .

(d) Summing up all the forces, we have the forces induced as a result of a unit rotation
only when the eects of both bending and shear deformations are included.
V1 = V2 =
6EI
L2
Due to Unit Rotation
12EI 1
(0.5L)
L3 1 +
bt
k11
(2.44-a)
vs
Due to Parasitic Shear
M1
6EI 1
= 2
L 1+
4EI
=
L
(2.44-b)
+
6EI 1
(0.5L)
L2 1 +
bt
k21
(2.44-c)
vs
=
M2 =
4 + EI
1+ L
2EI
L
(2.44-d)
+
48
2 EI
1+ L
6EI 1
(0.5L)
L2 1 +
bt
k21
(2.44-e)
vs
(2.44-f)
Thus, the element stiness matrix given in Eq. 2.41 becomes
[kbV ] =
Victor Saouma
v1
V1 L312EIz y )
(1+
6EIz
M1 L2 (1+y )
V2 L312EIz y )
(1+
M2 L26EIz y )
(1+
1
6EIz
L2 (1+y )
(4+y )EIz
(1+y )L
L26EIz y )
(1+
(2y )EIz
L(1+y )
v2
L312EIz y )
(1+
6EIz
L2 (1+y )
12EIz
L3 (1+y )
L26EIz y )
(1+
2
6EIz
L2 (1+y )
(2y )EIz
L(1+y )
L26EIz y )
(1+
(4+y )EIz
L(1+y )
(2.45)
Draft
2.6.3
215
2D Frame Element
49 The stiness matrix of the two dimensional frame element is composed of terms from the
truss and beam elements where kb and kt refer to the beam and truss element stiness matrices
respectively.
u1x v1y 1z u2x v2y 2z
t
t
P1x k11
0
0 k12
0
0
b
b
b
b
V1y 0 k11 k12
0 k13 k14
0
b
b
b
b
M1z
k21 k22
0 k23 k24
2df r
]=
(2.46)
[k
t
t
P2x k21
0
0 k22
0
0
b
b
b
b
0 k33 k34
V2y 0 k31 k32
b
b
b
b
M2z 0 k41 k42
0 k43 k44
Thus, we have:
[k2df r ] =
u1x
P1x EA
L
V1y 0
M1z 0
P2x EA
V2y 0
M2z 0
v1y
0
1z
0
12EIz
L3
6EIz
L2
6EIz
L2
4EIz
L
12EIz
L3
6EIz
L2
6EIz
L2
2EIz
L
u2x
EA
L
0
0
EA
L
0
0
v2y
0
2z
0
12EIz
L3
6EIz
L2
0
6EIz
L2
2EIz
L
12EIz
L3
6EIz
L2
6EI
L2
0
4EIz
L
(2.47)
Note that if shear deformations must be accounted for, the entries corresponding to shear
and exure must be modied in accordance with Eq. 2.45
50
2.6.4
Grid Element
51 The stiness matrix of the grid element is very analogous to the one of the 2D frame element,
except that the axial component is replaced by the torsional one. Hence, the stiness matrix is
1x
u1y
T1x Eq. 2.25 0
b
V1y
0
k11
b
M1z
0
k21
T2x Eq. 2.25 0
b
V2y
0
k31
b
M2z
0
k41
[kg ] =
Victor Saouma
1z
0
b
k12
b
k22
0
b
k32
b
k42
2x
u2y
Eq. 2.25 0
b
0
k13
b
0
k23
Eq. 2.25 0
b
0
k33
b
0
k43
2z
0
b
k14
b
k24
b
k34
b
k44
(2.48)
Draft
216
Upon substitution, the grid element stiness matrix is given by

1x
T1x GIx
L
V1y 0
M1z 0
T2x Gix
V2y 0
M2z 0
[kg ] =
u1y
0
1z
0
12EIz
L3
6EIz
L2
6EIz
L2
4EIz
L
12EIz
L3
6EIz
L2
6EIz
L2
2EIz
L3
2x
GIx
L
0
0
2z
0
u2y
0
12EIz
L3
6EIz
L2
0
12EIz
L3
6EIz
L2
GIx
L
0
0
6EIz
L2
2EIz
L
6EIz
L2
0
4EIz
L
(2.49)
52 Note that if shear deformations must be accounted for, the entries corresponding to shear
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
v1
0
b
k11
0
0
0
b
k21
0
b
k13
0
0
0
b
k12
[k3df r ] =
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
z2
0
k b 14
b
k24
b
k34
0
b
k44
(2.50)
For [k3D ] and with we obtain:

11
Px1
u1
EA
l
0
Vz1
Tx1
0
0
My1
0
Mz1
Px2 EA
L
0
Vy2
0
Vz2
0
Tx2
0
My2
Vy1
[k
3d r
f
]=
Mz2
v1
w1
x1
y1
z1
u2
v2
w2
x2
y2
z2
EA
L
12EIy
L3
6EIy
L2
12EIy
L3
EIy
6 2
L
12EIz
L3
0
0
12EIz
L3
0
6EIz
L2
0
0
4EIz
L
6EIy
L2
0
GI
EIy
L2
2EIy
L
0
0
Lx
0
12EIz
L3
12EIy
L3
4EIy
L
6EIz
L2
GIx
L
6EIy
L2
6EIz
L2
6EIz
L2
6EIy
L2
0
0
2EIz
L
12EIy
L3
2EIz
L
6EIz
L2
2EIy
L
12EIz
L3
GI
Lx
0
6EIz
L2
0
EA
L
6EIy
L2
6EIz
L2
6EIy
L2
GIx
L
4EIy
L
6EIz
L2
0
0
0
4EIz
L
(2.51)
Note that if shear deformations must be accounted for, the entries corresponding to shear
53
Victor Saouma
Draft
2.7 Remarks on Element Stiness Matrices
2.7
217
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 1; and L times row 2 is equal to the sum of row 3 and 6. This singularity
(not present in the exibility matrix) is caused by the linear relations introduced by the
equilibrium equations which are embedded in the formulation.
Symmetry: All matrices are symmetric due to Maxwell-Bettis reciprocal theorem, and the
stiness exibility relation.
54
More about the stiness matrix properties later.
Victor Saouma
Draft
218
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.
11
00
11
00
x
z
Y
O
Victor Saouma
Draft
Chapter 3
STIFFNESS METHOD; Part I:

ORTHOGONAL STRUCTURES
3.1
Introduction
1 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.
In the previous chapter our starting point was basic displacement-force relations resulting in
element stiness matrices [k].
3
4 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.
5
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.
6
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
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 .
8
If we were to analyse this problem by the force (or exibility) method, then
Draft
32
STIFFNESS METHOD; Part I: ORTHOGONAL STRUCTURES
Figure 3.1: Problem with 2 Global d.o.f. 1 and 2
Victor Saouma
Draft
3.3 Examples
33
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 ).
3. Apply a unit load at point B, and then C, and compute the deections fij at note i due
to a unit force at node j.
4. Write the compatibility of displacement equation
fBB fBC
fCB fCC
RB
RC
1
2
0
0
(3.1)
5. Invert the matrix, and solve for the reactions

10
We will analyze this simple problem by the stiness method.

1. The rst step consists in making it kinematically determinate (as opposed to statically
determinate in the exibility method). Kinematically determinate in this case simply
means restraining all the d.o.f. and thus prevent joint rotation, Fig 3.1-b.
2. We then determine the xed end actions caused by the element load, and sum them for
each d.o.f., Fig 3.1-c: FEM1 and FEM2 .
3. In the third step, we will apply a unit displacement (rotation in this case) at each degree
of freedom at a time, and in each case we shall determine the intenal reaction forces, K11 ,
K21 , and K12 , K22 respectively. Note that we use [K], rather than [k] since those are
forces in the global coordinate system, Fig 3.1-d. Again note that we are focusing only
on the reaction forces corresponding to a global degree of freedom. Hence, we are not
attempting to determine the reaction at node A.
4. Finally, we write the equation of equilibrium at each node:
M1
M2
FEM1
FEM2
K11 K12
K21 K22
1
2
(3.2)
Note that the FEM being on the right hand side, they are detemined as the reactions to the
applied load. Strictly speaking, it is a load which should appear on the left hand side of the
equation, and are the nodal equivalent loads to the element load (more about this later).
11
As with the element stiness matrix, each entry in the global stiness matrix Kij , corresponds
to the internal force along d.o.f. i due to a unit displacement (generalized) along d.o.f. j (both
in global coordinate systems).
12
3.3
Examples
Example 3-1: Beam

Victor Saouma
Draft
34
Considering the previous problem, Fig. 3.1-a, let P1 = 2P , M = P L, P2 = P , and P3 = P ,

Solve for the displacements.
Solution:
1. Using the previously dened sign convention:
FEM1 =
P1 L P2 L
2P L P L
PL
+
=
+
=
8
8
8
8
8
BA
FEM2 =
(3.3-a)
BC
PL
8
(3.3-b)
CB
BA
BC
2. If it takes 4EI (k44 ) to rotate BA (Eq. 2.41) and 4EI (k22 ) to rotate BC, it will take
L
L
8EI
a total force of L to simultaneously rotate BA and BC, (Note that a rigid joint is
assumed).
3. Hence, K11 which is the sum of the rotational stinesses at global d.o.f. 1. will be equal
BC
to K11 = 8EI ; similarly, K21 = 2EI (k42 ) .
L
L
4. If we now rotate dof 2 by a unit angle, then we will have K22 =
BC
(k42 ) .
4EI
L
BC
(k22 ) and K12 =
2EI
L
5. The equilibrium relation can thus be written as:

PL
0
P8L
P8L
F EM
or
P L + P8L
+ P8L
2EI
L
4EI
L
8EI
L
2EI
L
1
2
K
8EI
L
2EI
L
(3.4)
2EI
L
4EI
L
1
2
(3.5)
6. The two by two matrix is next inverted

1
2
8EI
L
2EI
L
2EI
L
4EI
L
P L + P8L
+ P8L
17 P L2
112 EI
2
5
112 P L
EI
(3.6)
7. Next we need to determine both the reactions and the internal forces.
8. Recall that for each element {p} = [k]{}, and in this case {p} = {P} and {} = {}
for element AB. The element stiness matrix has been previously derived, Eq. 2.41, and
in this case the global and local d.o.f. are the same.
9. Hence, the equilibrium equation for element AB, at the element level, can be written as
(note that we must include the xed end actions to maintain equilibrium):
p1
p3
p4
Victor Saouma
{p}
12EI
L3
6EI
L2
12EI
L3
6EI
L2
12EI
L3
6EI
L2
6EI
L2
4EI
L
6EI
L2
2EI
L
12EI
L3
6EI
L2
[k]
6EI
L2
2EI
L
6EI
L2
4EI
L
0
0
0
17 P L2
112 EI
2P
2
2P L
8
2P
2
2P L
8
(3.7)

FEM
{ }
Draft
3.3 Examples
35
solving
p1 p2 p3 p4
107
56 P
31
56 P L
5
56 P
5
14 P L
(3.8)
10. Similarly, for element BC:
p1
p3
p4
12EI
L3
6EI
L2
12EI
L3
6EI
L2
{p}
12EI
L3
6EI
L2
6EI
L2
4EI
L
6EI
L2
2EI
L
6EI
L2
2EI
L
6EI
L2
4EI
L
12EI
L3
6EI
L2
0
17 P L2
112 EI
5
112 P L
EI
{ }
[k]
P
2
PL
8
P
2
P8L
(3.9)
FEM
or
p1 p2 p3 p4
7
8P
P
7
9
14 P L
(3.10)
11. This simple example calls for the following observations:

(a) Node A has contributions from element AB only, while node B has contributions
from both AB and BC.
(b) We observe that pAB = pBC eventhough they both correspond to a shear force
3
1
at node B, the dierence betweeen them is equal to the reaction at B. Similarly,
pAB = pBC due to the externally applied moment at node B.
4
2
(c) From this analysis, we can draw the complete free body diagram, Fig. 3.1-e and
then the shear and moment diagrams which is what the Engineer is most interested
in for design purposes.
Example 3-2: Frame

Whereas in the rst example all local coordinate systems were identical to the global one,
in this example we consider the orthogonal frame shown in Fig. 5.1,
Solution:
1. Assuming axial deformations, we do have three global degrees of freedom, 1 , 2 , and 3 .
2. Constrain all the degrees of freedom, and thus make the structure kinematically determinate.
3. Determine the xed end actions for each element in its own local coordinate system:
PL
8
P
2
p 1 p2 p3 p4 p5 p6
wH
2
P
2
P8L
(3.11-a)
AB
p1 v1 m1 p2 v2 m2
wH 2
12
wH
2
wH
12
(3.11-b)
BC
(3.11-c)
4. Summing the xed end actions at node B in global coordinates we have
P1 P2 P3
Victor Saouma
wH
2
P
2
P8L +
wH 2
12
(3.12)
Draft
36
1
0A
1
0

P
M
1
0
1
0
P/2
EI
w
L/2
2
B
L/2
11
00
11
00
K21
K31
11
00
K
32
1
B
1
0A
1
0
K22
1
0
1
0
K23
K12
1
0
1
0
1
0
1
0
K 33
K
13
K11
C
11
00
11
00
11
00
Figure 3.2: Frame Example (correct K23 )

5. Next, we apply a unit displacement in each of the 3 global degrees of freedom, and we seek
to determine the structure global stiness matrix. Each entry Kij of the global stiness matrix
will correspond to the internal force in degree of freedom i, due to a unit displacement in degree
of freedom j.
6. Recalling the force displacement relations derived earlier, we can assemble the global stiness
matrix in terms of contributions from both AB and BC:
Ki1
1
K1j
K2j
K3j
AB
BC
AB
BC
AB
BC
EA
L
12EI c
H3
0
0
0
6EI c
H2
Ki2
2
0
0
12EI b
L3
EA
H
b
6EI
L2
Ki3
3
0
6EI c
H2
b
6EI
L2
0
4EI b
L
4EI c
H
7. Summing up, the structure global stiness matrix [K] is:
[K] =
P1
P2
M3
Victor Saouma
EA
L
1
c
+ 12EI
H3
0
6EI c
H2
AB
P1 k44
AB
P2 k
AB
M3 k64
1
BC
+ k22
+ k BC
BC
+ k32
2
0
12EI b
+ EA
H
L3
b
6EI
2
L
AB
k45
AB
k55
AB
k65
2
BC
+ k21
BC
+ k11
BC
+ k31
3
6EI c
H2
b
6EI
L2
4EI b
4EI c
L + H
AB
k46
AB
k56
AB
k66
BC
+ k23
BC
+ k13
BC
+ k33
(3.13-a)
(3.13-b)
Draft
3.3 Examples
37
8. The global equation of equilibrium can now be written
wH
2
P
2
2
P8L + wH
12
EA
L
+ 12EI
H3
0
0
12EI b
+ EA
H
L3
b
6EI
2
L
6EI c
H2
F EA
6EI c
H2
b
6EI
L2
4EI b
4EI c
L + H
2
(3.14)
[K]
9. Solve for the displacemnts
2

3
EA
L
6EI c
H2
b
6EI
L2
b
4EI
4EI c
L + H
0
12EI b
+ EA
H
L3
c
b
6EI
6EI
H2
L2
(L3 (84Ib+19AL2 )P )
+ 12EI
H3
0
32E(3Ib+AL2 )(12Ib+AL2 )
(L3 (12Ib+13AL2 )P )
32E(3Ib+AL2 )(12Ib+AL2 )
L2 (12Ib+AL2 )P
64EIb(3Ib+AL2 )
P wH
2
2
P
2
2
M + P8L wH
12
(3.15-a)
10. To obtain the element internal forces, we will multiply each element stiness matrix by the
local displacements. For element AB, the local and global coordinates match, thus
p1
p2
EA
L
0
3
= EA

p4
p5
p6
12EIy
L3
6EIy
L2
6EIy
L2
4EIy
L
12EI
L3 y
6EIy
L2
6EI
L2 y
2EIy
L
0
P
2
PL
8
P
2
P8L
EA
L
0
0
EA
L
12EI
L3 y
6EI
L2 y
6EIy
L2
2EIy
L
0
0
12EIy
L3
6EIy
L2
6EI
L2
4EIy
L
0
0
0
(L3 (84Ib+19AL2 )P )
32E(3Ib+AL2 )(12Ib+AL2 )
(L3 (12Ib+13AL2 )P )
32E(3Ib+AL2 )(12Ib+AL2 )
L2 (12Ib+AL2 )P
64EIb(3Ib+AL2 )
(3.16-a)
11. For element BC, the local and global coordinates do not match, hence we will need to
transform the displacements from their global to their local coordinate components. But since,
vector (displacement and load), and matrix transformation have not yet been covered, we note
by inspection that the relationship between global and local coordinates for element BC is
Local
Global
1
0
2
0
3
0
4
2
5
1
6
3
and we observe that there are no local or global displacements associated with dof 1-3; Hence
Victor Saouma
Draft
38
the internal forces for element BC are given by:
p1
p2
EA
L
0
12EIy
L3
6EIy
L2
0
3
= EA

p4
p5
p6
6EIy
L2
4EIy
L
12EI
L3 y
6EIy
L2
6EI
L2 y
2EIy
L
wH
wH 2

12
wH
2
wH 2
EA
L
0
0
EA
L
0
0
12EI
L3 y
6EI
L2 y
6EIy
L2
2EIy
L
12EIy
L3
6EI
L2 y
6EI
L2
4EIy
L
0
0
0
(L3 (12I b +13AL2 )P )
32E (3I b +AL2 )(12I b +AL2 )

(L3 (84I b +19AL2 )P )
b +AL2
b +AL2
32E (3I
)(12I
)
L2 (12I b +AL2 )P
b
b
2
64EI
(3I
+AL
(3.17-a)
12
Note that the element is dened as going from C to B hence, x, y, z correspond to Y, X, Z.

Mathematica:
Ic=Ib
M= 0
w= 0
H= L
alpha=
K={
{E A /L + 12 E Ic /H^3, 0,
6 E Ic/H^2},
{0,
12 E Ib/L^3 + E A/H, -6 E Ib/L^2},
{6 E Ic/H^2,
-6 E Ib/L^2,
4 E Ib/L + 4 E Ic/H}
}
d=Inverse[K]
load={-P/2 - w H/2, -P/2, M+P L/8 -w H^2/12}
displacement=Simplify[d . load]
Example 3-3: Grid

Analyse the orthogonal grid shown in Fig. 3.3. The two elements have identical exural
and torsional rigidity, EI and GJ.
Solution:
1. We rst identify the three degrees of freedom, 1 , 2 , and 3 .
Victor Saouma
Draft
3.3 Examples
39
Figure 3.3: Grid Example

2. Restrain all the degrees of freedom, and determine the xed end actions:
T1
2
M
3
P
2
PL
8
0
P
(3.18)
2 l
P
8
@node A
@node B
3. Apply a unit displacement along each of the three degrees of freedom, and determine the
internal forces:
1. Apply unit rotation along global d.o.f. 1.
AB
(a) AB (Torsion) K11 =
BC
(b) BC (Flexure) K11 =
GJ
AB
AB
L , K21 = 0, K31 = 0
4EI
6EI
BC
BC
L , K21 = L2 , K31 =
4. Apply a unit displacement along global d.o.f. 2.

12EI
AB
, K32 =
L3
BC
BC
K22 = 12EI , K32
L3
AB
AB
(a) AB (Flexure): K12 = 0, K22 =
BC
(b) BC (Flexure): K12 =
6EI
,
L2
6EI
L2
=0
5. Apply unit rotation along global d.o.f. 3.

AB
AB
AB
(a) AB (Flexure): K13 = 0, K23 = 6EI , K33 =
L2
BC
BC
BC
(b) BC (Torsion): K13 = 0, K23 = 0, K33 =
GJ
L
4. The structure stiness matrix will now be assembled:
GJ
K11 K12 K13
L
K21 K22 K23 = 0
K31 K32 K33
0
0
12EI
L3
6EI
L2
[KAB ]
Victor Saouma
4EI
L
6EI +
L2
4EI
L
4EI
L
6EI
L2
6EI
L2
12EI
L3
GJ
L
[KBC ]
Draft
310
L2
0
0
4L2 6L
0
EI
EI
12 6L + 3 6L 12
0
0
L3
L
0
6L 4L2
0
0 L2
(4 + )L2 6L
0
EI
6L
24
6L
L3
0
6L (4 + )L2
(3.19-a)
[KStructure ]
AB
k44
AB
k64
AB
k54
+
+
+
BC
k55
BC
k65
BC
k45
AB
BC
k46 + k56
AB
BC
k66 + k66
AB + k BC
k56
46
AB
BC
k45 + k54
AB
BC
k65 + k64
AB
BC
k55 + k44
(3.19-b)
where = GJ , and in the last equation it is assumed that for element BC, node 1 corresponds
EI
to C and 2 to B.
5. The structure equilibrium equation in matrix form:
{P}
0
P
2L
P
8
FEA @B
or
1
(4 + )L2 6L
0
EI
6L
24
6L
2
+ 3
0
6L (4 + )L2 3
{}
[K]
2

3
(3.20)
5+2
P L2
16EI (1+)(4+)
P L3
96EI 5+2
1+
2
1
3P L (1+)(4+)
16EI
(3.21)
6. Determine the element internal forces. This will be accomplished by multiplying each element
stiness matrix [k] with the vector of nodal displacement {}. Note these operations should
be accomplished in local coordinate system, and great care should be exercized in writing the
nodal displacements in the same local coordinate system as the one used for the derivation of
the element stiness matrix, Eq. 2.49.
7. For element AB and BC, the vector of nodal displacements are
0
0
0
1
3
2
AB
0
0
0
(3.22)
BC
8. Hence, for element AB we have
p1
p2
GIx
L
0
3
= Gix
p4 L
p5 0

p6
Victor Saouma
0
4EIy
L
6EI
L2 y
0
6EIy
L2
12EIy
L3
2EIy
L
6EIy
L2
6EI
L2 y
12EI
L3 y
GIx
L
0
0
GIx
L
0
0
2EIy
L
6EI
L2 y
6EIy
L2
12EI
L3 y
4EIy
L
6EIy
L2
6EIy
L2
12EIy
L3
P/2
P L/8
+

0

P/2

P L/8
(3.23)
Draft
3.4 Observations
311
9. For element BC:
p1
p2
GIx
L
0
3
= Gix
p4 L
p5 0

p6
3.4
0
4EIy
L
6EI
L2 y
0
6EIy
L2
12EIy
L3
2EIy
L
6EIy
L2
L2 y
12EI
L3 y
6EI
GIx
L
0
0
GIx
L
0
0
2EIy
L
6EI
L2 y
6EIy
L2
12EI
L3 y
4EIy
L
6EIy
L2
6EIy
L2
12EIy
L3
(3.24)
Observations
On the basis of these two illustrative examples we note that the global structure equilibrium
equation can be written as
{P} = {FEA} + [K]{}
(3.25)
13
where [K] is the global structure stiness matrix (in terms of the unrestrained d.o.f.) {P} the
vector containing both the nodal load and the nodal equivalent load caused by element loading,
{} is the vector of generalized nodal displacements.
14 Whereas the preceding two examples were quite simple to analyze, we seek to generalize the
method to handle any arbitrary structure. As such, some of the questions which arise are:
1. How do we determine the element stiness matrix in global coordinate systems, [Ke ],
from the element stiness matrix in local coordinate system [ke ]?
2. How to assemble the structure [KS ] from each element [KE ]?
3. How to determine the {FEA} or the nodal equivalent load for an element load?
4. How to determine the local nodal displacements from the global ones?
5. How do we compute reactions in the restrained d.o.f?
6. How can we determine the internal element forces (P , V , M , and T )?
7. How do we account for temperature, initial displacements or prestrain?
Those questions, and others, will be addressed in the next chapters which will outline the
general algorithm for the direct stiness method.
Victor Saouma
Draft
312
3.5
Homework
1. Solve (analytically) for the three displacements, reactions, and the internal forces for the
frame in terms of P , L, H, E, I, A. Assume both members to have the same elastic and
cross-sectional properties (you may want to use Mathematica).
Solve for P = 10 kip, L = H = 12 ft, E =30,000 Ksi, A = 10 in2 , I = 200 in4 .
Note:
(a) Show on three separate gures all the (global) stiness terms.
(b) Determine the xed end actions.
(c) Assemble the global stiness matrix.
(d) Dene Z = AL2 /I, factor the matrix, and invert.
(e) You may either proceed with Mathematica from that point onward, or give up,
substitute the numerical values and solve for internal forces and reactions.
(f) Draw the free body diagram for each member.
(g) Draw the shear and moment diagram for the structure.
(h) SOlve for the reaction
2. Using Mathematica, derive the analytical expression fo the stiness matrix of a frame
element in global coordinate system.
Victor Saouma
Draft
Chapter 4
TRANSFORMATION MATRICES
4.1
4.1.1
Preliminaries
[ke ] [Ke ] Relation
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 straightforward.
1
The determination of the element stiness matrix in global coordinates, from the element
stiness matrix in local coordinates requires the introduction of a transformation.
2
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).
4
Recalling that
{p} = [k(e) ]{}
{P} = [K
(4.1)
(4.2)
(e)
]{}
Let us dene a transformation matrix [(e) ] such that:

{} = [(e) ]{}
(4.3)
{p} = [
(4.4)
(e)
]{P}
Note that we use the same matrix (e) since both {} and {p} are vector quantities (or tensors
of order one).
6
Substituting Eqn. 4.3 and Eqn. 4.4 into Eqn. 4.1 we obtain
[(e) ]{P} = [k(e) ][(e) ]{}
premultiplying by [(e) ]1
(4.5)
{P} = [(e) ]1 [k(e) ][(e) ]{}
(4.6)
But since the rotation matrix is orthogonal, we have [(e) ]1 = [(e) ]T and
{P} = [(e) ]T [k(e) ][(e) ]{}
[K(e) ]
(4.7)
Draft
42
[K(e) ] = [(e) ]T [k(e) ][(e) ]
(4.8)
which is the general relationship between element stiness matrix in local and global coordinates.
4.1.2
Direction Cosines
The dot product of two vectors
A B = |A|.|B| cos = Ax Bx + Ay By + Az Bz
(4.9)
where |A| is the norm (length) of A, and is the angle between the two vectors.
A vector V
V = Vx i + Vy j + Vz k
(4.10)
can be normalized
Vn =
Vy
Vz
x2 x1
y2 y1
z 2 z1
Vx
i+
j+
k=
i+
j+
k
|V |
|V |
|V |
L
L
L
(4.11)
where L = (x2 x1 )2 + (y2 y1 )2 + (z2 z1 )2 and hence to obtain the three direction cosines
of vector V we simply take the dot product of its normalized form with the three unit vectors
forming the orthogonal coordinate system
Vn i =
Vn j =
Vn k =
x2 x1
= lvx
L
y2 y1
= lvy
L
z 2 z1
= lvz
L
(4.12-a)
(4.12-b)
(4.12-c)
Note that since both vectors are normalized, the dot product is equal to the direction cosine.
The problem confronting us is the general transformation of a vector V from (X, Y, Z)
coordinate system to (x, y, z), Fig. 4.1: where:
10
Vx
lxX
V
= lyX
y
V
lzX
z
lxY
lyY
lzY
lxZ VX
lyZ VY
lzZ VZ
(4.13)
[ ]
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.
11
With respect to Fig. 4.2, lxX = cos ; lxY = cos , and lxZ = cos or
Vx = VX lxX + VY lxY + VZ lxZ
= VX cos + VY cos + VZ cos
Victor Saouma
(4.14-a)
(4.14-b)
Draft
4.1 Preliminaries
43
Y
y
VY
Vy
Vx
X
VX
Vz
VZ
Z
Figure 4.1: Arbitrary 3D Vector Transformation
VY
VX
VZ
Z
z
Figure 4.2: 3D Vector Transformation
Victor Saouma
Draft
44
12
Direction cosines are unit orthogonal vectors satisfying the following relations:
3
lij lij = 1
i = 1, 2, 3
(4.15)
j=1
i.e:
2
2
2
l11 + l12 + l13 = 1 or cos2 + cos2 + cos2 = 1 = 11
and
(4.16)
i = 1, 2, 3
lij lkj
= 0
j=1
k = 1, 2, 3
(4.17-a)
i=k
l11 l21 + l12 l22 + l13 l23 = 0 = 12
(4.17-b)
By direct multiplication of []T and [] it can be shown that: []T [] = [I] []T =
[]1 [] is an orthogonal matrix.
13
14
The reverse transformation (from local to global) would be

{V} = []T {v}
VX
or
Y
V
Z
lxX
= lxY
lxZ
lyX
lyY
lyZ
(4.18)
lzX Vx
lxY Vy
lzZ Vz
(4.19)
[ ]1 =[ ]T
4.2
Transformation Matrices For Framework Elements
The rotation matrix, [], will obviously vary with the element type. In the most general case
(3D element, 6 d.o.f. per node), we would have to dene:
15
Fx1
Fy1
Fz1
Mx1
My1
Mz1
Fx2
Fy2
Fz2
Mx2
My2
Mz2
[]
[]
[]
[]
FX1
FY 1
FZ1
MX1
MY 1
MZ1
FX2
FY 2
FZ2
MX2
MY 2
MZ2
(4.20)
[]
and should distinguish between the vector transformation [] and the element transformation
matrix [].
16
In the next sections, we will examine the transformation matrix of each type of element.
Victor Saouma
Draft
4.2 Transformation Matrices For Framework Elements
4.2.1
4.2.1.1
17
45
2 D cases
2D Frame
For the 2D frame element, Fig. 4.3 and from Eq. 4.13 the vector rotation matrix is dened
Y
5
x
4
X
1
Figure 4.3: 2D Frame Element Rotation with respect to Z (or z)

in terms of 9 direction cosines of 9 dierent angles. However for the 2D case, we will note that
four angles are interrelated (lxX , lxY , lyX , lyY ) and can all be expressed in terms of a single one
, where is the direction of the local x axis (along the member from the rst to the second
node) with respect to the global X axis.
18 The remaining 5 terms are related to the angle between Z and z, which is zero, and the Z
x y plane, which is /2. Thus the rotation matrix can be written as:
lxX
[] = lyX
0
0
cos
cos( ) 0
cos sin 0
2

0 = cos( 2 + )
cos
0 = sin cos 0
1
0
0
1
0
0
1
lxY
lyY
0
(4.21)
and we observe that the angles are dened from the second subscript to the rst, and that
counterclockwise angles are positive.
19
The element rotation matrix [] will then be given by
p1
p2
3
=
p4
p5
p6
cos sin 0
0
0
0
0
0
0
sin cos 0
0
0
1
0
0
0
0
0
0 cos sin 0
0
0
0 sin cos 0
0
0
1
0
0
0
P1
P2
P3
P4
P5
P6
(4.22)
[]
4.2.1.2
Grid
20 The transformation matrix for the grid element, Fig. 4.4 is very analogous to the one of the
frame element except that rotation is with respect to the y axis instead of the z axis.
21 From Eq. 4.13 the vector rotation matrix is dened in terms of 9 direction cosines of 9
dierent angles. However, we will note that four angles are interrelated (lxX , lxZ , lzX , lzZ ) and
can all be expressed in terms of a single one , where is the direction of the local x axis
(along the member from the rst to the second node) with respect to the global X axis.
Victor Saouma
Draft
46
Y y
6
2
Figure 4.4: Grid Element Rotation

The remaining 5 terms are related to the angle between Y and y, which is zero, and the Y
x z plane, which is /2. Thus the rotation matrix can be written as:
22
lxX
[] = 0
lzX
0 lxZ
cos
0 cos( )
cos 0 sin
2

0
1
0
0
1
0
1 0 =
=
cos( 2 + ) 0
sin 0 cos
0 lzZ
cos
(4.23)
and we observe that the angles are dened from the second subscript to the rst, and that
counterclockwise angles are positive.
23
The element rotation matrix [] will then be given by
p1
p2
3
=
p4
p5
p6
cos
0
sin
0
0
0
0 sin
0
0
1
0
0 cos
0
0
0
cos
0
0
0
sin
0
0
0
0
0
0
0
0
0 sin
1
0
0 cos
P1
P2
P3
P4
P5
P6
(4.24)
[]
4.2.1.3
2D Truss
24 For the 2D truss element, the global coordinate system is two dimensional, whereas the local
one is only one dimensional, hence the vector transformation matrix is, Fig. 4.5.
2
Y
11
00
11
00
11
00
1
0
1
0
1
0
Figure 4.5: 2D Truss Rotation
Victor Saouma
Draft

[] =
25
lxX
lxY
cos cos( )
2
47
cos sin
(4.25)
The element rotation matrix [] will then be assembled from the vector rotation matrix [].
p1
p2
[] 0
0 []
P1
cos sin
0
0
0
0
cos sin
P3
P4
[]
4.2.2
(4.26)
3D Frame
Given that rod elements, are dened in such a way to have their local x axis aligned with
their major axis, and that the element is dened by the two end nodes (of known coordinates),
then recalling the denition of the direction cosines it should be apparent that the evaluation
of the rst row, only, is quite simple. However evaluation of the other two is more complex.
26
This generalized transformation from X, Y, Z to x, y, z was accomplished in one step in the

two dimensional case, but intermediary ones will have to be dened in the 3D case.
27
Starting with reference (X1 , Y1 , Z1 ) coordinate system which corresponds to the global coordinate system, we seek to transform an element originally along the X1 axis into its new
position along the X2 axis.
28
29 In the 2D case this was accomplished through one single rotation , and all other angles
where dened in terms of it.
In the 3D case, it will take a minimum of two rotations and , and possibly a third one
(dierent than the one in 2D) to achieve this transformation, Table 4.1. Those three angles
are also called the Eulerian angles.
30
From
X1 , Y1 , Z1
X , Y , Z
X , Y , Z
To
With respect to
X , Y , Z
Y1 Y
X , Y , Z
Z Z
Optional
X , Y , Z
X X
Angle
Table 4.1: 3D Transformations of Linear Elements

We can start with the rst row of the transformation matrix which corresponds to the
direction cosines of the reference axis (X1 , Y1 , Z1 ) with respect to X2 . This will dene the
rst row of the vector rotation matrix []:
31
CX
[] = l21
l31
where CX =
xj xi
L ,
Victor Saouma
CY =
yj yi
L ,
CZ =
zj zi
L ,
CY
l22
l32
L=
CZ
l23
l33
(4.27)
(xj xi )2 + (yj yi )2 + (zj zi )2 .
Draft
48
32 Note that this does not uniquely dene the new coordinate system. This will be achieved in
two ways: a reduced and a generalized one.
33
In the simple rotation, Fig. 4.6, the vertical axis of the member remains vertical, and the
Y
Y1 , Y
Y2
X2
j
i
L CY
X1
L CZ
L CX
Z2 , Z
CX
C XZ
Z
1
Z1
CZ
CY
C XZ
Figure 4.6: Reduced 3D Rotation

last two rows of Eq. 4.27 can be determined through two successive rotations (assuming that
(X1 , Y1 , Z1 and X2 , Y2 , Z2 are originally coincident):
1. Rotation by about the Y1 axis, this will place the X1 axis along X . This rotation [R ]
is made of the direction cosines of the axis (X , Y , Z ) with respect to (X1 , Y1 , Z1 ):
cos 0 sin
0
1
0
[R ] =
sin 0 cos
we note that: cos =
CX
CXZ ,
sin =
CZ
CXZ ,
2. Rotation by about the Z2 axis
and CXZ =
(4.28)
2
2
CX + CZ .
cos sin 0
[R ] = sin cos 0
0
0
1
(4.29)
where cos = CXZ , and sin = CY .

34
Combining Eq. 4.28 and 4.29 yields:
[] = [R ][R ] =
CX
CX CY
CXZ
CZ
CXZ
CY
CXZ
0
CZ
CY CZ
CXZ
CX
CXZ
(4.30)
For vertical member the preceding matrix is no longer valid as CXZ is undened. However
we can obtain the matrix by simple inspection, Fig. 4.7 as we note that either:
35
1. X2 axis is aligned with Y1

2. Y2 axis is aligned with -X1
Victor Saouma
Draft

Y1 X2
49
Y1
Z2
j
=90
Y2
X1
Z1 Z2
=270
Y2
X1
Z1
X2
Figure 4.7: Special Case of 3D Transformation for Vertical Members

3. Z2 axis is aligned with Z1
or
1. X2 axis is aligned with -Y1
2. Y2 axis is aligned with -X1
3. Z2 axis is aligned with -Z1
hence the rotation matrix with respect to the y axis, is similar to the one previously derived
for rotation with respect to the z axis, except for the reordering of terms:
[] = CY
0
CY
0
0
0
1
(4.31)
which is valid for both cases (CY = 1 for = 90 deg, and CY = 1 for = 270 deg).
36 In the generalized case, we need to dene an additional rotation to the preceding transformation of an angle about the X axis, Fig. 4.8.
This rotation is dened such that:
1. X is aligned with X2 and normal to both Y2 and Z2

2. Y makes an angle 0, and =
3. Z makes an angle 0,
, with respect to X2 , Y2 and Z2 respectively
+ and , with respect to X2 , Y2 and Z2 respectively
Noting that cos( + ) = sin and cos = sin , the direction cosines of this transforma2
tion are given by:
1
0
0
[R ] = 0 cos sin
(4.32)
0 sin cos
37
causing the Y2 Z2 axis to coincide with the principal axes of the cross section. This will yield:
Victor Saouma
Draft
410
Y
Y
Y Y
1
Z Z
X
Z
X X
Figure 4.8: Complex 3D Rotation
Victor Saouma
Draft
411
[] = [R ][R ][R ]
CX
CX CY cos CZ sin
CXZ
CX CY sin CZ cos
CXZ
CY
CXZ cos
CXZ sin
CZ
CY CZ cos +CX sin
CXZ
CY CZ sin +CX cos
CXZ
(4.33)
(4.34)
38 As for the simpler case, the preceding equation is undened for vertical members, and a
counterpart to Eq. 4.31 must be derived. This will be achieved in two steps:
1. Rotate the member so that:

(a) X2 axis aligned with Y1
(b) Y2 axis aligned with -X1
(c) Z2 axis aligned with Z1
this was previously done and resulted in Eq. 4.31
[R ] = CY
0
CY
0
0
0
1
(4.35)
2. The second step consists in performing a rotation of angle with respect to the new X2
as dened in Eq. 4.32.
3. Finally, we multiply the two transformation matrices [R ][R ] given by Eq. 4.35 and 4.32
to obtain:
0
CY
0
[] = [R ][R ] = CY cos 0 sin

(4.36)
CY sin
0 cos
Note with = 0, we recover Eq. 4.31.
4.2.3
3D Truss
With reference to the rst part of the derivation of the transformation of 3D frame element,
the transformation matrix of 3D truss elements is
39
[3D T ] =
Victor Saouma
CX
0
CY
0
CZ
0
0
CX
0
CY
0
CZ
(4.37)
Draft
412
Victor Saouma
Draft
Chapter 5
STIFFNESS METHOD; Part II

5.1
5.1.1
Direct Stiness Method

Global Stiness Matrix
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.
1
For instance, with reference to Fig. 5.1, we have three global degrees of freedom, 1 , 2 , and
11
00 A
11
00
11
00 A
11
00
P/2
EI
w
L/2
2
B
C
111
000
111
000
K22
1
K21
11
00
1
0A
1
0
1
0
11
00
L/2
1
0A
1
0
K31
11
00A
11
00
11
00
K
32
K23
K12
B
1
0
1
0
K11
C
11
00
11
00
Figure 5.1: Frame Example
111
000
K 33
K
13
1
Draft
52
3 . and the global (restrained or structures) stiness matrix is
K11 K12 K13
K = K21 K22 K23

K31 K32 K33
(5.1)
and the rst column corresponds to all the internal forces in the unrestrained d.o.f. when a
unit displacement along global d.o.f. 1 is applied.
5.1.1.1
Structural Stiness Matrix
3 The structural stiness matrix is assembled only for those active degrees of freedom which
are active (i.e unrestrained). It is the one which will be inverted (or rather decomposed) to
determine the nodal displacements.
5.1.1.2
Augmented Stiness Matrix
The augmented stiness matrix is expressed in terms of all the dof. However, it is partitioned into two groups with respective subscript u where the displacements are known (zero
otherwise), and t where the loads are known.
4
Pt
Ru ?
Ktt Ktu
Kut Kuu
t ?
(5.2)
We note that Ktt corresponds to the structural stiness matrix.

We will see later that the two types of loads (or displacements) will correspond to essential
and natural boundary conditions.
5
The rst equation enables the calculation of the unknown displacements.

t = K1 (Pt Ktu u )
tt
(5.3)
The second equation enables the calculation of the reactions

Ru = Kut t + Kuu u
(5.4)
8 For internal book-keeping purpose, since we are assembling the augmented stiness matrix,
we proceed in two stages:
1. First number all the global unrestrained degrees of freedom

2. Then number separately all the global restrained degrees of freedom (i.e. those with
known displacements, zero or otherwise), multiplied by -1, this will enable us later on to
distinguish the restrained from unrestrained dof.
Victor Saouma
Draft
5.2 Logistics
53
9 The element internal forces (axial and shear forces, and moment at each end of the member)
are determined from
(e)
pint = k(e) (e)
(5.5)
(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
Logistics
5.2.1
Boundary Conditions, [ID] Matrix
10 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:
Victor Saouma
Draft
54

4
11
4
10
[k]
[K]
Figure 5.2: Example for [ID] Matrix Determination

Node No.
1
2
3
4
[ID]T
000
110
000
100
2. At this stage, the [ID] matrix is equal to:
0 1 0 1
ID = 0 1 0 0
0 0 0 0
(5.7)
3. After we determined the equation numbers, we would have:
1 10 5 12
8
ID = 2 11 6
3
4
7
9
5.2.2
(5.8)
LM Vector
13 The LM vector of a given element gives the global degree of freedom of each one of the element
degree of freedoms. For the structure shown in Fig. 5.2, we would have:
LM
LM
LM
5.2.3
=
=
=
10 11 4 5 6 7
5 6 7 1 2 3
1 2 3 12 8 9
element 1 (2 3)
element 2 (3 1)
element 3 (1 4)
Assembly of Global Stiness Matrix
As for the element stiness matrix, the global stiness matrix [K] is such that Kij is the
force in degree of freedom i caused by a unit displacement at degree of freedom j.
14
Victor Saouma
Draft
5.2 Logistics
55
15 Whereas this relationship was derived from basic analysis at the element level, at the structure level, this term can be obtained from the contribution of the element stiness matrices
[K(e) ] (written in global coordinate system).
16 For each Kij term, we shall add the contribution of all the elements which can connect degree
of freedom i to degree of freedom j, assuming that those forces are readily available from the
individual element stiness matrices written in global coordinate system.
17
Kij is non-zero if degree of freedom i and degree of freedom j

1. Are connected by an element.
2. Share a node.
3. Are connected by an element and the corresponding value in the element stiness matrix
in the global coordinate system is zero.
18 There are usually more than one element connected to a dof. Hence, individual element
stiness matrices terms must be added up.
Because each term of all the element stiness matrices must nd its position inside the global
stiness matrix [K], it is found computationally most eective to initialize the global stiness
matrix [KS ](N EQAN EQA ) to zero, and then loop through all the elements, and then through
19
(e)
each entry of the respective element stiness matrix Kij .

(e)
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
S
columns of the 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).
20
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
4 kN/m
2
8m
11
00
11
00
11
00
11
00
3m
1
111
000
111
000
111
000
111
000
1
111
000
7.416 m
8m
Figure 5.3: Simple Frame Analyzed with the MATLAB Code
Victor Saouma
Draft
56
The ID matrix is initially set to:
1 0 1
[ID] = 1 0 1
1 0 1
(5.9)
We then modify it to generate the global degrees of freedom of each node:
4 1 7
[ID] = 5 2 8
6 3 9
(5.10)
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 5 6 1
2
3
1
2
3 7 8 9
(5.11)
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
A16
A26
A36
A46
A56
A66
1
1 B11
2 B21
3 B31
7 B41
8 B51
9 B61
2
B12
B22
B32
B42
B52
B62
3
B13
B23
B33
B43
B53
B63
7
B14
B24
B34
B44
B54
B64
8
B15
B25
B35
B45
B55
B65
B16
B26
B36
B46
B56
B66
K (1) =
K (2) =
(5.12-a)
(5.12-b)
We note that for each element we have shown the corresponding LM vector.
Now, we assemble the global stiness matrix
K=
A44 + B11 A45 + B12 A46 + B13

A54 + B21 A55 + B22 A56 + B23
A64 + B31 A65 + B32 A66 + B33
A14
A15
A16
A24
A25
A26
A35
A36
A34
B41
B42
B43
B51
B52
B53
B61
B62
B63
A41
A51
A61
A11
A21
A31
0
0
0
A42
A52
A62
A12
A22
A32
0
0
0
A43
A53
A63
A13
A23
A33
0
0
0
B14
B24
B34
0
0
0
B44
B54
B64
B15
B25
B35
0
0
0
B45
B55
B65
B16
B26
B36
0
0
0
B46
B56
B66
(5.13)
We note that some terms are equal to zero because we do not have a connection between
the corresponding degrees of freedom (i.e. node 1 is not connected to node 3).
Victor Saouma
Draft
5.2 Logistics
5.2.4
23
57
Algorithm
The direct stiness method can be summarized as follows:
Preliminaries: First we shall

1. Identify type of structure (beam, truss, grid or frame) and determine the
(a) Number of spatial coordinates (1D, 2D, or 3D)
(b) Number of degree of freedom per node (local and global)
(c) Number of cross-sectional and material properties
2. Determine the global unrestrained and restrained degree of freedom equation numbers for each node, Update the [ID] matrix (which included only 0s and 1s in the
input data le).
Analysis :
1. For each element, determine
(a)
(b)
(c)
(d)
(e)
Vector LM relating local to global degree of freedoms.

Element stiness matrix [k(e) ]
Angle between the local and global x axes (3D analysis).
Rotation matrix [(e) ]
Element stiness matrix in global coordinates [K(e) ] = [(e) ]T [k(e) ][(e) ]
2. Assemble the augmented stiness matrix [K(S) ] of unconstrained and constrained

degree of freedoms.
3. Extract [Ktt ] from [K(S) ] and invert (or decompose into into [Ktt ] = [L][L]T where
[L] is a lower triangle matrix.
4. Assemble load vector {P} in terms of nodal load and xed end actions.
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]{}.
24
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.
Solution:
1. Determine the structure ID matrix
Node #
Victor Saouma
1
2
3
4
5
Bound. Cond.
X
Y
0
1
0
0
1
1
0
0
0
0
Draft
58

4
5
1
6
8
50k
12
100k
16
16
Figure 5.4:
0 0 1 0 0
1 0 1 0 0
ID =
N ode 1 2
1 2
8 3
(5.14-a)
3
4 5
9 4 6
10 5 7
(5.14-b)
2. The LM vector of each element is evaluated next

LM
1 8 4 5
(5.15-a)
LM
1 8 2 3
(5.15-b)
LM
2 3 4 5
(5.15-c)
LM
4 5 6 7
(5.15-d)
LM
9 10 4 5
(5.15-e)
LM
2 3 6 7
(5.15-f)
LM
2 3 9 10
(5.15-g)
LM
9 10 6 7
(5.15-h)
3. Determine the element stiness matrix of each element in the global coordinate system noting
that for a 2D truss element we have
[K (e) ] = [(e) ]T [k(e) ][(e) ]
2
c
cs c2 cs
EA cs
s2 cs s2
2 cs
c2
cs
L c
cs s2 cs
s2
where c = cos =
Victor Saouma
x2 x1
L ;
s = sin =
(5.16-a)
(5.16-b)
Y2 Y1
L
Draft
5.2 Logistics
59
Element 1 L = 20 , c =
160
20
120
20
= 0.8, s =
= 0.6,
EA
L
(30,000
ksi)(10 in2 )
20
= 15, 000 k/ft.
1
8
4
5
1
9, 600
7200
9, 600 7, 200
8 7, 200
5, 400 7, 200 5, 400
4 9, 600 7, 200 9, 600

7, 200
5 7, 200 5, 400 7, 200
5, 400
[K1 ] =
Element 2 L = 16 , c = 1 , s = 0 ,
EA
L
= 18, 750 k/ft.
1
8
1
18, 750
0
8
0
0
2 18, 750 0
3
0
0
[K2 ] =
Element 3 L = 12 , c = 0 , s = 1 ,
EA
L
2 0
3 0
4 0
5 0
[K3 ] =
Element 4 L = 16 , c = 1 , s = 0 ,
(5.17)
(5.18)
= 25, 000 k/ft.
3
0
25, 000
0
25, 000
EA
L
2
3
18, 750 0
0
0
18, 750 0
0
0
4
5
0
0
0 25, 000
0
0
0 25, 000
(5.19)
= 18, 750 k/ft.
4
5
6
7
4 18, 750 0 18, 750 0

5
0
0
0
0
6 18, 750 0 18, 750 0

7
0
0
0
0
[K4 ] =
Element 5 L = 20 , c =
160
20
= 0.8 , s = 0.6 ,
EA
L
(5.20)
= 15, 000 k/ft.
9
10
4
5
9
9, 600 7, 200 9, 600 7, 200
10 7, 200 5, 400
7, 200 5, 400
4 9, 600 7, 200
9, 600 7, 200
5
7, 200 5, 400 7, 200 5, 400
[K5 ] =
Element 6 L = 20 , c = 0.8 , s = 0.6 ,
[K6 ] =
Victor Saouma
EA
L
(5.21)
= 15, 000 k/ft.
2
3
6
7
2 9, 600
7, 200 9, 600 7, 200
3 7, 200
5, 400 7, 200 5, 400
6 9, 600 7, 200 9, 600

7, 200
7 7, 200 5, 400 7, 200
5, 400
(5.22)
Draft
510
Element 7 L = 16 , c = 1 , s = 0 ,
EA
L
= 18, 750 k/ft.
2
3
9
10
2
18, 750 0 18, 750
0
3
0
0
0
0
9 18, 750 0 18, 750

0
10
0
0
0
0
[K7 ] =
Element 8 L = 12 , c = 0 , s = 1 ,
EA
L
9
9
0
10 0
6 0
7
0
[K8 ] =
(5.23)
= 25, 000 k/ft.

10
0
25, 000
0
25, 000
6
7
0
0
0 25, 000
0
0
0 25, 000
(5.24)
4. Assemble the global stiness matrix in k/ft Note that we are not assembling the augmented
stiness matrix, but rather its submatrix [Ktt ].
0
0
100k
0
0
50k
0
9, 600 + 18, 750
18, 750
9, 600 + (2) 18, 750
9, 600
0
0
18, 750 + (2)9, 600
0
7, 200
5, 400 + 25, 000
SYMMETRIC
7, 200
0
25, 000
7, 200 7, 200
25, 000 + 5, 400(2)
0
9, 600
7, 200
18, 750
0
18, 750 + 9, 600
0
7, 200
5, 400
0
0
7, 200
25, 000 + 5, 400
(5.25)
5. Convert to k/in and simplify
100
0
0
50
0
2, 362.5
=

SYMMETRIC

1, 562.5
3, 925.0
0
600
2, 533.33
Pt
800
600
0
0
0
800
0
2, 083.33
600
3, 162.5
0
1, 562.5
2, 983.33
0
2, 362.5
0
600
450
0
0
600
2, 533.33
(5.26)
6. Invert stiness matrix and solve for displacements
0.0223 in
0.00433 in
0.116 in
U4
0.0102 in
=
V 0.0856 in
U6 0.00919 in
V 0.0174 in
7
U1
U2
V
3
(5.27)
7. Solve for member internal forces (in this case axial forces) in local coordinate systems
c
s c s
c s
c
s
U1
V
1
U2
(5.28)
V2
Victor Saouma
U1
U2
V3
U4
V5
U6
V7
ut
Ktt
u1
u2
u1
u2
v3
u4
v5
u6
v7
Draft
5.2 Logistics
511
Element 1
p1
p2
= (15, 000 k/ft)(
52.1 k
52.1 k
1 ft
)
12 in
0.8
0.6 0.8 0.6
0.8 0.6
0.8
0.6
0.0223
(5.29-a)
0.0102
0.0856
Compression
(5.29-b)
Element 2
p1
p2
= 18, 750 k/ft(
1 ft
)
12 in
43.2 k
43.2 k
1 0 1 0
1 0
1 0
0.0233
0.00433
(5.30-a)
0.116
Tension
(5.30-b)
Element 3
p1
p2
= 25, 000 k/ft(
1 ft
)
12 in
63.3 k
63.3 k
0
1 0 1
0 1 0
1
0.00433
0.116
0.0102
(5.31-a)
0.0856
Tension
(5.31-b)
Element 4
p1
p2
= 18, 750 k/ft(
1 ft
)
12 in
1.58 k
1.58 k
1 0 1 0
1 0
1 0
0.0102
0.0856
0.00919
(5.32-a)
0.0174
Tension
(5.32-b)
Element 5
p1
p2
= 15, 000 k/ft(
=
Victor Saouma
54.0 k
54.0 k
1 ft
)
12 in
0.8
0.6
0.8 0.6
0.8 0.6 0.8
0.6
Compression
0.0102
0.0856
0
0
(5.33-a)
(5.33-b)
Draft
512
Element 6
p1
p2
= 15, 000 k/ft(
1 ft
)
12 in
60.43 k
60.43 k
0.00433
0.8
0.6 0.8 0.6
0.8 0.6
0.8
0.6
0.116
(5.34-a)
0.00919
0.0174
Tension
(5.34-b)
Element 7
p1
p2
= 18, 750 k/ft(
6.72 k
6.72 k
1 ft
)
12 in
1 0 1 0
1 0
1 0
0.00433
0.116
0
0
Compression
(5.35-a)
(5.35-b)
Element 8
p1
p2
= 25, 000 k/ft(
36.3 k
36.3 k
1 ft
)
12 in
0
1 0 1
0 1 0
1
0
0
0.00919
0.0174
Compression
8. Determine the structures MAXA vector
1 3
9 14
2 5 8 13 19
4 7 12 18
6 11 17
[K] =
10 16
15
25
24
23
22
21
20
(5.36-a)
(5.36-b)
MAXA =
6
10
15
20
(5.37)
Thus, 25 terms would have to be stored.
Example 5-3: Analysis of a Frame with MATLAB

Victor Saouma
Draft
5.2 Logistics
513
The simple frame shown in Fig. 5.5 is to be analyzed by the direct stiness method. Assume:
E = 200, 000 MPa, A = 6, 000 mm2 , and I = 200 106 mm4 . The complete MATLAB solution
is shown below along with the results.
50kN
4 kN/m
2
8m
1
0
1
0
1
0
1
0
3m
1
111
000
111
000
111
000
111
0001
111
000
7.416 m
8m
Figure 5.5: Simple Frame Analyzed with the MATLAB Code
% zero the matrices

k=zeros(6,6,2); K=zeros(6,6,2); Gamma=zeros(6,6,2);
% Structural properties units: mm^2, mm^4, and MPa(10^6 N/m)
A=6000;II=200*10^6;EE=200000;
% Convert units to meter and kN
A=A/10^6;II=II/10^12;EE=EE*1000;
% Element 1
i=[0,0];j=[7.416,3]; [k(:,:,1),K(:,:,1),Gamma(:,:,1)]=stiff(EE,II,A,i,j);
% Element 2
i=j;j=[15.416,3]; [k(:,:,2),K(:,:,2),Gamma(:,:,2)]=stiff(EE,II,A,i,j);
% Define ID matrix
ID=[
-4 1 -7;
-5 2 -8;
-6 3 -9];
% Determine the LM matrix
LM=[
-4 -5 -6 1 2 3;
1 2 3 -7 -8 -9];
% Assemble augmented stiffness matrix
Kaug=zeros(9); for elem=1:2
for r=1:6
lr=abs(LM(elem,r));
for c=1:6
lc=abs(LM(elem,c));
Kaug(lr,lc)=Kaug(lr,lc)+K(r,c,elem);
end
end
end
% Extract the structures Stiffness Matrix
Ktt=Kaug(1:3,1:3);
% Determine the fixed end actions in local coordinate system
fea(1:6,1)=0; fea(1:6,2)=[0,8*4/2,4*8^2/12,0,8*4/2,-4*8^2/12];
% Determine the fixed end actions in global coordinate system
Victor Saouma
Draft
514
FEA(1:6,1)=Gamma(:,:,1)*fea(1:6,1); FEA(1:6,2)=Gamma(:,:,2)*fea(1:6,2);
% 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;
0;
This simple proigram will produce the following results:

Displacements =
0.0010
-0.0050
Victor Saouma
Draft
5.2 Logistics
515
-0.0005
Reactions =
130.4973
55.6766
13.3742
-149.2473
22.6734
-45.3557
int_forces =
int_forces =
141.8530
2.6758
13.3742
-141.8530
-2.6758
8.0315
149.2473
9.3266
-8.0315
-149.2473
22.6734
-45.3557
We note that the internal forces are consistent with the reactions (specially for the second node
of element 2), and amongst themselves, i.e. the moment at node 2 is the same for both elements
(8.0315).
Example 5-4: Analysis of a simple Beam with Initial Displacements

The full stiness matrix of a beam element is given by
v1
V1
12EI/L3
6EI/L2
M1
V2 12EI/L3
M2 6EI/L2
[ke ] =
1
6EI/L2
4EI/L
6EI/L2
2EI/L
v2
12EI/L3
6EI/L2
12EI/L3
6EI/L2
6EI/L2
2EI/L
6EI/L2
4EI/L
(5.38)
This matrix is singular, it has a rank 2 and order 4 (as it embodies also 2 rigid body motions).
25
We shall consider 3 dierent cases, Fig. 5.6
Cantilivered Beam/Point Load

1. The element stiness matrix is
k=
Victor Saouma
3 12EI/L3
4 6EI/L2
1 12EI/L3
2
6EI/L2
4
6EI/L2
4EI/L
6EI/L2
2EI/L
1
12EI/L3
6EI/L2
12EI/L3
6EI/L2
6EI/L2
2EI/L
6EI/L2
4EI/L
(5.39)
Draft
516
-3
1
0
1
0
1
0
1
0
-2

P
-4
-3
-4
-3
1
0
1
0
1
0
1
0
-4
Figure 5.6: Stiness Analysis of one Element Structure

2. The structure stiness matrix is assembled
K=
1
12EI/L2
2 6EI/L2
3 12EI/L3
4 6EI/L2
2
6EI/L2
4EI/L
6EI/L2
2EI/L
3
12EI/L3
6EI/L2
12EI/L3
6EI/L2
6EI/L2
2EI/L
6EI/L2
4EI/L
(5.40)
3. The global matrix can be rewritten as
6EI/L2
4EI/L
6EI/L2
2EI/L
12EI/L2
0
6EI/L2
=
12EI/L3
R3 ?
6EI/L2
R4 ?
12EI/L3
6EI/L2
12EI/L3
6EI/L2
6EI/L2
1 ?
2EI/L
2 ?
2
6EI/L
3
4
4EI/L
(5.41)
4. Ktt is inverted (or actually decomposed) and stored in the same global matrix
L3 /3EI
L2 /2EI
12EI/L3
6EI/L2
L2 /2EI
12EI/L3
6EI/L2
L/EI
6EI/L2
2EI/L
6EI/L2
2EI/L
12EI/L3
6EI/L2
6EI/L2
4EI/L
(5.42)
5. Next we compute the equivalent load, Pt = Pt Ktu u , and overwrite Pt by Pt
Pt Ktu u
0
0
0
0
0
0
Victor Saouma
L3 /3EI
L2 /2EI
12EI/L3
L2 /2EI
12EI/L3
6EI/L2
L/EI
6EI/L2
2EI/L
12EI/L3
6EI/L2
6EI/L2
4EI/L
6EI/L2
6EI/L
2EI/L
Draft
5.2 Logistics
517
6. Now we solve for the displacement t = K1 Pt , and overwrite Pt by t

tt
2
0
0
L2 /2EI
L2 /2EI
12EI/L3
6EI/L2
2EI/L
12EI/L3
6EI/L2
6EI/L2
4EI/L
2
6EI/L
P L3 /3EI
12EI/L3
L/EI
L3 /3EI
P L2 /2EI
0
0
6EI/L
2EI/L
6EI/L2
7. Finally, we solve for the reactions, Ru = Kut tt + Kuu u , and overwrite u by Ru
3
P L2 /3EI
P L /2EI
R3
R4
12EI/L3
6EI/L2
6EI/L2
12EI/L3
2EI/L
6EI/L2
1
6EI/L2
4EI/L
6EI/L2
2EI/L
4
12EI/L3
6EI/L2
12EI/L3
6EI/L2
12EI/L3
6EI/L2
L2 /2EI
L/EI
L3 /3EI
L2 /2EI
3
P L /3EI
P L2 /2EI
P
PL
P L3 /3EI
6EI/L2
2EI/L
P L2 /2EI
2
6EI/L
4EI/L
0
Simply Supported Beam/End Moment

k=
3 12EI/L3
1 6EI/L2
4 12EI/L3
2
6EI/L2
6EI/L2
2EI/L
6EI/L2
4EI/L
(5.43)
K=
1
4EI/L
2 2EI/L
3 6EI/L2
4 6EI/L2
2
2EI/L
4EI/L
6EI/L2
6EI/L2
3
6EI/L2
6EI/L2
12EI/L3
12EI/L3
6EI/L2
2
6EI/L
12EI/L3
12EI/L3
(5.44)
3. The global stiness matrix can be rewritten as

0
4EI/L
2EI/L
=
6EI/L2
R ?
3
6EI/L2
R4 ?
4. Ktt is inverted
L3 /3EI
L/6EI
6EI/L2
6EI/L2
Victor Saouma
2EI/L
4EI/L
6EI/L2
6EI/L2
6EI/L2
6EI/L2
12EI/L3
12EI/L3
6EI/L2
1 ?
6EI/L2
2 ?
12EI/L3
3
4
12EI/L3
(5.45)
L/6EI
6EI/L2
6EI/L2
L/3EI
6EI/L2
6EI/L2
6EI/L2
12EI/L3
12EI/L3
6EI/L2
12EI/L3
12EI/L3
(5.46)
Draft
518
5. We compute the equivalent load, Pt = Pt Ktu u , and overwrite Pt by Pt

Pt Ktu u
0
M
0
0
M
0
L/6EI
6EI/L2
6EI/L2
L/3EI
6EI/L2
6EI/L2
6EI/L2
6EI/L2
12EI/L3
12EI/L3
12EI/L3
12EI/L3
L3 /3EI
L/6EI
6EI/L2
6EI/L2
6. Solve for the displacements, t = K1 Pt , and overwrite Pt by t

tt
2
0
L/6EI
L/6EI
6EI/L2
6EI/L2
L/3EI
L3 /3EI
6EI/L2
6EI/L2
6EI/L2 6EI/L2
M L/6EI
M L/3EI
0
0
12EI/L3
12EI/L3
M
6EI/L2
12EI/L3 0
6EI/L2
12EI/L3
7. Solve for the reactions, Rt = Kut tt + Kuu u , and overwrite u by Ru
M L/6EI
M L/3EI
R1
R2
L/6EI
L/3EI
6EI/L2
6EI/L2
6EI/L2
6EI/L2
6EI/L2
6EI/L2
12EI/L3
12EI/L3
6EI/L
L3 /3EI
L/6EI
6EI/L
12EI/L
M L/6EI
M L/3EI
M/L
M/L
12EI/L
M L/6EI
M L/3EI
Cantilivered Beam/Initial Displacement and Known Displacements

2
2 12EI/L3
3 6EI/L2
4 12EI/L3
1
6EI/L2
k=
3
6EI/L2
4EI/L
6EI/L2
2EI/L
4
12EI/L3
6EI/L2
12EI/L3
6EI/L2
6EI/L2
2EI/L
6EI/L2
4EI/L
3
2EI/L
6EI/L2
4EI/L
6EI/L2
6EI/L2
12EI/L3
6EI/L2
12EI/L3
(5.47)

1
1
4EI/L
2 6EI/L2
3 2EI/L
4 6EI/L2
K=
2
6EI/L2
12EI/L3
6EI/L2
12EI/L3
3. The global matrix can be rewritten as

M
4EI/L
R2 ?
6EI/L2
=
2EI/L
R3 ?
6EI/L2
R4 ?
Victor Saouma
6EI/L2
12EI/L3
6EI/L2
12EI/L3
2EI/L
6EI/L2
4EI/L
6EI/L2
(5.48)
6EI/L2
1 ?

12EI/L3
2
6EI/L2 3

4
12EI/L3
(5.49)
Draft
5.3 Skyline Storage of Global Stiness Matrix, MAXA Vector
519
4. Ktt is inverted (or actually decomposed) and stored in the same global matrix
6EI/L2
L/4EI
6EI/L2
2EI/L
6EI/L
12EI/L3
6EI/L2
12EI/L3
6EI/L2
4EI/L
6EI/L2
6EI/L2
2EI/L
2
12EI/L3
6EI/L2
12EI/L3
(5.50)
5. Next we compute the equivalent load, Pt = Pt Ktu u , and overwrite Pt by Pt
Pt Ktu u
M + 6EI0 /L2
2EI/L
6EI/L2
12EI/L3
6EI/L2
12EI/L3
2
6EI/L
2EI/L
6EI/L2
0
0
0
6EI/L2
6EI/L2
4EI/L
6EI/L2
12EI/L3
6EI/L2
12EI/L3
L/4EI
0
0
0
L/4EI
6EI/L2
2EI/L
6EI/L2
12EI/L3
6EI/L2
12EI/L3
6EI/L2
4EI/L
6EI/L2
6EI/L2
2EI/L
M L/4EI + 30 /2L
0
0
0
6. Now we solve for the displacements, t = K1 Pt , and overwrite Pt by t

tt
1
0
0
M + 6EI0 /L2
3
12EI/L
0
6EI/L2
0
3
0
6EI/L2
12EI/L
0
0
0
7. Finally, we solve for the reactions, Rt = Kut tt + Kuu u , and overwrite u by Ru
M L/4EI + 30 /2L
R2
R3
R4
2EI/L
6EI/L2
12EI/L3
6EI/L2
12EI/L3
2EI/L
6EI/L2
4EI/L
6EI/L2
6EI/L2
12EI/L3
6EI/L2
12EI/L3
M L/4EI + 30 /2L
25
6EI/L2
5.3
6EI/L2
L/4EI
0
0
M L/4EI + 30 /2L
3M/2L 3EI0 /L3
M/2 3EI0 /L2
3M/2L + 3EI0 /L3
Skyline Storage of Global Stiness Matrix, MAXA Vector
The stiness matrix of a structure will be a square matrix of dimension NEQxNEQ.
Victor Saouma
Draft
520
Figure 5.7: Example of Bandwidth

We rst observe that the matrix is symmetric, thus only the upper half needs to be stored.
Furthermore, we observe that this matrix has a certain bandwidth, BW, dened as | Kij
Kii |max , when Kij = 0, Fig. 5.7.
26
Thus, we could as a rst space saving solution store the global stiness matrix inside a
rectangular matrix of length NEQ and width BW, which can be obtained from the LM vector
(largest dierence of terms of LM for all the elements).
27
It is evident that numbering of nodes is extremely important as it controls the size of the
bandwidth, and hence the storage requirement, Fig. 5.8. In this context, we observe that the
stiness matrix really has a variable bandwidth, or variable skyline. Hence if we want to
store only those entries below the skyline inside a vector rather than a matrix for maximum
storage eciency, then we shall dene a vector MAXA which provides the address of the diagonal
terms.
28
In the following global stiness matrix, the individual entries which must be stored in the
global stiness matrix are replaced by their address in the vector representation of this same
29
Victor Saouma
Draft
5.3 Skyline Storage of Global Stiness Matrix, MAXA Vector
521
Figure 5.8: Numbering Schemes for Simple Structure

matrix. Also shown is the corresponding MAXA vector.
1
1 x
2
3
4
5
6
7
8
9
10
11
12
K =
2
x
x
1 3
2 5
Victor Saouma
3
x
x
5
x
x
x
x x
x
11
10
9
6 8
7
17
16
15
14
13
12
6
x
x
x
x
x
x
24
23
22
21
20
19
18
7 8 9 10
x x x
x x x
x x x
x
x
x
x
x
x
x
x
x x
x
x
32
31
30
29
28
27
26
25
41
40
39
38
37
36
35
34
33
48
47
46
45
44
43
42
11
x
x
x
x
x
x
56
55
54
53
52
51
50
49
12
x
x
x
x
x
x
68
67
66
65
64
63
62
61
60
59
58
57
MAXA =
1
2
4
6
7
12
18
25
33
42
49
57
Draft
522
Thus, to locate an element within the stiness matrix, we use the following formula:
Kij = MAXA(j) + (j i)
(5.51)
if i j (since we are storing only the upper half).

30
Using this formula, we will have:

K58 = MAXA(8) + (8 5) = 18 + 3 = 21
(5.52)
K42 = MAXA(4) + (4 2) = 6 + 2 = 8
(5.53)
We should note that the total number of non-zero entries inside the global stiness matrix
is always the same, irrespective of our numbering scheme. However by properly numbering
the nodes, we can minimize the number of zero terms1 which would fall below the skyline and
which storage would be ineective.
31
5.4
32
Computer Program Organization
The main program should,

1. Read
(a) title card
(b) control card which should include:
i.
ii.
iii.
iv.
v.
Number of nodes
Number of elements
Type of structure: beam, grid, truss, or frame; (2D or 3D)
Number of dierent element properties
Number of load cases
2. Determine:
(a) Number of spatial coordinates for the structure
(b) Number of local and global degrees of freedom per node
3. For each node read:
(a) Node number
(b) Boundary conditions of each global degree of freedom [ID]
(c) Spatial coordinates
Note that all the above are usually written on the same data card
4. For each element, read:
1
As we shall see later, all the terms below the skyline (including the zeros) must be stored. Following matrix
decomposition, all zero terms outside the skyline terms remain zero, and all others are altered.
Victor Saouma
Draft
5.4 Computer Program Organization
523
(a) Element number

(b) First and second node
(c) Element Property number
5. For each element property group read the associated elastic and cross sectional characteristics. Note these variables will depend on the structure type.
6. Determine the vector u which stores the initial displacements.
7. Loop over all the elements and for each one:
(a) Retrieve its properties
(b) Determine the length
(c) Call the appropriate subroutines which will determine:
i. The stiness matrix in local coordinate systems [k(e) ].
ii. The angle and the transformation matrix [(e) ].
8. Assembly of the global stiness matrix
(a) Initialize the global stiness matrix to zero
(b) Loop through each element, e, and for each element:
i. Retrieve its stiness matrix (in local coordinates) [k(e) ] and transformation matrix [(e) ].
ii. Compute the element stiness matrix in global coordinates from [K(e) ] = [(e) ]T [k(e) ][(e) ].
iii. Dene the {LM} array of the element
iv. Loop through each row i and column j of the element stiness matrix, and for
those degree of freedom not equal to zero, add the contributions of the element
to the structures stiness matrix K S [LM (i), LM (j)] = K S [LM (i), LM (j)] +
K (e) [i, j]
9. Extract the structures stiness matrix [Ktt ] from the augmented stiness matrix.
10. Invert the structures stiness matrix (or decompose it).
11. For each load case:
(a) Determine the nodal equivalent loads (xed end actions), if any.
(b) Assemble the load vector
(c) Load assembly (once for each load cae) once the stiness matrix has been decomposed, than the main program should loop through each load case and,
i. Initialize the load vector (of length NEQ) to zero.
ii. Read number of loaded nodes. For each loaded node store the non-zero values
inside the load vector (using the [ID] matrix for determining storage location).
iii. Loop on all loaded elements:
A. Read element number, and load value
B. Compute the xed end actions and rotate them from local to global coordinates.
Victor Saouma
Draft
524

C. Using the LM vector, add the xed end actions to the nodal load vector
(unless the corresponding equation number is zero, ie. restrained degree of
freedom).
D. Store the xed end actions for future use.
(d) Apply Eq. 5.3 to determine the nodal displacements t = K1 (Pt Ktu u )
tt
(e) Apply Eq. 5.4 to determine the nodal reactions Rt = Kut t + Kuu u
(f) Determine the internal forces (axial, shear and moment)
i. For each element retrieve:
A. nodal coordinates
B. rotation matrix [(e) ].
C. element stiness matrix [k(e) ].
ii. Compute nodal displacements in local coordinate system from (e) = [(e) ] {}
iii. Compute element internal forces from {p} = [k(e) ] (e)
iv. If the element is loaded, add corresponding xed end actions
v. print the interior forces
5.5
Computer Program Flow Charts (FORTRAN)
Section must be edited

33
The main program should, Fig. 5.9:

1. Read
(a) TITLE CARD
(b) CONTROL CARD which should include:
i.
ii.
iii.
iv.
v.
Number of nodes
Number of elements
Type of structure: beam, grid, truss, or frame; (2D or 3D)
Number of dierent element properties
Number of load cases
2. Determine:
(a) Number of spatial coordinates for the structure
(b) Number of local and global degrees of freedom per node
3. Set up the pointers of the dynamic memory allocation (if using f77) for:
(a) Nodal coordinates
(b) Equation number matrix (ID)
(c) Element connectivity
(d) Element properties
(e) Element stiness matrices
Victor Saouma
Draft
5.5 Computer Program Flow Charts (FORTRAN)
525
(f) Element rotation matrices

4. Loop over all the elements and determine the element stiness matrices (in local coordinates), and rotation angles.
5. Determine the column heights, and initialize the global stiness vector to zero.
6. Loop through all the elements, and for each one
(a) Determine the element stiness matrices in global coordinates
(b) Determine the LM vector
(c) Assemble the structures global stiness matrix.
7. Decompose the global stiness matrix using a Choleskys decomposition).
8. For each load case:
(a) Determine the nodal equivalent loads (xed end actions), if any.
(b) Assemble the load vector
(c) Backsubstitute and obtain the nodal displacements
(d) Loop through each element and:
i. Determine the nodal displacements in local coordinates
ii. Determine the internal forces (include eects of xed end actions).
34
The tree structure of the program is illustrated in Fig. 5.10
5.5.1
35
Input
The input subroutine should:

1. For each node read:
(a) Node number
(b) Boundary conditions of each global degree of freedom [ID]
(c) Spatial coordinates
Note that all the above are usually written on the same data card
2. Determine equation numbers associated with each degree of freedom, and the total number
of equations (NEQ).
3. For each element, read:
(a) Element number
(b) First and second node
(c) Element Property number
4. For each element property group read the associated elastic and cross sectional characteristics. Note these variables will depend on the structure type.
Victor Saouma
Draft
526
Figure 5.9: Program Flowchart
Victor Saouma
Draft
527
Figure 5.10: Programs Tree Structure
Victor Saouma
Draft
528
5.5.2
Element Stiness Matrices
For each element:

1. Retrieve its properties
2. Determine the length
3. Call the appropriate subroutines which will determine:
(a) The stiness matrix in local coordinate systems [ke ].
(b) The direction cosines.
5.5.3
Assembly
Since a skyline solver will be used, we rst need to determine the appropriate pointers which
will enable us to eciently store the global stiness matrix ({MAXA}). This is accomplished
as follows, Fig. 5.11:
1. Determine the maximum height of the skyline for each column of the global stiness
matrix by rst assigning a very large number to each row of {MAXA}, and then looping
through each element, and for each one:
(a) Determine the lowest associated global degree of freedom number (from the {LM}
vectors)
(b) Compare this height with the one currently associated with those degree of freedom
stored in the element {LM}; if lower overwrite
2. Determine the total height of each skyline (i.e. each column) by determining the dierence between MAXA (IEQ) (Skyline elevation), and IEQ (BottomLine). Overwrite
MAXA with this height.
3. Determine the total length of the vector storing the compacted structure global stiness
matrix by summing up the height of each skyline
4. Assign to MAXA(NEQ+1) this total length +1.
5. Loop backward from the last column to the rst, and for each one determine the address
of the diagonal term from MAXA(IEQ) = MAXA(IEQ + 1) MAXA(IEQ)
36 Once the MAXA vector has been determine, then term K(i, j) in the square matrix, would
be stored in KK(MAXA(j)+j-i) (assuming j > i) in the compacted form of {K}.
37
The assembly of the global stiness matrix is next described, Fig. 5.12:
1. Initialize the vector storing the compacted stiness matrix to zero.
2. Loop through each element, e, and for each element:
(a) Retrieve its stiness matrix (in local coordinates) [ke ], and direction cosines.
(b) Determine the rotation matrix [] of the element.
(c) Compute the element stiness matrix in global coordinates from [bK e ] = []T [ke ][].
Victor Saouma
Draft
529
Figure 5.11: Flowchart for the Skyline Height Determination
Victor Saouma
Draft
530
Figure 5.12: Flowchart for the Global Stiness Matrix Assembly
Victor Saouma
Draft
531
(d) Dene the {LM} array of the element

(e) Loop through each row and column of the element stiness matrix, and for those
degree of freedom not equal to zero, add the contributions of the element to the
structures stiness matrix (note that we assemble only the upper half).
K S [LM (i), LM (j)] = K S [LM (i), LM (j)] + K e [i, j]
5.5.4
(5.54)
Decomposition
38 Decompose the global stiness matrix. Since the matrix is both symmetric and positive
denite, the matrix can be decomposed using Choleskys method into: [K] = [L][L]T . Should a
division by zero occur, or an attempt to extract the square root of a negative number happen,
then this would be an indication that either the global stiness matrix is not properly assembled,
or that there are not enough restraint to prevent rigid body translation or rotation of the
structure.
5.5.5
Load
Once the stiness matrix has been decomposed, than the main program should loop through
each load case and, Fig. 5.13
39
1. Initialize the load vector (of length NEQ) to zero.

2. Read number of loaded nodes. For each loaded node store the non-zero values inside the
load vector (using the [ID] matrix for determining storage location).
3. Loop on all loaded elements:
(a) Read element number, and load value
(b) Compute the xed end actions and rotate them from local to global coordinates.
(c) Using the LM vector, add the xed end actions to the nodal load vector (unless the
corresponding equation number is zero, ie. restrained degree of freedom).
(d) Store the xed end actions for future use.
5.5.6
Backsubstitution
Backsubstitution is achieved by multiplying the decomposed stiness matrix with the load
vector. The resulting vector stores the nodal displacements, in global coordinate system, corresponding to the unrestrained degree of freedom.
40
5.5.7
Internal Forces and Reactions
The internal forces for each element, and reactions at each restrained degree of freedom, are
determined by, Fig. 5.15
41
1. Initialize reactions to zero

Victor Saouma
Draft
532
Figure 5.13: Flowchart for the Load Vector Assembly
Victor Saouma
Draft
533
Figure 5.14: Flowchart for the Internal Forces
Victor Saouma
Draft
534
Figure 5.15: Flowchart for the Reactions
Victor Saouma
Draft
5.6 Computer Implementation with MATLAB
535
2. For each element retrieve:

(a) nodal coordinates
(b) rotation matrix
(c) element stiness matrix
3. Compute nodal displacements in local coordinate system from e = [](LM)
4. Compute element internal forces from p = [ke ] e
5. If the element is loaded, add corresponding xed end actions
6. print the interior forces
7. check if any of its degree of freedom is restrained, if so:
(a) rotate element forces to global coordinates
(b) update appropriate reaction
5.6
Computer Implementation with MATLAB
42 You will be required, as part of your term project, to write a simple MATLAB (or whatever
other language you choose) program for the analysis of two dimensional frames with nodal load
and initial displacement, as well as element load.
5.6.1
Program Input
43 It is essential that the structure be idealized such that it can be discretized. This discretization should dene each node and element uniquely. In order to decrease the required amount
of computer storage and computation it is best to number the nodes in a manner that minimizes the numerical separation of the node numbers on each element. For instance, an element
connecting nodes 1 and 4, could be better dened by nodes 1 and 2, and so on. As it was
noted previously, the user is required to have a decent understanding of structural analysis and
structural mechanics. As such, it will be necessary for the user to generate or modify an input
le input.m using the following directions. Open the le called input.m and set the existing
variables in the le to the appropriate values. The input le has additional helpful directions
given as comments for each variable. After setting the variables to the correct values, be sure
to save the le. Please note that the program is case-sensitive.
In order for the program to be run, the user must supply the required data by setting certain
variables in the le called indat.m equal to the appropriate values. All the user has to do
is open the text le called indat.txt, ll in the required values and save the le as indat.m
in a directory within MATABs path. There are helpful hints within this le. It is especially
important that the user keep track of units for all of the variables in the input data le. All
of the units MUST be consistent. It is suggested that one always use the same units for all
problems. For example, always use kips and inches, or kilo- newtons and millimeters.
44
Victor Saouma
Draft
536
5.6.1.1
Input Variable Descriptions
A brief description of each of the variables to be used in the input le is given below:
npoin This variable should be set equal to the number of nodes that comprise the structure.
A node is dened as any point where two or more elements are joined.
nelem This variable should be set equal to the number of elements in the structure. Elements
are the members which span between nodes.
istrtp This variable should be set equal to the type of structure. There are six types of
structures which this program will analyze: beams, 2-D trusses, 2-D frames, grids, 3-D trusses,
and 3-D frames. Set this to 1 for beams, 2 for 2D-trusses, 3 for 2D- frames, 4 for grids, 5 for
3D-trusses, and 6 for 3D-frames. An error will occur if it is not set to a number between 1 and
6. Note only istrp=3 was kept.
nload This variable should be set equal to the number of dierent load cases to be analyzed.
A load case is a specic manner in which the structure is loaded.
ID (matrix) The ID matrix contains information concerning the boundary conditions for
each node. The number of rows in the matrix correspond with the number of nodes in the
structure and the number of columns corresponds with the number of degrees of freedom for
each node for that type of structure type. The matrix is composed of ones and zeros. A one
indicates that the degree of freedom is restrained and a zero means it is unrestrained.
nodecoor (matrix) This matrix contains the coordinates (in the global coordinate system)
of the nodes in the structure. The rows correspond with the node number and the columns
correspond with the global coordinates x, y, and z, respectively. It is important to always
include all three coordinates for each node even if the structure is only two- dimensional. In
the case of a two-dimensional structure, the z-coordinate would be equal to zero.
lnods (matrix) This matrix contains the nodal connectivity information. The rows correspond with the element number and the columns correspond with the node numbers which the
element is connected from and to, respectively.
E,A,Iy (arrays) These are the material and cross-sectional properties for the elements. They
are arrays with the number of terms equal to the number of elements in the structure. The
index number of each term corresponds with the element number. For example, the value of
A(3) is the area of element 3, and so on. E is the modulus of elasticity, A is the cross-sectional
area, Iy is the moment of inertia about the y axes
Pnods This is an array of nodal loads in global degrees of freedom. Only put in the loads
in the global degrees of freedom and if there is no load in a particular degree of freedom, then
put a zero in its place. The index number corresponds with the global degree of freedom.
Pelem This an array of element loads, or loads which are applied between nodes. Only
one load between elements can be analyzed. If there are more than one element loads on
the structure, the equivalent nodal load can be added to the nodal loads. The index number
corresponds with the element number. If there is not a load on a particular member, put a zero
in its place. These should be in local coordinates.
a This is an array of distances from the left end of an element to the element load. The index
number corresponds to the element number. If there is not a load on a particular member, put
a zero in its place. This should be in local coordinates.
w This is an array of distributed loads on the structure. The index number corresponds
with the element number. If there is not a load on a particular member, put a zero in its place.
This should be in local coordinates
dispflag Set this variable to 1 if there are initial displacements and 0 if there are none.
45
Victor Saouma
Draft
537
initial displ This is an array of initial displacements in all structural degrees of freedom.
This means that you must enter in values for all structure degrees of freedom, not just those
restrained. For example, if the structure is a 2D truss with 3 members and 3 node, there would
be 6 structural degrees of freedom, etc. If there are no initial displacements, then set the values
equal to zero.
angle This is an array of angles which the x-axis has possibly been rotated. This angle
is taken as positive if the element has been rotated towards the z-axis. The index number
corresponds to the element number.
drawflag Set this variable equal to 1 if you want the program to draw the structure and 0
if you do not.
5.6.1.2
Sample Input Data File
The contents of the input.m le which the user is to ll out is given below:
%**********************************************************************************************
%
Scriptfile name:
indat.m
(EXAMPLE 2D-FRAME INPUT DATA)
%
%
Main Program:
casap.m
%
%
This is the main data input file for the computer aided
%
structural analysis program CASAP. The user must supply
%
the required numeric values for the variables found in
%
this file (see users manual for instructions).
%
%**********************************************************************************************
%
HELPFUL INSTRUCTION COMMENTS IN ALL CAPITALS
SET NPOIN EQUAL TO THE NUMBER OF NODES IN THE STRUCTURE
npoin=3;
%
SET NELEM EQUAL TO THE NUMBER OF ELEMENTS IN THE STRUCTURE
nelem=2;
%
SET NLOAD EQUAL TO THE NUMBER OF LOAD CASES
nload=1;
%
INPUT THE ID MATRIX CONTAINING THE NODAL BOUNDARY CONDITIONS (ROW # = NODE #)
ID=[1 1 1;
0 0 0;
1 1 1];
%
INPUT THE NODE COORDINATE (X,Y) MATRIX, NODECOOR (ROW # = NODE #)
nodecoor=[
0 0;
7416 3000;
15416 3000
];
%
INPUT THE ELEMENT CONNECTIVITY MATRIX, LNODS (ROW # = ELEMENT #)
Victor Saouma
Draft
538
lnods=[
1 2;
2 3
];
%
%
%
INPUT THE MATERIAL PROPERTIES ASSOCIATED WITH THIS TYPE OF STRUCTURE

PUT INTO ARRAYS WHERE THE INDEX NUMBER IS EQUAL TO THE CORRESPONDING ELEMENT NUMBER.
COMMENT OUT VARIABLES THAT WILL NOT BE USED
E=[200 200];
A=[6000 6000];
Iz=[200000000 200000000];
%
%
%
%
%
%
%
%
%
INPUT THE LOAD DATA. NODAL LOADS, PNODS SHOULD BE IN MATRIX FORM. THE COLUMNS CORRESPOND
TO THE GLOBAL DEGREE OF FREEDOM IN WHICH THE LOAD IS ACTING AND THE THE ROW NUMBER CORRESPONDS
WITH THE LOAD CASE NUMBER. PELEM IS THE ELEMENT LOAD, GIVEN IN A MATRIX, WITH COLUMNS
CORRESPONDING TO THE ELEMENT NUMBER AND ROW THE LOAD CASE. ARRAY "A" IS THE DISTANCE FROM
THE LEFT END OF THE ELEMENT TO THE LOAD, IN ARRAY FORM. THE DISTRIBUTED LOAD, W SHOULD BE
IN MATRIX FORM ALSO WITH COLUMNS = ELEMENT NUMBER UPON WHICH W IS ACTING AND ROWS = LOAD CASE.
ZEROS SHOULD BE USED IN THE MATRICES WHEN THERE IS NO LOAD PRESENT. NODAL LOADS SHOULD
BE GIVEN IN GLOBAL COORDINATES, WHEREAS THE ELEMENT LOADS AND DISTRIBUTED LOADS SHOULD BE
GIVEN IN LOCAL COORDINATES.
Pnods=[18.75 -46.35 0];

Pelem=[0 0];
a=[0 0];
w=[0 4/1000];
%
%
IF YOU WANT THE PROGRAM TO DRAW THE STUCTURE SET DRAWFLAG=1, IF NOT SET IT EQUAL TO 0.
THIS IS USEFUL FOR CHECKING THE INPUT DATA.
drawflag=1;
%
END OF INPUT DATA FILE
5.6.1.3
Program Implementation
In order to run the program, open a new MATLAB Notebook. On the rst line, type the
name of the main program CASAP and evaluate that line by typing ctrl-enter. At this point, the
main program reads the input le you have just created and calls the appropriate subroutines
to analyze your structure. In doing so, your input data is echoed into your MATLAB notebook
and the program results are also displayed. As a note, the program can also be executed directly
from the MATAB workspace window, without Microsoft Word.
5.6.2
5.6.2.1
Program Listing
Main Program
%**********************************************************************************************
%Main Program: casap.m
%
%
This is the main program, Computer Aided Structural Analysis Program
%
CASAP. This program primarily contains logic for calling scriptfiles and does not
%
perform calculations.
%
%
All variables are global, but are defined in the scriptfiles in which they are used.
%
Victor Saouma
Draft

%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
539
Associated scriptfiles:
(for all stuctures)
indat.m (input data file)
idrasmbl.m
elmcoord.m
draw.m
(3 - for 2D-frames)
length3.m
stiffl3.m
trans3.m
assembl3.m
loads3.m
disp3.m
react3.m
%**********************************************************************************************
%
COMMENT CARDS ARE IN ALL CAPITALS
SET NUMERIC FORMAT
format short e
%
CLEAR MEMORY OF ALL VARIABLES
clear
% INITIALIZE OUTPUT FILE
fid = fopen(casap.out, wt);
%
%
SET ISTRTP EQUAL TO THE NUMBER CORRESPONDING TO THE TYPE OF STRUCTURE:

3
=
2DFRAME
istrtp=3;
%
READ INPUT DATA SUPPLIED BY THE USER
indat
%
%
REASSAMBLE THE ID MATRIX AND CALCULATE THE LM VECTORS

CALL SCRIPTFILE IDRASMBL
idrasmbl
%
ASSEMBLE THE ELEMENT COORDINATE MATRIX
elmcoord
%
2DFRAME CALCULATIONS
%
%
CALCULATE THE LENGTH AND ORIENTATION ANGLE, ALPHA FOR EACH ELEMENT
CALL SCRIPTFILE LENGTH3.M
length3
%
CALCULATE THE 2DFRAME ELEMENT STIFFNESS MATRIX IN LOCAL COORDINATES
Victor Saouma
Draft
540
%
CALL SCRIPTFILE STIFFL3.M
stiffl3
%
%
CALCULATE THE 2DFRAME ELEMENT STIFFNESS MATRIX IN GLOBAL COORDINATES

CALL SCRIPTFILE TRANS3.M
trans3
%
%
ASSEMBLE THE GLOBAL STRUCTURAL STIFFNESS MATRIX

CALL SCRIPTFILE ASSEMBL3.M
assembl3
%
PRINT STRUCTURAL INFO
print_general_info
%
LOOP TO PERFORM ANALYSIS FOR EACH LOAD CASE
for iload=1:nload
print_loads
%
%
DETERMINE THE LOAD VECTOR IN GLOBAL COORDINATES

CALL SCRIPTFILE LOADS3.M
loads3
%
%
CALCULATE THE DISPLACEMENTS

CALL SCRIPTFILE DISP3.M
disp3
%
%
CALCULATE THE REACTIONS AT THE RESTRAINED DEGREES OF FREEDOM

CALL SCRIPTFILE REACT3.M
react3
% CALCULATE THE INTERNAL FORCES FOR EACH ELEMENT
intern3
%
END LOOP FOR EACH LOAD CASE
end
%
%
DRAW THE STRUCTURE, IF USER HAS REQUESTED (DRAWFLAG=1)

CALL SCRIPTFILE DRAW.M
draw
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
%************************************************************************************************
Victor Saouma
Draft

%SCRIPTFILE NAME:
%
%MAIN FILE
:
%
%Description
:
%
%
%
%
541
IDRASMBL.M
CASAP
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
%**********************************************************************************************
%SCRIPTFILE NAME:
ELEMCOORD.M
%
%MAIN FILE
:
CASAP
%
%Description
:
This file assembles a matrix, elemcoor which contains the coordinates
%
of the first and second nodes on each element, respectively.
%
%
%**********************************************************************************************
%
ASSEMBLE THE ELEMENT COORDINATE MATRIX, ELEMCOOR FROM NODECOOR AND LNODS
for ielem=1:nelem
Victor Saouma
Draft
542

end
%
END OF ELMCOORD.M SCRIPTFILE
5.6.2.4
Element Lengths
%**********************************************************************************************
%
Scriptfile name :
length3.m
(for 2d-frame structures)
%
%
Main program
:
casap.m
%
%
When this file is called, it computes the length of each element and the
%
angle alpha between the local and global x-axes. This file can be used
%
for 2-dimensional elements such as 2-D truss, 2-D frame, and grid elements.
%
This information will be useful for transformation between local and global
%
variables.
%
%
Variable descriptions: (in the order in which they appear)
%
%
nelem
=
number of elements in the structure
%
ielem
=
counter for loop
%
L(ielem)
=
length of element ielem
%
elemcoor(ielem,4)
=
xj-coordinate of element ielem
%
elemcoor(ielem,1)
=
xi-coordinate of element ielem
%
elemcoor(ielem,5)
=
yj-coordinate of element ielem
%
elemcoor(ielem,2)
=
yi-coordinate of element ielem
%
alpha(ielem)
=
angle between local and global x-axes
%
%
%**********************************************************************************************
%
COMPUTE THE LENGTH AND ANGLE BETWEEN LOCAL AND GLOBAL X-AXES FOR EACH ELEMENT
for ielem=1:nelem
L(ielem)=
alpha(ielem)=
%
END OF LENGTH3.M SCRIPTFILE
5.6.2.5
Element Stiness Matrices
%**********************************************************************************************
%
Scriptfile name:
stiffl3.m
%
%
Main program:
casap.m
%
%
When this file is called, it computes the element stiffenss matrix
%
of a 2-D frame element in local coordinates. The element stiffness
%
matrix is calculated for each element in the structure.
%
%
The matrices are stored in a single matrix of dimensions 6x6*i and
%
can be recalled individually later in the program.
%
%
Variable descriptions: (in the order in which the appear)
%
Victor Saouma
Draft

%
%
%
%
%
%
%
ielem
=
nelem
=
k(ielem,6,6)=
E(ielem)
=
A(ielem)
=
L(ielem)
=
Iz(ielem)
=
543
counter for loop

number of element in the structure
element stiffness matrix in local coordinates
modulus of elasticity of element ielem
cross-sectional area of element ielem
lenght of element ielem
moment of inertia with respect to the local z-axis of element ielem
%
%**********************************************************************************************
for ielem=1:nelem
k(1:6,1:6,ielem)=...
end
%
END OF STIFFL3.M SCRIPTFILE
5.6.2.6
Transformation Matrices
%**********************************************************************************************
%
Scriptfile name :
trans3.m
%
%
Main program
:
casap.m
%
%
This file calculates the rotation matrix and the element stiffness
%
matrices for each element in a 2D frame.
%
%
%
%
ielem
=
counter for the loop
%
nelem
=
%
rotation
=
rotation matrix containing all elements info
%
Rot
=
rotational matrix for 2d-frame element
%
alpha(ielem)
=
angle between local and global x-axes
%
K
=
element stiffness matrix in global coordinates
%
k
=
element stiffness matrix in local coordinates
%
%**********************************************************************************************
%
%
CALCULATE THE ELEMENT STIFFNESS MATRIX IN GLOBAL COORDINATES

FOR EACH ELEMENT IN THE STRUCTURE
for ielem=1:nelem
%
SET UP THE ROTATION MATRIX, ROTATAION
rotation(1:6,1:6,ielem)=...
ktemp=k(1:6,1:6,ielem);
%
CALCULATE THE ELEMENT STIFFNESS MATRIX IN GLOBAL COORDINATES
K(1:6,1:6,ielem)=
end
%
END OF TRANS3.M SCRIPTFILE
Victor Saouma
Draft
544
5.6.2.7
Assembly of the Augmented Stiness Matrix
%**********************************************************************************************
%
Scriptfile name :
assembl3.m (for 2d-frame structures)
%
%
Main program
:
casap.m
%
%
This file assembles the global structural stiffness matrix from the
%
element stiffness matrices in global coordinates using the LM vectors.
%
In addition, this file assembles the augmented stiffness matrix.
%
%
Variable Descritpions (in order of appearance):
%
%
ielem
=
Row counter for element number
%
nelem
=
Number of elements in the structure
%
iterm
=
Counter for term number in LM matrix
%
LM(a,b)
=
LM matrix
%
jterm
=
Column counter for element number
%
temp1
=
Temporary variable
%
temp2
=
Temporary variable
%
temp3
=
Temporary variable
%
temp4
=
Temporary variable
%
number_gdofs
=
Number of global dofs
%
new_LM
=
LM matrix used in assembling the augmented stiffness matrix
%
aug_total_dofs =
Total number of structure dofs
%
K_aug
=
Augmented structural stiffness matrix
%
Ktt
=
Structural Stiffness Matrix (Upper left part of Augmented structural stiffness matrix)
%
Ktu
=
Upper right part of Augmented structural stiffness matrix
%
Kut
=
Lower left part of Augmented structural stiffness matrix
%
Kuu
=
Lower rigth part of Augmented structural stiffness matrix
%
%
%**********************************************************************************************
% RENUMBER DOF INCLUDE ALL DOF, FREE DOF FIRST, RESTRAINED NEXT
new_LM=LM;
number_gdofs=max(LM(:));
new_LM(find(LM<0))=number_gdofs-LM(find(LM<0));
aug_total_dofs=max(new_LM(:));
%
ASSEMBLE THE AUGMENTED STRUCTURAL STIFFNESS MATRIX
K_aug=zeros(aug_total_dofs);
for ielem=1:nelem
Tough one!
end
%
SET UP SUBMATRICES FROM THE AUGMENTED STIFFNESS MATRIX
Ktt=
XXXXXXXXXXXXXXXXXXXXXX
Ktu=
Kut=
Kuu=
Victor Saouma
COMPLETE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
Draft

%
545
END OF ASSEMBL3.M SCRIPTFILE
5.6.2.8
Print General Information
%**********************************************************************************************
%
Scriptfile name :
print_general_info.m
%
%
Main program
:
casap.m
%
%
Prints the general structure info to the output file
%**********************************************************************************************
XXXXXXXXXXXXXXXXXXXXXX COMPLETE and EDIT XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
fprintf(fid,\n\nNumber of Nodes: %d\n,npoin);
fprintf(fid,Number of Elements: %d\n,nelem);
fprintf(fid,Number of Load Cases: %d\n,nload);
fprintf(fid,Number of Restrained dofs: %d\n,abs(min(LM(:))));
fprintf(fid,Number of Free dofs: %d\n,max(LM(:)));
fprintf(fid,\nNode Info:\n);
for inode=1:npoin
fprintf(fid,
Node %d (%d,%d)\n,inode,nodecoor(inode,1),nodecoor(inode,2));
freedof= ;
if(ID(inode,1))>0
freedof=strcat(freedof, X );
end
if(ID(inode,2))>0
freedof=strcat(freedof, Y );
end
if(ID(inode,3))>0
freedof=strcat(freedof, Rot);
end
if freedof==
freedof= none; node is fixed;
end
fprintf(fid,
Free dofs:%s\n,freedof);
end
fprintf(fid,\nElement Info:\n);
for ielem=1:nelem
fprintf(fid,
Element %d (%d->%d),ielem,lnods(ielem,1),lnods(ielem,2));
fprintf(fid,
E=%d A=%d Iz=%d \n,E(ielem),A(ielem),Iz(ielem));
end
5.6.2.9
Print Load
%**********************************************************************************************
%
Scriptfile name :
print_loads.m
%
%
Main program
:
casap.m
%
%
Prints the current load case data to the output file
%**********************************************************************************************
Load_case=iload
if iload==1
fprintf(fid,\n_________________________________________________________________________\n\n);
Victor Saouma
Draft
546
end
fprintf(fid,Load Case: %d\n\n,iload);
fprintf(fid,
Nodal Loads:\n);
for k=1:max(LM(:));
%WORK BACKWARDS WITH LM MATRIX TO FIND NODE# AND DOF
LM_spot=find(LM==k);
elem=fix(LM_spot(1)/(nterm+1))+1;
dof=mod(LM_spot(1)-1,nterm)+1;
node=lnods(elem,fix(dof/4)+1);
switch(dof)
case {1,4}, dof=Fx;
case {2,5}, dof=Fy;
otherwise, dof= M;
end
%PRINT THE DISPLACEMENTS
if Pnods(k)~=0
fprintf(fid,
Node: %2d %s = %14d\n,node, dof, Pnods(k));
end
end
fprintf(fid,\n
Elemental Loads:\n);
for k=1:nelem
fprintf(fid,
Element: %d
Point load = %d at %d from left\n,k,Pelem(k),a(k));
fprintf(fid,
Distributed load = %d\n,w(k));
end
fprintf(fid,\n);
5.6.2.10
Load Vector
%**********************************************************************************************
%
Scriptfile name:
loads3.m
%
%
Main program:
casap.m
%
%
When this file is called, it computes the fixed end actions for elements which
%
carry distributed loads for a 2-D frame.
%
%
%
%
ielem
=
counter for loop
%
nelem
=
%
b(ielem)
=
distance from the right end of the element to the point load
%
L(ielem)
=
length of the element
%
a(ielem)
=
distance from the left end of the element to the point load
%
Ffl
=
fixed end force (reaction) at the left end due to the point load
%
w(ielem)
=
distributed load on element ielem
%
L(ielem)
=
length of element ielem
%
Pelem(ielem)
=
element point load on element ielem
%
Mfl
=
fixed end moment (reaction) at the left end due to the point load
%
Ffr
=
fixed end force (reaction) at the right end due to the point load
%
Mfr
=
fixed end moment (reaction) at the right end due to the point load
%
feamatrix_local
=
matrix containing resulting fixed end actions in local coordinates
%
feamatrix_global
=
matrix containing resulting fixed end actions in global coordinates
%
fea_vector
=
vector of feas in global dofs, used to calc displacements
%
fea_vector_abs
=
vector of feas in every structure dof
%
dispflag
=
flag indicating initial displacements
Victor Saouma
Draft
547
%
Ffld
=
fea (vert force) on left end of element due to initial disp
%
Mfld
=
fea (moment) on left end of element due to initial disp
%
Ffrd
=
fea (vert force) on right end of element due to initial disp
%
Mfrd
=
fea (moment) on right end of element due to initial disp
%
fea_vector_disp
=
vector of feas due to initial disp, used to calc displacements
%
fea_vector_react
=
vector of feas due to initial disp, used to calc reactions
%
%**********************************************************************************************
%
%
%
CALCULATE THE FIXED END ACTIONS AND INSERT INTO A MATRIX IN WHICH THE ROWS CORRESPOND
WITH THE ELEMENT NUMBER AND THE COLUMNS CORRESPOND WITH THE ELEMENT LOCAL DEGREES
OF FREEDOM
for ielem=1:nelem
feamatrix_local(ielem,1:6)=[0 Ffl Mfl 0 Ffr Mfr];

%
ROTATE THE LOCAL FEA MATRIX TO GLOBAL
feamatrix_global=...
end
%
CREATE A LOAD VECTOR USING THE LM MATRIX
INITIALIZE FEA VECTOR TO ALL ZEROS
for idofpn=1:ndofpn
fea_vector(idofpn,1)=0;
end
for ielem=1:nelem
end
END OF LOADS3.M SCRIPTFILE
5.6.2.11
Nodal Displacements
%**********************************************************************************************
%
Scriptfile name :
disp3.m (for 2d-frame structures)
%
%
Main program
:
casap.m
%
%
When this file is called, it computes the displacements in the global
%
degrees of freedom.
%
%
%
%
Ksinv
=
inverse of the structural stiffness matrix
%
Ktt
=
structural stiffness matrix
%
Delta
=
vector of displacements for the global degrees of freedom
%
Pnods
=
vector of nodal loads in the global degrees of freedom
%
fea_vector =
vector of fixed end actions in the global degrees of freedom
Victor Saouma
Draft
548
%
%
Simplified for 2D Frame Case only
%
%**********************************************************************************************
%
CREATE A TEMPORARY VARIABLE EQUAL TO THE INVERSE OF THE STRUCTURAL STIFFNESS MATRIX
Ksinv=inv(Ktt);
%
CALCULATE THE DISPLACEMENTS IN GLOBAL COORDINATES
Delta=
%
PRINT DISPLACEMENTS WITH NODE INFO
fprintf(fid,
Displacements:\n);
for k=1:size(Delta,1)
end
%PRINT THE DISPLACEMENTS
fprintf(fid,
(Node: %2d %s) %14d\n,node, dof, Delta(k));
end
fprintf(fid,\n);
%
END OF DISP3.M SCRIPTFILE
5.6.2.12
Reactions
%**********************************************************************************************
%
Scriptfile name :
react3.m
%
%
Main program
:
casap.m
%
%
When this file is called, it calculates the reactions at the restrained degrees of
%
freedom.
%
%
Variable Descriptions:
%
%
Reactions
=
Reactions at restrained degrees of freedom
%
Kut
=
Upper left part of aug stiffness matrix, normal structure stiff matrix
%
Delta
=
vector of displacements
%
fea_vector_react
=
vector of feas in restrained dofs
%
%
%
%**********************************************************************************************
%
CALCULATE THE REACTIONS FROM THE AUGMENTED STIFFNESS MATRIX
Reactions=
fprintf(fid,
Reactions:\n);
for k=1:size(Reactions,1)
%WORK BACKWARDS WITH LM MATRIX TO FIND NODE# AND DOF
Victor Saouma
Draft
549
LM_spot=find(LM==-k);
elem=fix(LM_spot(1)/(nterm+1))+1;
dof=mod(LM_spot(1)-1,nterm)+1;
node=lnods(elem,fix(dof/4)+1);
switch(dof)
case {1,4}, dof=Fx;
case {2,5}, dof=Fy;
otherwise, dof=M ;
end
%PRINT THE REACTIONS
fprintf(fid,
(Node: %2d %s) %14d\n,node, dof, Reactions(k));
end
fprintf(fid,\n);
%
END OF REACT3.M SCRIPTFILE
5.6.2.13
Internal Forces
%**********************************************************************************************
%
Scriptfile name :
intern3.m
%
%
Main program
:
casap.m
%
%
When this file is called, it calculates the internal forces in all elements
%
freedom.
%
%
%**********************************************************************************************
Pglobe=zeros(6,nelem);
Plocal=Pglobe;
fprintf(fid,
Internal Forces:);
%LOOP FOR EACH ELEMENT
for ielem=1:nelem
%FIND ALL 6 LOCAL DISPLACEMENTS
elem_delta=zeros(6,1);
for idof=1:6
gdof=LM(ielem,idof);
if gdof<0
elem_delta(idof)=
else
elem_delta(idof)=
end
end
%SOLVE FOR ELEMENT FORCES (GLOBAL)
Pglobe(:,ielem)=K(:,:,ielem)*elem_delta+feamatrix_global(:,ielem);
%ROTATE FORCES FROM GLOBAL TO LOCAL COORDINATES
%ROTATE FORCES TO LOCAL COORDINATES
Plocal(:,ielem)=
%PRINT RESULTS
fprintf(fid,\n
Victor Saouma
Element: %2d\n,ielem);
Draft
550
for idof=1:6
if idof==1
fprintf(fid,
At Node: %d\n,lnods(ielem,1));
end
if idof==4
fprintf(fid,
At Node: %d\n,lnods(ielem,2));
end
switch(idof)
case {1,4}, dof=Fx;
case {2,5}, dof=Fy;
otherwise, dof=M ;
end
fprintf(fid,
(Global : %s ) %14d,dof, Pglobe(idof,ielem));
fprintf(fid,
(Local : %s ) %14d\n,dof, Plocal(idof,ielem));
end
end
fprintf(fid,\n_________________________________________________________________________\n\n);
5.6.2.14
Sample Output File
CASAP will display gure 5.16.

8000
6000
4000
2
2000
1
0
2000
4000
6000
2000
4000
6000
8000
10000
12000
14000
16000
Figure 5.16: Structure Plotted with CASAP

Number
Number
Number
Number
Number
of
of
of
of
of
Nodes: 3
Elements: 2
Load Cases: 1
Restrained dofs: 6
Free dofs: 3
Node Info:
Node 1 (0,0)
Free dofs: none; node is fixed
Node 2 (7416,3000)
Free dofs: X Y Rot
Node 3 (15416,3000)
Free dofs: none; node is fixed
Victor Saouma
Draft
Element Info:
Element 1 (1->2)
Element 2 (2->3)
551
E=200 A=6000 Iz=200000000

E=200 A=6000 Iz=200000000
_________________________________________________________________________
Load Case: 1
Nodal Loads:
Node: 2 Fx = 1.875000e+001
Node: 2 Fy = -4.635000e+001
Elemental Loads:
Element: 1
Point load = 0 at 0 from left
Distributed load = 0
Element: 2
Point load = 0 at 0 from left
Distributed load = 4.000000e-003
Displacements:
(Node: 2 delta X) 9.949820e-001
(Node: 2 delta Y) -4.981310e+000
(Node: 2 rotate ) -5.342485e-004
Reactions:
(Node: 1
(Node: 1
(Node: 1
(Node: 3
(Node: 3
(Node: 3
Fx) 1.304973e+002
Fy) 5.567659e+001
M ) 1.337416e+004
Fx) -1.492473e+002
Fy) 2.267341e+001
M ) -4.535573e+004
Internal Forces:
Element: 1
At Node: 1
(Global :
(Global :
(Global :
At Node: 2
(Global :
(Global :
(Global :
Element: 2
At Node: 2
(Global
(Global
(Global
At Node: 3
(Global
(Global
(Global
Fx )
Fy )
M )
1.304973e+002
5.567659e+001
1.337416e+004
(Local : Fx )
(Local : Fy )
(Local : M )
1.418530e+002
2.675775e+000
1.337416e+004
Fx ) -1.304973e+002
Fy ) -5.567659e+001
M ) 8.031549e+003
(Local : Fx ) -1.418530e+002
(Local : Fy ) -2.675775e+000
(Local : M ) 8.031549e+003
: Fx ) 1.492473e+002
: Fy ) 9.326590e+000
: M ) -8.031549e+003
(Local : Fx ) 1.492473e+002
(Local : Fy ) 9.326590e+000
(Local : M ) -8.031549e+003
: Fx ) -1.492473e+002
: Fy ) 2.267341e+001
: M ) -4.535573e+004
(Local : Fx ) -1.492473e+002
(Local : Fy ) 2.267341e+001
(Local : M ) -4.535573e+004
_________________________________________________________________________
Victor Saouma
Draft
552
5.7
Homework
1. For the following structure

(a) Determine the structure ID matrix, the LM [k(e) ] and [(e) ], and [K(e) ] for each
element.
(b) Assemble the augmented stiness matrix.
(c) Determine the vector of nodal forces.
(d) Solve for the nodal displacements and reactions.
(e) Determine the internal forces in each element.
Assume
A mm2
4 103
6 103
1
2
Iz mm4
50 106
200 106
J mm4
100 103
300 103
E=200,000 MPa, =0.3

Members are orthogonal.
1
0
1
0
5 kN
1
0
1
0
m 2
3 kN/m
1
5m
You are strongly encouraged to write a special code in MATLAB to solve this problem.
2. For the truss shown below
(a) Determine the Structure ID matrix.
(b) For each element, determine the LM vector, transformation matrix, and element
stiness matrix in global coordinates.
(c) Assemble the global stiness matrix, decompose it, and determine nodal displacements.
(d) Determine the axial forces in each element.
(e) Determine the reactions.
Victor Saouma
Draft
5.7 Homework
553
(f) Determine the structure MAXA vector, and the number of terms which would have
to be stored in a vector should a variable skyline solver be used.
(g) Can you suggest an alternative numbering order for the node? What would be the
number of terms to be stored in the global stiness matrix.
Assume A = 10in2 , E = 30, 000 ksi for all members.
Note that the transformed element stiness matrix in global coordinates is given by:
[Ke ] =
EA
cos2
sin cos
cos2
sin cos
2
sin cos
sin
sin cos
sin2
sin cos
cos2
sin cos
cos2
2
sin cos
sin
sin cos
sin2
(5.55)
3. For the non-prismatic element shown below,

(a) Write a computer program to derive its stiness matrix, using E = 4, 000 ksi, t = 10
in. Neglect axial deformation.
5
1
50k
6
8
12
100k
16
Victor Saouma
16
Draft
554
12"
18"
Victor Saouma
12
Draft
Chapter 6
EQUATIONS OF STATICS and

KINEMATICS
Note: This section is largely based on chapter 6 of Mc-Guire and Gallagher, Matrix Structural
Analysis, John Wiley
1 Having developed the stiness method in great details, and prior to the introduction of energy
based methods (which will culminate with the nite element formulation), we ought to revisit
the exibility method. This will be done by rst introducing some basic statics and kinematics
relationship.
2 Those relations will eventually enable us not only to formulate the exibility/stiness relation,
but also other by-products such as algorithms for: 1) the extraction of a statically determinate
structure from a statically indeterminate one; 2) checking prior to analysis whether a structure
is kinematically unstable; 3) providing an alternative method of assembling the global stiness
matrix.
6.1
6.1.1
Statics Matrix [B]

Statically Determinate
The statics matrix [B] relates the vector of all the structures {P} nodal forces in global
coordinates to all the unknown forces (element internal forces in their local coordinate system
and structures external reactions) {F}, through equilibrium relationships and is dened as:
3
{P} [B] {F}
(6.1)
4 [B] would have as many rows as the total number of independent equations of equilibrium;
and as many columns as independent internal forces. This is reminiscent of the equilibrium
matrix obtained in analyzing trusses by the method of joints.
5 Depending on the type of structure, the internal element forces, and the equilibrium forces will
vary according to Table 6.1. As with the exibility method, there is more than one combination
of independent element internal forces which can be selected.
Draft
62
EQUATIONS OF STATICS and KINEMATICS
Type
Truss
Beam 1
Beam 2
Beam 3
2D Frame 1
Internal Forces
Axial force at one end
Shear and moment at one end
Shear at each end
Moment at each end
Axial, Shear, Moment at each end
Equations of Equilibrium
FX = 0, FY = 0
A
A
Fy = 0, Mz = 0
A
B
Fy = 0, Fy = 0
A = 0, M B = 0
Mz
z
A
A
A
Fx = 0, Fy = 0, Mz = 0
Table 6.1: Internal Element Force Denition for the Statics Matrix
6 Matrix [B] will be a square matrix for a statically determinate structure, and rectangular
(more columns than rows) otherwise.
Example 6-1: Statically Determinate Truss Statics Matrix

Considering the truss shown in Fig. 6.1, it has 8 unknown forces (4 internal member forces
and 4 external reactions), and 8 equations of equilibrium (2 at each of the 4 nodes). Assuming
all the element forces to be tensile, and the reactions as shown in the gure, the equilibrium
equations are:
Node
Node 1
FX = 0
Px1 +F3 C Rx1 = 0
Node 2
Node 3
Px2 + F2 = 0
Px3 F2 F3 C = 0
Node 4
Px4 +Rx4 = 0
L
L2 +H 2
Py2 F1 = 0
Py3 F4 F3 S = 0
0
Py4 +F4 Ry4 = 0
where cos =
FY = 0
Py1 +F1 + F3 S Ry1 =0
= C and sin =
H
L2 +H 2
= S Those equations of equilibrium can be
Y
P
y2
P
y2
Px2
Px2
Py3
f2
f1
1
Py1 f1
Rx1
1
4
L
Px3
f2
3
f4
f3
f3
Px1
f4
Py4
Rx4
Px4
4
R y1
R y4
Figure 6.1: Example of [B] Matrix for a Statically Determinate Truss
Victor Saouma
Draft
6.1 Statics Matrix [B]
63
cast in matrix form

1
Fx
1
Fy
2
Fx
2
Fy
3
Fx
3
Fy
4
Fx
4
Fy
:
:
:
:
:
:
:
:
Px1
Py1
Px2
Py2
Px3
Py3
Px4
Py4
0
0 C 0
1 0 S 0
0 1 0
0
1
0
0
0
0
1
C
0
0
0
S
1
0
0
0
0
0
0
0 1
{P}
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
F1
F2
F3
F4
Rx1
Ry1
Rx4
Ry4
(6.2)
{F}
[B]
the unknown forces and reactions can be determined through inversion of [B]:
F1
F2
F3
F4
Rx1
Ry1
Rx4
Ry4
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
{F}
1 0
0 0
1
0 C
S
0 C
0 1
S
1 C
0 0
S
0 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
Px2
Py2
Py2
Px2
Px2
C
S
C Px2
Px2
S
C Px2
+ Py2
0
S
C Px2
(6.3)
{P}
[B]1
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:
P1
M1
P2
M2
P3
M3
P4
M4
{P}
Victor Saouma
1 0
0
0
0
0 1 0
8 1 0
0
0
0 0 0
1
0 1 0
0
0 0 1
0
1 2 1 0
0 0 0
0
0
1
0 1 0 0 0
0
0
0
1 3 1 0 0
0
0
0
0
1
0 0 0
0
0
0
0
0
1 0 0
[B]
v1
m1
v2
m2
v3
m3
R1
R2
(6.4)
{F}
Draft
64
Figure 6.2: Example of [B] Matrix for a Statically Determinate Beam

Inverting this 8 by 8 matrix would yield
v1
m1
v2
m2
v3
m3
R1
R2
{F}
6.1.2
7
0 1
8
0 0
0 0
0 0
0 0
0 0
1 1
8
0 1
8
0 1
8
0 1
0 0
0 0
0
0 0
0 1
8
1 1
8
1
4
2
1
0
0
0
1
4
5
4
1
8
1
0
1
0
0
1
8
1
8
5
8
5
1
3
1
0
5
8
13
8
1
8
1
0
1
0
1
1
8
1
8
0 5
0 40

0 20
0 0

=
20 0

0 0
0 5

25
[B]1
(6.5)
{P}
Statically Indeterminate
For the case of a statically indeterminate structure, Eq. 6.1 can be generalized as:
{P}2n1 = [ [B0 ]2n2n [Bx ]2nr ]
F0
Fx
(6.6)
(2n+r)1
where [B0 ] is a square matrix, {F0 } the vector of unknown internal element forces or external
reactions, and {Fx } the vector of unknown redundant internal forces or reactions.
8
Hence, we can determine {F0 } from

{F0 } = [B0 ]1 {P} [B0 ]1 [Bx ] {Fx }
= [C1 ]2n2n {P} [C2 ]2nr {Fx }
(6.7)
(6.8)
Note the following denitions which will be used later:
Victor Saouma
Draft
65
[B0 ]1 [C1 ]
[B0 ]
(6.9)
[Bx ] [C2 ] (6.10)
Example 6-3: Statically Indeterminate Truss Statics Matrix

Revisiting the rst example problem, but with an additional member which makes it statically indeterminate, Fig. 6.3, it now has 9 unknown forces (5 internal member forces and 4
Y
Py3
P
y2
P
y2
Px2
f2
f5
f3
3
f4
f5 f
4
Py4
f1
3
1
Py1 f1
Rx1
5
1
Px3
f2
Px2
f3
Px1
Rx4
Px4
4
R y4
R y1
Figure 6.3: Example of [B] Matrix for a Statically Indeterminate Truss

external reactions), and only 8 equations of equilibrium. Selecting the fth element force as the
redundant force, and with r = 1, we write Eq. 6.6
F0
Fx
{P}2n1 = [ [B0 ]2n2n [Bx ]2nr ]
(6.11-a)
(2n+r)1
{P} = [B0 ] {F0 } + [Bx ] {Fx }

Px1
Py1
Px2
Py2
Px3
Py3
Px4
Py4
{P}
0
0 C 0
1 0 S 0
0 1 0
0
1
0
0
0
0
1
S
0
0
0
S
1
0
0
0
0
0
0
0 1
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
F1
F2
F3
F4
Rx1
Ry1
Rx4
Ry4
{F0 }2nx1
[B0 ]2nx2n
0
0
C
S
0
0
C
S
{F5 } (6.11-b)
{Fx }
[Bx ]nx1
We can solve for the internal forces in terms of the (still unknown) redundant force
{F0 } = [B0 ]1 {P} [B0 ]1 [Bx ] {Fx }
Victor Saouma
(6.12-a)
Draft
66
F1
F2
F3
F4
Rx1
Ry1
Rx4
Ry4
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
{F0 }
1 0
0 0
1
0 C
S
0 C
0 1
S
1 C
0 0
S
0 C
0
0
0
1
0
0
0
1
0
0
0
0
0
0
1
0
Px2

Py2
0
0
0
0
0
0
0
1
{P}
[B0 ]1 [C1 ]
{F }(6.12-b)
5
0
0 {Fx }
[Bx ]
Or using the following relations [B0 ]1 [C1 ] and [B0 ]1 [Bx ] [C2 ] we obtain
F1
F2
F3
F4
Rx1
Ry1
x4
Ry4
{F0 }
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
1
0 C
S
0 C
0 1
S
1 C
0 0
S
0 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
Px2
Py2
{P}
[C1 ]
{F5 }
(6.13)
{Fx }
[C2 ]
Note, that this equation is not sucient to solve for the unknown forces, as {Fx } must be
obtained through force displacement relations ([D] or [K]).
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.
10
11
Starting with
{P}2n1 = [B]2n(2n+r) {F}(2n+r)1
[B]2n(2n+r) {F}2n+r1 [I]2n2n {P}2n1 = {0}
B I
where
B I
2n(4n+r)
F
P
= {0}
(6.14-a)
corresponds to the augmented matrix.
If we apply a Gauss-Jordan elimination process to the augmented matrix, Eq. 6.14-a is then
transformed into:
12
I2n2n C22nr
C12n2n
F0
x
P
4n+r1
= {0}
(6.15)
or:
{F0 }2n1 = [C1 ]2n2n {P}2n1 + [C2 ]2nr {Fx }r1
Victor Saouma
(6.16)
Draft
67
which is identical to Eq. 6.8; As before, Fx are the redundant forces and their solution obviously
would depend on the elastic element properties.
Example 6-4: Selection of Redundant Forces

Revisiting the statically determined truss of Example 1, but with the addition of a fth
element, the truss would now be statically indeterminate to the rst degree. The equation of
equilibrium 6.2 will then be written as:
1
Fx : A
1 :B
Fy
2
Fx : C
2 :D
Fy
3
Fx : E
3
Fy : F
4 :G
Fx
4
Fy : H
0
0 C 0
0 1 0 0 0
1 0 S 0
0 0 1 0 0
0 1 0
0 C 0 0 0 0
1
0
0
0
S 0 0 0 0
0
1
C
0
0 0 0 0 0
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
[
B I
0
0
0
0
1 0
0 1
0
0
0
0
0
0
0
0
F1
F2
F3
F4
F5
Rx1
Ry1
Rx4
Ry4
Px2
Py2
(6.17)
{F}
Note that since load is applied only on node 2, we have considered a subset of the identity
matrix [I].
1. We start with the following matrix
F1
A 0
B 1
C 0
D 1
E 0
F 0
G 0
H 0
F2 F3
0 C
0 S
1
0
0
0
1
C
0
S
0
0
0
0
F4 F5
0
0
0
0
0 C
0
S
0
0
1
0
0
C
1 S
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
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
0
F4
0
0
0
0
0
1
0
1
F5
0
0
C
S
0
0
C
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
0
(6.18)
2. Interchange columns
Rx1
A 1
B 0
C 0
D 0
E 0
F 0
G 0
H 0
Victor Saouma
Ry1
0
1
0
0
0
0
0
0
F2 F1 F3
0
0 C
0 1 S
1 0
0
0
1
0
1
0
C
0
0
S
0
0
0
0
0
0
(6.19)
Draft
68
3. Operate as indicates
B = B + DB
E+C
E = C
G
H
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
F2 F1
0
0
0
0
1 0
0
1
1
0
0
0
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
Py2
0
1
0
0
(6.20)
4. Operate as indicated
A
B
C
D
E
F
G
H
Rx1
= A + CE
1
0
= B + SE
0
= C
= F SE 0
0
=H +F
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
Py2
0
1
1
(6.21)
0
0
0
5. Interchange columns and observe that F5 is the selected redundant.

Rx1
A 1
B 0
C 0
D 0
E 0
F 0
G 0
H 0
Ry1
0
1
0
0
0
0
0
0
From Eq. 6.8 we have
Rx1
Ry1
F2
F1
F3
F4
Rx4
Ry4
{F0 }
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
1
S/C
1
0
1/C
S/C
0
S/C
[C1 ]
F4
0
0
0
0
0
1
0
0
0
1
0
1
0
0
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
Py2
{P}
Px2
1
S/C
1
0
1/C
S/C
0
S/C
Py2
0
1
0
0
{F5 }
(6.22)
(6.23)
{Fx }
[C2 ]
which is identical to the results in Eq. 6.13 except for the order of the terms.
Victor Saouma
Draft
6.2 Kinematics Matrix [A]
6.1.3
69
Kinematic Instability
13 Kinematic instability results from a structure with inadequate restraint in which rigid body
motion can occur. Kinematic instability will result in a matrix which is singular, and decomposition of this matrix will result in a division by zero causing a computer program to crash.
Hence, it is often desirable for large structures to determine a priori whether a structure is
kinematically instable before the analysis is performed.
14 Conditions for static determinacy and instability can be stated as a function of the rank
of [B]. If [B] has n rows (corresponding to the number of equilibrium equations), u columns
(corresponding to the number of internal forces and reactions), and is of rank r, then conditions
of kinematic instability are summarized in Table 6.2
n > u Kinematically Instable

n = u Statically Determinate
n = u = r Stable
n = u > r Instable with n r modes of kinematic instability
n < r Statically Indeterminate (degree u n)
n=r
Stable
n>r
Instable with n r modes of kinematic instability
Table 6.2: Conditions for Static Determinacy, and Kinematic Instability
Note that kinematic instability is not always synonymous with structure collapse. In some
cases equilibrium will be recovered only after geometry would have been completely altered (such
as with a exible cable structures) and equations of equilibrium would have to be completely
rewritten with the new geometry.
15
6.2
Kinematics Matrix [A]
16 The kinematics matrix [A] relates all the structures {} nodal total displacements in global
coordinates to the element relative displacements in their local coordinate system and the support displacement (which may not be zero if settlement occurs) {}, through kinematic relationships and is dened as:
{} [A] {}
(6.24)
[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.
17
18 It can be easily shown that for trusses (which corresponds to shortening or elongation of the
member):
e = (u2 u1 ) cos + (v2 v1 ) sin
(6.25)
Victor Saouma
Draft
610
where is the angle between the element and the X axis. whereas for exural members:
v21 = v2 v1 z1 L
(6.26-a)
z21 = z2 z1
(6.26-b)
Example 6-5: Kinematics Matrix of a Truss

Considering again the statically indeterminate truss of the previous example, the kinematic
matrix will be given by:
e
1
e
2
e
3
e
4
e
5
u1
v1
u4
v4
0
1
0
1 0 0 0
0
u1
0
0
1 0 1 0 0
0
v1
C S
0
0 C S 0
0
u2
0
0
0
0 0 1 0 1
v2
0
0 C S 0 0 C S
u
1
0
0
0 0 0 0
0 3
v3
0
1
0
0 0 0 0
0
u
4
0
0
0
0 0 0 1
0
v4
0
0
0
0 0 0 0
1
(6.27)
[A]
Applying the constraints: u1 = 0; v1 = 0; u4 = 0; and v4 = 0 we obtain:
e
1
e
2
e
3
e
4
e
5
0
0
0
0
0
1
0
1 0 0 0
0
u1
0
0
1 0 1 0 0
0
v1
C S
0
0 C S 0
0
u2
0
0
0
0 0 1 0 1
v
2
0
0 C S 0 0 C S
u
1
0
0
0 0 0 0
0 3
v3
0
1
0
0 0 0 0
0
u4
0
0
0
0 0 0 1
0
v4
0
0
0
0 0 0 0
1
(6.28)
[A]
We should observe that [A] is indeed the transpose of the [B] matrix in Eq. 6.17
6.3
Statics-Kinematics Matrix Relationship
19 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.
Victor Saouma
Draft
6.3 Statics-Kinematics Matrix Relationship
6.3.1
20
Statically Determinate
The external work being dened as

Wext = 1 P {}
2
{P} = [B] {F}
21
Wext =
1
F [B]T {}
2
(6.29)
Wint =
1
F [A] {}
2
(6.30)
Alternatively, the internal work is given by:

Wint = 1 F {}
2
{} = [A] {}
22
611
Equating the external to the internal work Wext = Wint we obtain:

1
1
F [B]T {} = F [A] {}
2
2
(6.31)
[B]T = [A]
(6.32)
23 This relationship is analogous to the duality which exists between statics and kinematics as
expressed through the curvature-area theorems.
6.3.2
Statically Indeterminate
24 Whereas in the preceding case we used Eq. 6.1 for [B], for the most general case of statically
indeterminate structures we can start from Eq. 6.6 and write:
F0
Fx
{P} = [ B0 Bx ]
(6.33)
where Fx correspond to the redundant forces. The external work will then be
Wext =
25
1
2
[B0 ]t
[Bx ]t
F0 Fx
{}
(6.34)
Again, we can generalize Eq. 6.24 and write

0
x
A0
Ax
{}
(6.35)
where {0 } and {x } are relative displacements corresponding to {F0 } and {Fx } respectively.
26
Consequently the internal work would be given by:

Wint =
27
1
2
F0 Fx
[A0 ]
[Ax ]
{}
(6.36)
As before, equating the external to the internal work Wext = Wint and simplifying, we obtain:
[B0 ]T
= [A0 ]
(6.37)
= [Ax ]
(6.38)
[Bx ]
Victor Saouma
Draft
612
6.4
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.
29
Rewriting Eq. 6.35 as

{} = [A0 ]1 {0 } = [B0 ]t
{0 } = [B0 ]1 {0 }
(6.39)
we can solve for {F0 } from Eq. 6.8

{F0 } = [B0 ]1 {P} [B0 ]1 [Bx ] {Fx }
[C1 ]
30
(6.40)
[C2 ]
Combining this equation with [B0 ]1 = [C1 ] from Eq. 6.40, and with Eq. 6.39 we obtain
{} = [C1 ]t {0 }
31
(6.41)
Similarly, we can revisit Eq. 6.35 and write

{x } = [Ax ] {}
(6.42)
When the previous equation is combined with the rightmost side of Eq. 6.39 and 6.38 we obtain
t
{x } = [Bx ]t [B0 ]1 {0 }
32
(6.43)
Thus, with [B0 ]1 [Bx ] = [C2 ] from Eq. 6.40

{x } = [C2 ]t {0 }
(6.44)
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
{Pe } =
P1
P2
{ } =
...
...
Pn
n
T
T
(6.45-a)
(6.45-b)
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.
Victor Saouma
Draft
6.5 Congruent Transformation Approach to [K]
613
34 Similarly, we can dene the unconnected (or unassembled) global stiness matrix of the
structure as [Ke ]:
{F} = [Ke ] {}
[K1 ]
[K2 ]
[Ke ] =
(6.46-a)
[K3 ]
..
(6.46-b)
[Kn ]
Note that all other terms of this matrix are equal to zero, and that there is no intersection
between the various submatrices. Hence, this matrix does not reect the connectivity among
all the elements.
35
36
We recall the following relations (Eq. 6.1, 6.24, and 6.32 respectively)
{} = [A]{}
(6.47-a)
{P} = [B]{F}
(6.47-b)
[B]
= [A]
(6.47-c)
We now combine those matrices with the denition of the stiness matrix:
{P} = [B]{F} [A]T {F} = [K]{}
{P} = [K]{}
{F} = [Ke ]{}
[K] = [A]T [Ke ][A]
{} = [A]{}
[B] = [A]
(6.48)
37 Thus, we have just dened a congruent transformation on the unconnected global stiness
matrix written in terms of [Ke ] to obtain the structure stiness matrix. We shall note that:
1. If [Ke ] is expressed in global coordinates, then [A] is a boolean matrix.

2. If [Ke ] is in local coordinates, then [A] must include transformation from element to global
coordinate systems, and is no longer boolean.
3. [K] accounts for the B.C. as those terms associated with the restrained d.o.f. are not
included.
4. Note the similarity between the direct stiness method: [K] =
congruent transformation approach: [K] = [A]T [Ke ][A].
[]T [k][] and the

n
[]T [k]66 []66

5. If the structure is a frame with n elements, then we would have [K]neqneq =
66
1
and the congruent transformation approach: [K]neqneq = [A]T
[Ke ]6n6n [A]6nneq .
neq6n
6. Congruent approach appears to be less ecient than the direct stiness method as both
[Ke ] and [A] are larger than [K].
Example 6-6: Congruent Transformation

Victor Saouma
Draft
614
Figure 6.4: Example 1, Congruent Transfer

Assemble the global stiness matrix of the grid shown in Fig. 6.4 using the direct stiness
method and the congruent transformation method.
Solution:
The 2 element stiness matrices in global coordinate system are given by:
[KAB ]
[KBC ]
7.692
0.
.4 105
1 105
0.
14.423
0.
7.692
0.
0.
12.
0.
.2 105
12.
.0048
0.
12.
.0048
7.692
0.
0.
.4 105
12.
.0048
(6.49-a)
18.75 .5 105
0.
18.75
0.
14.423
0.
0.
.00469 18.75
0.
.00469
1 105
0.
18.75
14.423
0.
.00469
(6.49-b)
We shall determine the global stiness matrix using the two approaches:
Direct Stiness
0 1 0
[ID] = 0 2 0
0 3 0
{LM1 } =
{LM } =
2
(6.50-a)
0 0 0 1 2 3
(6.50-b)
1 2 3 0 0 0
(6.50-c)
(7.692 + 1 105 )
(0. + 0.)
(18.75 + 0.)
(0. + 0.)
(.4 105 + 14.423)
(12. + 0.)
[K] =
(6.51-a)
(0. + 18.75)
(12. + 0.)
(.0048 + .00469)
Victor Saouma
Draft
615
1 105
0.
18.75
12.
0.
.4 105
=
18.75
12.
.00949
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
[Ke ] {}
7.692
0
.4 105
sym
0
12.
.0048
0.
1 105
0.
0.
14.423
sym
18.75
0.
.00469
1
x
1
y
1
Wz
2
x
2
y
2
Wz
(6.52-a)
element 1
(6.52-b)
element 2
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] {}
1
x
1

y
wz
2
x
wz
1
0
0
1
0
0
0
1
0
0
1
0
0
0
1
0
0
1
(6.53-a)
y
wz
(6.53-b)
As for the kinematics matrix, we are relating the local displacements of each
element to the global ones. Hence this matrix is analogous to the connectivity matrix.
Whereas the connectivity matrix dened earlier reected the element connection, this
one reects the connectivity among all the unrestrained degrees of freedom.
3. If we take the product: [A]T [Ke ][A] then we will recover [K] as shown above.
Example 6-7: Congruent Transformation of a Frame

Assemble the stiness matrix of the frame shown in Fig. 6.5 using the direct stiness
method, and the two congruent approaches.
Solution:
The stiness matrices of elements AB and BC in local coordinate system are given by:
[k]AB = [k]BC
Victor Saouma
.75
= 200
sym
0.
.00469
0.
.75
0.
0.
18.75
0
.0048 18.75
1 105
0
18.75 .5 105
(6.54)
.75
0.
0.
.00469 18.75
1 105
Draft
616
Figure 6.5: Example 2

while the rotation matrix is given by:
[]AB
[]BC
.9272 .375 0.
0.
0.
0.
0.
.375 .927 0.
0.
0.
1
0.
0.
0.
0. 0. .9272 .375
0.
0. 0. .375 .9272
0.
0.
0.
0. 0.
0.
0.
0.
0.
0.
1
= [I]
(6.55-a)
(6.55-b)
The element stiness matrices in global coordinates will then be given by:
[K]AB = []T [k]AB []AB
AB
.645 .259 7.031 .645 .259

7.031
.109 17.381 .259 .109

17.381
5
1 10 7.031 17.381 .5 105
= 200
.645
.259
7.031
sym
.109
17.381
1 105
(6.56-a)
(6.56-b)
and [K]BC = [k]BC

Direct Stiness: We can readily assemble the global stiness matrix:
(.645 + .75)
[K] = 200
sym
1.395
= 200
sym
.259
.1137
(.259 + 0.)
(7.031 + 0.)
(.109 + .00469) (17.38 + 18.75)

(1 + 1) 105
(6.57-a)
7.031
1.37
2 105
(6.57-b)
Congruent Transformation, global axis, Boolean [A]

Victor Saouma
Draft
617
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
[Ke ] {}
200
.645
.259
.109
sym
7.031
17.381
1 105
0
.75
0.
.00469
0
sym
0.
18.75
1 105
1
UX
1
VY
1
Z
2
UX
2
VY
2
Z
(6.58-a)
(6.58-b)
2. Next we determine the kinematics matrix A:

{} = [A] {}
1 0
0 1
0 0
=
1 0
u2
0 1
v2
2
0 0
1
u
1
v
0
0
1
0
0
1
(6.59-a)
(6.59-b)
3. Finally, if we take the product [A]T [Ke ] [A] we obtain the structure global stiness
36 66 63
matrix [K] in Eq. 6.57-b

Congruent Transformation (local axis):
1. Unconnected stiness matrix in local coordinates:
{pe }
1
PX
1
PY
1
MZ
2
PX
2
PY
2
MZ
[ke ] {e }
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
(6.60-a)
(6.60-b)
2. The kinematics matrix [A] is now given by:
{ e } = [A] {}
u1
.9272 .375 0.
x
1
.375 .9272 0.
vy
0.
1
0.
1. X
z
V
=
1.
u2
0.
0. Y
Z
2
0.
vy
1.
0.
2
0.
0.
1.
z
(6.61-a)
(6.61-b)
3. When the product: [A]T [ke ][A] we recover the structure global stiness matrix
Victor Saouma
Draft
618
Victor Saouma
Draft
Chapter 7
FLEXIBILITY METHOD
7.1
1
Introduction
Recall the denition of the exibility matrix

{} [d]{p}
(7.1)
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) } =
{
(e)
} =
F(1)
F(2)
(1)
...
(2)
...
F(n)
(n)
(7.2-a)
T
(7.2-b)
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.
Denoting by {R} the reaction vector, and by {R } the corresponding displacements, we
dene the unassembled structure exibility matrix as:
3
(e)
R
d(e)
[0]
F(e)
R
(7.3)
where d(e) is the unassembled global exibility matrix.

In its present form, Eq. 7.3 is of no help as the element forces F(e) and reactions {R} are
not yet known.
4
7.2
Flexibility Matrix
We recall from Sect. 6.1.2 that we can automatically identify the redundant forces [Fx ] and
rewrite Eq. 7.3 as:
(e)
0
d00 [0]
F0
=
(7.4)
(e)
x
Fx
[0] dxx
5
Draft
72
(e)
FLEXIBILITY METHOD
(e)
where d00 and dxx correspond to the unassembled global exibility matrix, and {F0 } and
{Fx } are the corresponding forces.
Next we must relate the redundant and nonredundant forces (which together constitute the
unknown element forces and reactions) to the externally applied load {P}. Hence we recall
from Eq. 6.16:
(7.5)
{F0 } = [C1 ] {P} + [C2 ] {Fx }
6
which can be rewritten (for convenience:) as:

F0
Fx
C1 C2
0 I
P
Fx
(7.6)
From Eq. 6.41 we had: {} = [C1 ]t {0 } and from Eq. 6.44: {x } = [C2 ]t {0 } which
lead to
t
p
C1 0
0
=
(7.7)
t
0
C2 I
x
where the subscript p in {p } has been added to emphasize that we are referring only to the
global displacements corresponding to {P}.
8
Finally, we substitute Eq. 7.6 into Eq. 7.4 and the results into Eq. 7.7 to obtain:
p
0
(e)
t
C1 0
t
C2 I
d00
0
C1 C2
0 I
0
(e)
dxx
or:
?
p
Dpp Dpx
Dxp Dxx
P
Fx
(7.8)
P
F?
x
(7.9)
where:
(e)
(7.10-a)
(e)
(7.10-b)
(e)
(7.10-c)
[Dpp ] = [C1 ]t d00 [C1 ]

[Dpx ] = [Dxp ]t = [C1 ]t d00 [C2 ]
(e)
[Dxx ] = [C2 ]t d00 [C2 ] + [dxx ]
This equation should be compared with Eq. 7.1 and will be referred to as the unsolved global
assembled exibility equation.
9
10 Finally, this equation should also be contrasted with its counterpart in the stiness method,
Eq. 5.2.
7.2.1
Solution of Redundant Forces
We can solve for the redundant forces (recall that in the exibility method, redundant forces
are the primary unknowns as opposed to displacements in the stiness method) by solving the
lower partition of Eq. 7.9:
11
{Fx } = [Dxx ]1 [Dxp ] {P}
(7.11)
Again, this equation should be compared with Eq. 5.3.

Victor Saouma
Draft
7.2 Flexibility Matrix
7.2.2
12
73
Solution of Internal Forces and Reactions
The internal forces and reactions can in turn be obtained through Eq. 6.16:
{F0 } = [C1 ] {P} + [C2 ] {Fx }
(7.12)
which is combined with Eq. 7.11 to yield:

{F0 } = [C1 ] [C2 ] [Dxx ]1 [Dxp ] {P}
7.2.3
13
(7.13)
Solution of Joint Displacements
Joint displacements are in turn obtained by considering the top partition of Eq. 7.9:
{p } = [Dpp ] [Dpx ] [Dxx ]1 [Dxp ] {P}
(7.14)
[D]
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 f5 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
(e)
u1
(e)

2
(e)
u4
(e)
(e)
3
(e)
4
Victor Saouma
d00
0
AE
0
(e)
dxx
F0
Fx
(7.15-a)
0
0
2
1.5
1.25
1.5
0
0
1.25
Rx1
Ry1
f2
f1
f3
f4
Rx4
Ry4
f5
(7.15-b)
Draft
74
FLEXIBILITY METHOD
From Example 6.1.2 we have
[C1 ] =
1
S/C
1
0
1/C
S/C
0
S/C
0
1
0
1
0
0
0
0
1
0.75
1
0
1.25
0.75
0
0.75
0
1
0
1
0
0
0
0
[C2 ] =
0.8
0.8
0.6
0.6
0.8
(7.16)
From Eq. 7.10-a

(e)
[Dpp ] = [C1 ]t d00 [C1 ] =
L
AE
4.797
0
0
1.50
(7.17)
3.838
0.900
(7.18)
From Eq. 7.10-b

(e)
[Dpx ] = [C1 ]t d00 [C2 ] =
L
AE
From Eq. 7.10-c

L
(4.860)
AE
We can now solve for the redundant force f5 from Eq. 7.11
(e)
(e)
[Dxx ] = [C2 ]t d00 [C2 ] + [dxx ] =
{Fx } = [Dxx ]1 [Dxp ] {P} =
1
4.860
3.8387 0.90
(7.19)
Px2
Py2
f5 = 0.790Px2 + 0.185Py2
(7.20-a)
(7.20-b)
The nonredundant forces are now obtained from Eq. 7.13
{F0 } =
Rx1
Ry1
f2
f1
f3
f4
Rx4
Ry4
f5
[C1 ] [C2 ] [Dxx ]1 [Dxp ] {P}
Victor Saouma
1
0.75
1
0
1.25
0.75
0
0.75
0
1
0
1
0
0
0
0
(7.21-a)
0.632 0.148
0
0
0.632 0.148
0.474 0.111
0.790 0.185
0.474 0.111
0.632 0.148
0
0
Px2
Py2
(7.21-b)
0.368
0.148
0.75
1.000
0.368 0.148
0.474
0.889
0.460
0.185
0.276 0.111
0.632 0.148
0.750
0
Px2
Py2
(7.21-c)
Draft
7.3 Stiness Flexibility Relations
75
Figure 7.1: Stable and Statically Determinate Element

Finally, the displacements are obtained from Eq. 7.14
{p } =
[Dpp ] [Dpx ] [Dxx ]1 [Dxp ] {P}
(7.22-a)
[D]
L
AE
u2
v2
L
AE
4.797
0
0
1.500
1.766 0.711
0.711 1.333
1
4.860
14.730 3.454
3.454 0.810
Px2
Py2
Px2
Py2
(7.22-b)
(7.22-c)
It should be noted that whereas we have used the exibility method in its algorithmic
implementation (as it would lead itself to computer implementation) to analyse this simple
problem, the solution requires a formidable amount of matrix operations in comparaison with
the classical (hand based) exibility method.
7.3
Stiness Flexibility Relations
14 Having introduced both the stiness and exibility methods, we shall rigorously consider the
relationship among the two matrices [k] and [d] at the structure level.
Let us generalize the stiness relation by partitioning it into two groups: 1) subscript s for
those d.o.f.s which are supported, and 2) subscript f for those dof which are free.
15
pf
ps
kf f
ksf
kf s
kss
f
s
(7.23)
We note that this is the same as equation 5.2, however it is expressed at the element level
and in the local coordinate system.
16
7.3.1
From Stiness to Flexibility
To obtain [d] the structure must be supported in a stable and statically determinate way, as
for the beam in Fig. 7.1. for which we would have:
17
Victor Saouma
Draft
76
FLEXIBILITY METHOD
{ f } =
{pf } =
M1
M2
(7.24-b)
{ps } =
18
1
2
V1
V2
(7.24-c)
and { s } =
v1
v2
(7.24-a)
Since { s } = {0} the above equation reduces to:

pf
ps
kf f
ksf
{ f }
(7.25)
and we would have:

{pf } = [kf f ] { f }
(7.26)
[d] = [kf f ]1
(7.27)
Example 7-2: Flexibility Matrix

From Fig. 7.1
M1
M2
EI
L
4 2
2 4
1
2
(7.28)
[kf f ]
[kf f ]1 = [d] =
7.3.2
L 1
EI 12
4 2
2
4
L
6EI
2 1
1
2
(7.29)
From Flexibility to Stiness
[kf f ]: From Eq. 7.23, [k] was subdivided into free and supported d.o.f.s, and we have shown
that [kf f ] = [d]1 , or {pf } = [kf f ] { f } but we still have to determine [kf s ], [ksf ], and [kss ].
19
20
[ksf ]: Since [d] is obtained for a stable statically determinate structure, we have:
{ps } = [B] {pf }
(7.30-a)
{ps } = [B] [kf f ]{ f }
(7.30-b)
[ksf ]
[ksf ] = [B] [d]1
21
(7.30-c)
[kf s ]: Equating the external to the internal work:
Victor Saouma
Draft
7.3 Stiness Flexibility Relations

1. External work: Wext =
2. Internal work: Wint =
1
2
1
2
77
f {pf }
ps { s }
Equating Wext to Wint and combining with

ps = f [ksf ]T
(7.31)
from Eq. 7.23 with { s } = {0} (zero support displacements) we obtain:

[kf s ] = [ksf ]T = [d]1 [B]T
22
(7.32)
[kss ]: This last term is obtained from

{ps } = [B] {pf }
(7.33-a)
{pf } = [kf s ] { s }
1
[kf s ] = [d]
[B]
(7.33-b)
(7.33-c)
Combining Eqns. 7.33-b, 7.33-a, and 7.33-c we obtain:

{ps } = [B][d]1 [B]T { s }
(7.34)
[kss ]
23
In summary we have:
[k] =
[d]1
[d]1 [B]T
1 [B][d]1 [B]T
[B][d]
(7.35)
A very important observation, is that the stiness matrix is obviously singular, since the
second row is linearly dependent on the rst one (through [B]) and thus, its determinent is
equal to zero.
24
Example 7-3: Flexibility to Stiness

With reference to Fig. 7.1, and with both M1 and M2 assumed to be positive (ccw):
1. The exibility matrix is given by:
1
2
L
6EI
2 1
1
2
M1
M2
(7.36)
[d]
2. The statics matrix [B] relating external to internal forces is given by:
V1
V2
1
L
1
1
1 1
M1
M2
(7.37)
[B]
Victor Saouma
Draft
78
FLEXIBILITY METHOD
3. [kf f ]: would simply be given by:

[kf f ] = [d]1 =
EI
L
4 2
2 4
(7.38)
4. [kf s ]: The upper o-diagonal

EI
L
[kf s ] = [d]1 [B]T =
1 1
1 1
1
L
4 2
2 4
EI
L2
6 6
6 6
(7.39)
5. [ksf ]: Lower o-diagonal term

[ksf ] = [B][d]1 =
1
L
1
1
1 1
EI
L
4 2
2 4
EI
L2
6
6
6 6
(7.40)
6. [kss ]:
[kss ] = [B] [d]1 [B]T = [ksf ] [B]T
EI 1
L2 L
1 1
1 1
6
6
6 6
(7.41-a)
=
12 12
12
12
EI
L3
(7.41-b)
Let us observe that we can rewrite:
M1
V1
EI
L3
V2
4L2 2L2
6l 6l
6l 6l
2L2 4L2
6l
6l
12 12
12
6l 6l 12
v1
(7.42)
v2
If we rearrange the stiness matrix we would get:
V1
V2
EI
L
M2
12
L2
6
L
12
L2
6
L
6
L
6
L
12
L2
6
L
12
L
6
L
v1
2 1
6
v2
6
L
(7.43)
[k]
and is the same stiness matrix earlier derived.

25
Alternatively, if we begin with Equation 7.43 we observe that it can be expressed as
[k] =
r
r
r
r
k11 +k12 +k21 +k22
L2 r
r
k11 +k21
L
kr +kr +kr +kr
11 12L2 21 22
r
r
k12 +k22
L
r
r
k11 +k21
L
r
k11
r
r
k12 +k22
L
r
k21
r
r
r
r
k11 +k12 +k21 +k22
L2 r
kr +k
22 L 12
r
r
r
r
k11 +k12 +k21 +k22
L2 r
kr +k
22 L 12
r
r
k22 +k12
L
r
k12
r
r
k22 +k12
L
r
k22
(7.44)
where k r corresponds to kf f
Victor Saouma
Draft
7.4 Stiness Matrix of a Curved Element
7.4
79
Stiness Matrix of a Curved Element
25 We seek to determine the stiness matrix of a circular arc of radius R and sustaining an
angle .
26
First, we determine the exibility matrix of a cantilevered arc from

U =
M
s
R
M
dx =
EI
EI
M M d
(7.45)
where M is the real moment at arbitrary point A caused by loads and M is the virtual moment
at A caused by unit load
27
The exibility matrix will thus be given by:
f11 f12 f13 N
v
V
= f21 f22 f23
f31 f32 f33
(7.46)
and
M
M
fij
= M3 + N (R R cos ) + V (R2 sin )
(7.47-a)
+ 13 = R(1 cos1 ) + R sin2
(7.47-b)
(7.47-c)
f11 =
=
=
=
=
Disp. in DOF i caused by unit load in DOF j

R
Mp1 MD1 d
EI o
R 2
R (1 cos )2 d
EI o
R3
(1 2 cos + cos2 )d
EI o
R3
[ 2 sin + /2 = 1/4 sin 2]
o
EI
R3
EI
3
2
f12 = f21 =
=
=
=
=
Victor Saouma
R
EI
R2 sin (1 cos )d
R3
(sin cossin)d
EI o
1
R3
cos sin2
EI
2
o
3
R
1
( cos sin2 ) (1 0)
EI
2
f21 =
f13 = f31 =
=
2 sin + 1 sin 2
4
R3
EI
R
EI
1 cos 1 sin2
2
R(1 cos )d
R2
[ sin ]
o
EI
(7.47-d)
(7.47-e)
(7.47-f)
(7.47-g)
(7.47-h)
(7.47-i)
(7.47-j)
(7.47-k)
(7.47-l)
(7.47-m)
(7.47-n)
(7.47-o)
Draft
710
FLEXIBILITY METHOD
=
f22 =
=
=
R2
EI
[ sin ]
R 2 2
R sin d
EI o
R3 1
sin 2
EI
4
o
3
R
EI
(7.47-q)
(7.47-r)
1 sin 2
4
(7.47-s)
R sin d
(7.47-t)
R2
[ cos ]
o
EI
R2
EI
f33 =
R
EI
7.5
R
f22 EI
f23 = f32 =
=
(7.47-p)
(7.47-u)
R
EI
[ cos + 1]
(7.47-v)
(7.47-w)
(7.47-x)
Duality between the Flexibility and the Stiness Methods
Indeterminancy
Gov. Equat.
Primary Unknows
Variational Principle
Victor Saouma
FLEXIBILITY
Static
Kinematic
Nodal Forces
Virtual Force
{p} = [B ] {p}
[d]
[D] = [B]T [d][B]
D11 D12
p
P
=
D21 D22
R?
R
{R} = [D22 ]1 ({R } [D21 ] {p})
{P } = [D11 ]1 ({p} + [D12 ] {R})
{p} = [B] {p}
STIFFNESS
Kinematic
Equilibrium
Nodal Displacements
Virtual Displacement
{} = [] {}
[k]
[k] = []T [k][]
kf f Df r
p
f
=
?
Drf Drr
R
r
{f } = [Kf f ]1 ({p} [kf r ] {r })
{R} = [Krf ] {f } + [krr ] {r }
{p} = [k][] {}
Draft
7.6 Homework
7.6
711
Homework
Derive the statics matrix for the following semi-circular box-girder of radius R.
Using the, previously derived, exibility matrix, derive the element stiness matrix in terms
of R and
11
00
11
00
x
z
Y
O
Victor Saouma
Draft
712
Victor Saouma
FLEXIBILITY METHOD
Draft
Chapter 8
SPECIAL ANALYSIS
PROCEDURES
To be edited
8.1
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
1
2 The stiness relationship for the beam element (with shear deformation) was previously derived in Eq. 2.45
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
Flexible
1
0
1
0
1
0
1
0
1
0
1
0
11
00
11
00
11
00
Rigid
1
0
1
0
1
0
Semi-Rigid
Figure 8.1: Flexible, Rigid, and Semi-Rigid Beams
Draft
82
SPECIAL ANALYSIS PROCEDURES

1
11
00
11
00
11
00
M1
y
11
00
11
00
11
00
M2
M1 1
M1 1
M2 2
M2 2
Figure 8.2: Deformations of a Semi-Rigid Beams

w
M1
1
0
1
0
1
0
1
0
1
0
1
0
R1
M2
R2
L
Figure 8.3: Fixed End Forces in a Semi-Rigid Beams

eliminating 1 and 2 from above, yielding
M1 =
EI
EI
(S11 1 + S12 2 ) and M2 =
(S21 1 + S22 2 )
L
L
(8.3)
where
S11 =
S22 =
S12 =
D =
1 =
2 =
4 + y + 122
D
4 + y + 121
D
2 y
D
1 + y + (4 + y )(1 + 2 ) + 121 2
EI
s
k1 L
EI
s
k2 L
(8.4-a)
(8.4-b)
(8.4-c)
(8.4-d)
(8.4-e)
(8.4-f)
For the xed end actions, we have a net rotation at the end of the beam of 1 and 2 . From
beam theory, the rotation of a simply supported uniformly loaded beam is wL3 /24EI, and from
Eq. 2.6, Fig. 8.3,
4
1 =
2 =
Victor Saouma
M1 L M2 L
wL3
24EI
3EI
6EI
M2 L M1 L
wL3
24EI
3EI
6EI
(8.5-a)
(8.5-b)
Draft
8.2 Nonuniform Torsion
83
Yet, for the two springs, the moment rotation relationship is given by
M1 = k1 1 and M2 = k2 2
(8.6)
As before, we eliminate 1 and 2 from the above equations, and solve for M1 , M2 , and the
reactions (from Statics), yielding
wL2 3(1 + 62 )
12
D1
wL2 3(1 + 61 )
=
12
D1
M1 M2
= 0.5wL +
L
M1 M2
= 0.5wL
L
M1 =
(8.7)
M2
(8.8)
R1
R2
(8.9)
(8.10)
where D1 = (2 + 61 )(2 + 62 ) 1
5
Similarly for concentrated midspan load, it can be shown that

P L 6(1 + 62 )
8
D1
P L 6(1 + 61 )
=
8
D1
M 1 M2
= 0.5P +
L
M 1 M2
= 0.5P
L
M1 =
(8.11)
M2
(8.12)
R1
R2
8.2
Inclined Supports
8.4
Condensation
8.5
Substructuring
8.6
Reanalysis
8.7
(8.14)
Nonuniform Torsion
8.3
(8.13)
Constraints
Victor Saouma
Draft
84
Victor Saouma
SPECIAL ANALYSIS PROCEDURES
Draft
Part II
Introduction to Finite Elements
Draft
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
X3
33
32
31
23
13
21
X3
22
X2
12
11
X1
X2
X1
Figure 9.1: Stress Components on an Innitesimal Element

the direction of outward facing normal, and
force.
11
= ij = 21
31
the second one (j) to the direction of component
t1
12 13
22 23 =
t
2
t
32 33
3
(9.1)
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.
2
Draft
92
X3
X3
V3
33
32
t3
23
31
t2
13
X1
X2
22
V1
X2
11
V2
X1
21
t1
(Components of a vector are scalars)
12
Stresses as components of a traction vector
(Components of a tensor of order 2 are vectors)
Figure 9.2: Stresses as Tensor Components

3 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.
9.1.1
Stress Traction Relation
4 The relation between stress tensor ij at a point and the stress vector ti (or traction) on a
plane of arbitrary orientation, can be established through the following, Fig. 9.3.
(9.2)
ti = nj ij
t1 t2 t3
n1 n2 n3
direction cosines
11 12 13
21 22 23
31 32 33
(9.3)
stress tensor
where n is a unit outward vector normal to the plane.
Note that the stress is dened at a point, and a traction is dened at a point and with respect
to a given plane orientation.
5
When expanded in cartesian coordinates,, the previous equation yields

tx = xx nx + xy ny + xz nz
ty = yx nx + yy ny + yz nz
tz = zx nx + zy ny + zz nz
(9.4)
Example 9-1: Stress Vectors

Victor Saouma
Draft
9.1 Stress
93
X2
-t
1
-t
S3
*
tn S
h N
n
A
X1
-t 2 S2
*
X3
*
Figure 9.3: Stress Traction Relations

if the stress tensor at point P is given by
t1
7 5 0
t
= 5 3 1 =
2
t
0
1 2
3
(9.5)
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 = ABAC =
e1 e2 e3
4 2 0
4 0 6
(9.6-a)
= 12e1 + 24e2 + 8e3
(9.6-b)
The unit normal of N is given by

3
6
2
n = e1 + e2 + e3
7
7
7
(9.7)
Hence the stress vector (traction) will be
3
7
and thus t = 9 e1 + 5 e2 +
7
7
Victor Saouma
6
7
2
7
7 5 0
5 3 1 =
0
1 2
9
7
5
7
10
7
(9.8)
10
7 e3
Draft
94
9.2
7
Strain
Given the displacement ui of a point, the strain ij is dened as

ij =
1
(ui,j + uj,i )
2
(9.9)
1
2
uj
ui
+
xj
xi
(9.10)
or
ij =
8
When expanded in 2D, this equation yields:

11 =
12 =
22 =
21 =
1
2
1
2
1
2
1
2
u1
x1
u1
x2
u2
x2
u2
x1
u1
x1
u2
+
x1
u2
+
x2
u1
+
x2
+
u1
x1
12
=
2
u2
=
x2
21
=
2
=
(9.11-a)
(9.11-b)
(9.11-c)
(9.11-d)
Initial (or thermal strain)
ij =
T
= (1 + ) T
Plane Stress
(9.12)
Plane Strain
note there is no shear strains caused by thermal expansion.

10
The strain may also be expressed as

= Lu
or
xx
yy

zz
xy
xz
yz
Victor Saouma
0
=
y

(9.13)
0
0
ux
uy
0
u
(9.14)
Draft
9.3 Fundamental Relations in Elasticity
9.3
9.3.1
95
Fundamental Relations in Elasticity

Equation of Equilibrium
Starting with the set of forces acting on an innitesimal element of dimensions dx1 dx2 dx3 ,
Fig. 9.4 and writing the summation of forces, will yield
11
(9.15)
ij,j + bi = 0
where is the density, bi is the body force (including inertia).
yy
+
yy y
d
y
+
yx
dy
yx y
d
y
xx
+
xx
+
xy
xy
xx
dx
x
xy x
d
x
yx
yy
dx
Figure 9.4: Equilibrium of Stresses, Cartesian Coordinates
12
When expanded in 3D, this equation yields:

11 12 13
+
+
+ b1 = 0
x1
x2
x3
21 22 23
+
+
+ b2 = 0
x1
x2
x3
31 32 33
+
+
+ b3 = 0
x1
x2
x3
13
(9.16-a)
Alternatively, the equation of equilibrium can be written as

LT + b = 0
Victor Saouma
(9.17)
Draft
96
or
14
0
0
xx
yy
zz
xy
xz
yz
bx
y
b
z
=0
(9.18)
Expanding
xx
x
yx
x
zx
x
xy
+ y + xz + bx = 0
z
yy
yz
+ y + z + by = 0
zy
+ y + zz + bz = 0
z
(9.19)
Example 9-2: Equilibrium Equation

In the absence of body forces, does the following stress distribution
x2 + (x2 x2 )
2x1 x2
0
x
2
1
2 + (x2 x2 )
2x1 x2
x1
0
2
1
0
0
(x2 + x2 )
1
2
(9.20)
where is a constant, satisfy equilibrium?

Solution:
T1j
xj
T2j
xj
T3j
xj
=
=
=
T11 T12 T13

+
+
= 2x1 2x1 = 0
x1
x2
x3
T21 T22 T23

+
+
= 2x2 + 2x2 = 0
x1
x2
x3
T31 T32 T33

+
+
=0
x1
x2
x3
(9.21-a)
(9.21-b)
(9.21-c)
Therefore, equilibrium is satised.
9.3.2
Compatibility Equation
If ij = 1 (ui,j + uj,i ) then we have six dierential equations (in 3D the strain tensor has a
2
total of 9 terms, but due to symmetry, there are 6 independent ones) for determining (upon
integration) three unknowns displacements ui . Hence the system is overdetermined, and there
must be some linear relations between the strains.
15
It can be shown (through appropriate successive dierentiation of Eq. ??) that the compatibility relation for strain reduces to:
16
2 jk
2 jj
2 ij
2 ik
+
= 0.
xj xj
xi xk
xi xj
xj xk
Victor Saouma
(9.22)
Draft
9.4 Stress-Strain Relations in Elasticity
97
17 In 3D, this would yield 9 equations in total, however only six are distinct. In 2D, this results
in (by setting i = 2, j = 1 and l = 2):
2 12
2 11 2 22
+
=
x1 x2
x2
x2
2
1
(9.23)
(recall that 212 = 12 .

18
When he compatibility equation is written in term of the stresses, it yields:

2 11
22 2 2 22
2 11
2 21
= 2 (1 + )
x1 x2
x2
x2
x2
x2
2
2
1
1
9.4
19
(9.24)
Stress-Strain Relations in Elasticity
The Generalized Hookes Law can be written as:

ij = Dijkl kl
i, j, k, l = 1, 2, 3
(9.25)
The (fourth order) tensor of elastic constants Cijkl has 81 (34 ) components however, due to
the symmetry of both and , there are at most 36 9(91) distinct elastic terms.
2
20
For the purpose of writing Hookes Law, the double indexed system is often replaced by a
simple indexed system with a range of six:
21
62 =36
k = Dkm m
k, m = 1, 2, 3, 4, 5, 6
(9.26)
For isotropic bodies (elastic properties independent of reference system used to describe it),
it can be shown that the number of independent elastic constants is two. The stress strain
relations can be written in terms of E and as:
22
ij
ij
1+
ij ij kk
E
E
E
ij kk
ij +
1+
1 2
(9.27)
(9.28)
where ij is the kroneker delta and is equal to 1 if i = j and to 0 if i = j. When the strain
equation is expanded in 3D cartesian coordinates it would yield:
xy
yy
zz
xy
yz
zx
Victor Saouma
=
=
=
=
=
=
1
E [xx (yy + zz )]
1
E [yy (zz + xx )]
1
E [zz (xx + yy )]
1+
E xy
1+
E yz
1+
E zx
(9.29)
Draft
98
23
When the stress equation is expanded in 3D cartesian coordinates, it will yield:

xx
yy
zz
xy
yz
zx
E
= (12)(1+) [(1 )xx + (yy + zz )]
E
= (12)(1+) [(1 )yy + (zz + xx )]
E
= (12)(1+) [(1 )zz + (xx + yy )]
= xy
= yz
= zx
(9.30)
where is the shear modulus and xy = 2xy .

24
In terms of Lam constant we would have

e
ij
ij
= ii ij + 2ij
1
kk
=
ij ij
2
3 + 2
(9.31)
(9.32)
where
=
=
25
E
(1 + )(1 2)
E
2(1 + )
(9.33)
(9.34)
In terms of initial stresses and strains

0
ij = Dijkl (kl 0 ) + ij
kl
9.5
(9.35)
Strain Energy Density
Any elastically deforming body possesses a uniquely dened strain energy1 density which
can be expressed as:
26
U0 =
9.6
27
1
1
ij dij = ij ij = Dijkl ij kl
2
2
0
(9.36)
Summary
Fig. 9.5 illustrates the fundamental equations in solid mechanics.
Used in the evaluation of Energy release rate later on.
Victor Saouma
Draft
9.6 Summary
99
Essential B.C.
ui : u
Body Forces
Displacements
bi
ui
Equilibrium
Kinematics
ij,j + bi = 0
ij =
?
Stresses
ij
1
2
ui
xj
uj
xi
?
-
Constitutive Rel.
ij = Dijkl kl
Strain
ij
Natural B.C.
ti : t
Figure 9.5: Fundamental Equations in Solid Mechanics
Victor Saouma
Draft
910
Victor Saouma
Draft
Chapter 10
VARIATIONAL AND ENERGY

METHODS
10.1
Work, Energy & Potentials; Denitions
10.1.1
Introduction
Work is dened as the product of a force and displacement

W
def
F.ds
(10.1-a)
dW
Fx dx + Fy dy
(10.1-b)
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.
3
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
102
VARIATIONAL AND ENERGY METHODS
1111111111
0000000000 U
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
1111111111
0000000000
U0
11111
d 00000
11111
00000
1111111111
d 0000000000
1111111111
0000000000
11111
00000
11111
00000
11111
00000
11111
00000
d
Nonlinear
d
Linear
Figure 10.1: Strain Energy and Complementary Strain Energy
10.1.2
6
Internal Strain Energy
The strain energy density of an arbitrary material is dened as, Fig. 10.1
def
U0 =
(10.4)
The complementary strain energy density is dened
U0 =
def
(10.5)
The strain energy and the complementary strain energy are equal to
U
def
def
U0 d
(10.6)
U0 d
(10.7)
where is the volume of the system.

9 To obtain a general form of the internal strain energy, we rst dene a stress-strain relationship
accounting for both initial strains and stresses
= D(
0)
+ 0
(10.8)
where D is the constitutive matrix; is the strain vector due to the displacements u;
initial strain vector; 0 is the initial stress vector; and is the stress vector.
is the
10 The initial strains and stresses are the result of conditions such as heating or cooling of a
system or the presence of pore pressures in a system.
11
The strain energy U for a linear elastic system is obtained by substituting

= D
(10.9)
with Eq. 10.4 and 10.8

U=
Victor Saouma
1
2
D d
D 0 d +
0 d
(10.10)
Draft
10.1 Work, Energy & Potentials; Denitions
103
where is the volume of the system.

Considering uniaxial stresses, in the absence of initial strains and stresses, and for linear
elastic systems, Eq. 10.10 reduces to
12
1
2
U=
13
(10.11)
E d
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
U=
1
2
P2
dx
0 AE
L
(10.12)
Torsional Members:
1
2
1
2
xy
G
Tr
J
E d
xy Gxy d
xy
xy =
xy =
d = rddrdx
r
U=
1
2
T2
dx
0 GJ
(10.13)
M2
dx
0 EIz
(10.14)
r2 ddr
Flexural Members:
1
2
x = Mz y
Iz
y
= Mzz
EI
d = dAdx
A
10.1.2.1
y 2 dA = Iz
U=
1
2
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.
14
Wi =
Victor Saouma
d d
(10.15)
Draft
104

x
x
C,D
11
00
11
00
11
00 x dx
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00x dx
11
00
11
00
11
00
11
00
11
00
11
00
11
00
00
11
A,B
x
0
x
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
111
000
0
( x) F
( x) F
(b)
(a)
Figure 10.2: Eects of Load Histories on U and Wi

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.
16
Thus, in the absence of initial strains,

U = Wi
(10.16)
17 The internal work depends on the load history, this is illustrated by considering an axial
member subjected to two load cases, Fig. 10.2: a) Initial thermal strains (with no corresponding
stress increase), followed by an external force; and b) External force, followed by thermal strain.
In both cases the internal work is equal to the area under the curve ABCD.
Ui =
(x )F
0
xx dx Adx
Wia = Ui
(10.17-a)
(10.17-b)
Wib = Ui
L
0
xx dx Adx
(10.17-c)
Hence, Wi is not always equal to Ui .
10.1.3
18
External Work
External work W performed by the applied loads on an arbitrary system is dened as

def
We =
uT bd +
uT
td
(10.18)
where b is the body force vector; is the applied surface traction vector; and t is that portion
t
of the boundary where is applied, and u is the displacement.
t
Victor Saouma
Draft

19
For point loads and moments, the external work is

We =
20
105
P d +
M d
(10.19)
For linear elastic systems, we have for point loads

P
= K
We =
P d
We = K
1
d = K2
f
2
(10.20)
When this last equation is combined with Pf = Kf we obtain

1
We = Pf f
2
(10.21)
where K is the stiness of the structure.

21
Similarly for an applied moment we have

1
We = Mf f
2
10.1.3.1
(10.22)
Path Independence of External Work
22 In this section we seek to prove that the total work performed in going from state A to B is
independent of the path.
23
In 2D the dierential expression of the work is given by

dW = Fx dx + Fy dy
24
(10.23)
From calculus, a necessary and sucient condition for dW to be an exact dierential is that
Fy
Fx
=
y
x
(10.24)
25 If the force were to move along a closed contour (or from A to B and then back to A along
any arbitrary path), corresponding to , then from Greens theorem (Eq. 4.3) we have
(Rdx + Sdy) =
S R
dxdy
x
y
(10.25)
If we let R = Fx and S = Fy , then

W =
(Fx dx + Fy dy) =
Fx
Fy
dxdy
x
y
(10.26)
0
26
Thus, from Eq. 10.24 the work is equal to zero,
Victor Saouma
Draft
106
27
If we decompose the path

B
W =
+
A
=0
B
A
(10.27)
B
then, the integration for the work leads to

B
W =
A
(10.28)
(Fx dx + Fy dy)
which is path independent.

28 Note that if no net work is done in moving around a closed path, the system is said to be
conservative. This is the case for purely elastic systems.
When friction or plastic (or damped) deformations occur, then we would have a nonconservative system.
29
10.1.4
Virtual Work
30 We dene the virtual work done by the load on a body during a small, admissible (continuous
and satisfying the boundary conditions) change in displacements.
Internal Virtual Work Wi
def
External Virtual Work We
def
(10.29)
tud +
bud
(10.30)
where all the terms have been previously dened and b is the body force vector.
10.1.5
Complementary Virtual Work
10.1.5.1
Internal Virtual Work
31 Next we shall derive a displacement based expression of U for each type of one dimensional
structural member. It should be noted that the Virtual Force method would yield analogous
ones but based on forces rather than displacements.
32 Two sets of solutions will be given, the rst one is independent of the material stress strain
relations, and the other assumes a linear elastic stress strain relation.
10.1.5.1.1
Elastic Systems
33 In this set of formulation, we derive expressions of the virtual strain energies which are
independent of the material constitutive laws. Thus U will be left in terms of forces and
displacements.
Axial Members:
=
0
d = Adx
Victor Saouma
U = A
dx
(10.31)
Draft
107
Figure 10.3: Torsion Rotation Relations

b
h/2
M+dM
h
y1
h/2
11111
00000
11111
00000
11111
00000
11111
00000
11111
00000
11111
00000
11111
00000
dx
z
y
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
00000000
11111111
dA
max
Figure 10.4: Flexural Member

Torsional Members: With reference to Fig. 10.3
=
A
xy xy d
xy rdA
xy = r
d = dAdx
Shear Members:
U
V
xy xy d
=
A
xy dA
d = dAdx
U =
(
A
xy rdA) dx U =
T dx
(10.32)
U =
(
0
xy dA) xy dx U =
L
0
V xy dx (10.33)
Flexural Members: With reference to Fig. 10.4.

U
M
x x d
=
=
xx ydA
y =
dAdx
d =
0
Victor Saouma
M
=
y
xx dA
U =
M dx
(10.34)
Draft
108
10.1.5.1.2
34
Linear Elastic Systems
Should we have a linear elastic material ( = E) then:
Axial Members:
U
x = Ex = E du
dx
= d(u)
dx
d = Adx
U =
du d(u)
Adx
dx dx
d
(10.35)
r d
(10.36)
dx
dx
(10.37)

U =
xy xy d =
For an innitesimal element:

Tr
J
T dx
GJ
=
d =
= Gr
since the rate of change of rotation is the strain.

d = rddrdx
L
Gr
0
=
2
dx
dx
real
d(x )
rddrdx
dx
virtual
r ddr
o
U =
GJ
0
dx d(x )
dx
dx dx
(10.38)

U
x =
M =
My
Iz
d2 v
dx2 EIz
x =
x x d
d2 v
Ey
dx2
x
E
x =
=
d = dAdx
d2 (v)
dx 2 y
U =
0
d2 (v)
d2 v
Ey
ydAdx
2
dx2
A dx
(10.39)
or:
Eq. 10.39
2
Victor Saouma
y dA = Iz
U =
0
EIz
d2 v d2 (v)
dx
dx2 dx2
(10.40)
Draft

10.1.5.2
35
109
External Virtual Work W
For concentrated forces (and moments):

W =
qdx +
(i )Pi +
(10.41)
(i )Mi
where: i = virtual displacement.
10.1.6
Complementary Virtual Work
36 We dene the complementary virtual work done by the load on a body during a small,
admissible (continuous and satisfying the boundary conditions) change in displacements.
Complementary Internal Virtual Work Wi
Complementary External Virtual Work We
def
utd
(10.42)
(10.43)
Internal Complementary Virtual Strain Energy U
10.1.6.1
37
def
Again we shall consider two separate cases.
10.1.6.1.1
Arbitrary System
38 In this set of formulation, we derive expressions of the complemetary virtual strain energies
which are independent of the material constitutive laws. Thus U will be left in terms of forces
and displacements.
Axial Members:
U =
U = A
d = Adx
dx
(10.44)

U
xy xy d
xy rdA
xy = r
d = dAdx
Shear Members:
U
U =
xy dA
d = dAdx
(
A
xy rdA) dx U =
T dx
(10.45)
xy xy d
U =
(
0
xy dA) xy dx U =
L
0
V xy dx
(10.46)
Victor Saouma
Draft
1010

U =
M
x x d
y =
dAdx
M dx
(10.47)
Linear Elastic Systems
Should we have a linear elastic material ( = E) then:
Axial Members:
U =
= P
A
P
= AE
d = Adx
Torsional Members:
U =
xy
Tr
J
xy
G
xy =
xy =
d = rddrdx
r
Flexural Members:
U
r2 ddr
Mz y
x =
z
y
= Mzz
EI
d = dAdx
10.1.6.2
P
dx
AE
(10.48)
U =
T
0
T
dx
GJ
(10.49)
E dvol
U =
L
0
xy Gxy dvol
U =
40
U =
39
d =
10.1.6.1.2
M
=
y
xx ydA
xx dA
y 2 dA = Iz
U =
M
0
M
dx
EIz
(10.50)
External Complementary Virtual Work W
For concentrated forces (and moments):

W =
(i )Pi
(10.51)
(i )Mi
(10.52)
where: i = actual displacement. Or:

W =
Victor Saouma
Draft
10.2 Principle of Virtual Work and Complementary Virtual Work

41
1011
For distributed load:

W =
10.1.7
10.1.7.1
qdx +
i=1
(i )Pi +
n
i=1
(i )Mi
(10.53)
Potential Energy
Potential Functions
If during loading and unloading, U and U are independent of the path of deformation (i.e.
no intial strains), but depend only on the initial and nal states, then the dierential dU0 and
dU0 are exact dierentials and U0 and U0 are then potential functions.
42
10.1.7.2
43
Potential of External Work
The potential of external work W in an arbitrary system is dened as

def
We =
uT bd +
uT + uP
td
(10.54)
where u are the displacements, b is the body force vector; is the applied surface traction
t
vector; t is that portion of the boundary where is applied, and P are the applied nodal
t
forces.
Note that the potential of the external work is dierent from the external work itself (usually
by a factor of 1/2)
44
10.1.7.3
45
Potential Energy
The potential energy of a system is dened as
def
U We
U0 d
(10.55)
u + uP
td
ubd +
(10.56)
46 Note that in the potential the full load is always acting, and through the displacements of
its points of application it does work but loses an equivalent amount of potential, this explains
the negative sign.
10.2
Principle of Virtual Work and Complementary Virtual

Work
47 The principles of Virtual Work and Complementary Virtual Work relate force systems which
satisfy the requirements of equilibrium, and deformation systems which satisfy the requirement
of compatibility:
Victor Saouma
Draft
1012
1. In any application the force system could either be the actual set of external loads dp or
some virtual force system which happens to satisfy the condition of equilibrium p. This
set of external forces will induce internal actual forces d or internal hypothetical forces
compatible with the externally applied load.
2. Similarly the deformation could consist of either the actual joint deections du and compatible internal deformations d of the structure, or some hypothetical external and internal deformation u and which satisfy the conditions of compatibility.
48
Thus we may have 4 possible combinations, Table 10.1: where: d corresponds to the actual,
1
2
3
4
Force
External Internal
dp
d
p
dp
d
p
Deformation
External Internal
du
d
du
d
u
IVW
Formulation
U
U
CVW/Flexibility
VW/Stiness
Table 10.1: Possible Combinations of Real and Hypothetical Formulations

and (with an overbar) to the hypothetical values.
This table calls for the following observations
1. The second approach is the same one on which the method of virtual or unit load is based.
It is simpler to use than the third as a internal force distribution compatible with the
assumed virtual force can be easily obtained for statically determinate structures. This
approach will yield exact solutions for staticaally determinate structures.
2. The third approach is favoured for kinematically indeterminate problems or in conjunction
with approximate solution. It requires a proper guess of a displacement shape and is
the basis of the stifness method.
10.2.1
10.2.1.1
Principle of Virtual Work

Derivation
49 Derivation of the principle of virtual work starts with the assumption of that forces are in
equilibrium and satisfaction of the static boundary conditions.
50
The Equation of equilibrium (Eq. 9.15) which is rewritten as

xx xy
+
+ bx = 0
x
y
yy
xy
+
+ by = 0
y
x
(10.57-a)
(10.57-b)
where b representing the body force. In matrix form, this can be rewritten as
0
Victor Saouma
xx
yy

xy
bx
by
=0
(10.58)
Draft
1013
or
LT + b = 0
(10.59)
Note that this equation can be generalized to 3D.

The surface of the solid can be decomposed into two parts t and u where tractions and
displacements are respectively specied.
51
= t u
t = on t Natural B.C.
t
(10.60-b)
u = u on u Essential B.C.
(10.60-c)
(10.60-a)
Equations 10.59 and 10.60-b constitute a statically admissible stress eld.

52 We now express the local condition of equilibrium Eq. 10.59 and the static boundary condition Eq, 10.60-b in global (or integral) form. This is accomplished by multiplying both equations
by a virtual displacement u and integrating the rst equation over and the second one over
t , and we then take the sum of these two integrals (each of which must be equal to zero)
uT LT + b d +
uT (t d = 0
t)
(10.61)
B.C.
Equil.
Note that since each term is equal to zero, the negative sign is introduced to maintain later
on consistency with previous results. Furthermore, according to the fundamental lemma of the
calculus of variation (Eq. 10.198), this equation is still equivalent to Eq. 10.59 and 10.59
53
Next, we will focus our attention on
by
uT td which will be replaced (from Eq. 10.60-a)
uT td =
t
uT td
uT td
(10.62)
and which we seek to convert into a volume integral through Gauss Theorem, Eq. 4.6 and 4.7.
54
But rst let us recall the denition of the traction vector

t = .n
or ti = ij nj
(10.63-a)
applying Gauss theorem we obtain

uT td =
(uT )nd =
divuT d +
div(uT )d
(10.64-a)
uT divd
(10.64-b)
However, from Eq. 10.59 we have div = LT thus

uT td =
55
divuT d +
uT LT d
(10.65)
Combining Eq. 10.62 with the previous equation, leads to

uT td =
t
Victor Saouma
divuT d +
uT LT d
uT td
(10.66)
Draft
1014
56
We next substitute this last equation into Eq. 10.61 and reduce
uT LT d
uT bd +
divuT d +
57
uT bd +
divuT d
u td
uT LT d
u
u
uT td
uT
td = 0 (10.67-a)
t
uT
td = 0 (10.67-b)
t
The strain displacement relation can be written as

divu =
(10.68)
Substituting in the previous equation, we get
uT bd +
T d
uT td
uT = 0
td
(10.69)
Virtual displacement must be kinematically admissible, i.e. u must satisfy the essential
boundary conditions u = 0 on u , (note that the exact solution had to satisfy the natural B.C.
instead), hence the previous equation reduces to
58
T d
uT bd
Wi =Ui
uT = 0
td
(10.70)
We
Each of the preceding equations is a work expression, (Eq. 10.30). The rst one corresponds to
the internal virtual work, and the last two are expressions of the work done by the body forces
and the surface tractions through the corresponding virtual displacement u, hence
Wi = We
(10.71)
Ui = We
(10.72)
or
which is the expression of the principle of virtual work (or more specically of virtual
displacement) which can be stated as
A deformable system is in equilibrium if the sum of the external virtual work and
the internal virtual work is zero for virtual displacements u which are kinematically
admissible.
The major governing equations are summarized
T d
uT bd
Victor Saouma
(10.73)
in
(10.74)
u = 0
Wi
= Lu
uT
td
on
(10.75)
We
Draft
1015
P2
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
I=I1(1-x/2L)
x,u
y,v
Figure 10.5: Tapered Cantilivered Beam Analysed by the Vitual Displacement Method
59 Note that the principle is independent of material properties, and that the primary unknowns
are the displacements.
60
For one dimensional elements, with no initial strains (U = Wi )

d = P v
(10.76)
W
U
Example 10-1: Tapered Cantiliver Beam, Virtual Displacement

Analyse the problem shown in Fig. 10.5, by the virtual displacement method.
Solution:
For this exural problem, we must apply the expression of the virtual internal strain energy
as derived for beams in Eq. 10.40. And the solutions must be expressed in terms of the
displacements which in turn must satisfy the essential boundary conditions.
The approximate solutions proposed to this problem are
v =
v =
x
v2
2L
x 2
x
3
2
L
L
1 cos
(10.77-a)
3
v2
(10.77-b)
Note that these equations do indeed satisfy the essential B.C.

Using the virtual displacement method we evaluate the displacements v2 from three dierent
combination of virtual and actual displacement:
Solution
1
2
3
Total
Eqn. 10.77-a
Eqn. 10.77-a
Eqn. 10.77-b
Virtual
Eqn. 10.77-b
Eqn. 10.77-a
Eqn. 10.77-b
Where actual and virtual values for the two assumed displacement elds are given below.
Victor Saouma
Draft
1016

Trigonometric (Eqn. 10.77-a)
x
1 cos 2L v2
Polynomial (Eqn. 10.77-b)

x 2
x 3
3 L 2 L
v2
x
1 cos 2L v2
2
4L2
2
4L2
x 2
L
x 3
L
6
12x
L2
L3
6
12x
L3
L2
x
cos 2L v2
cos
x
2L v2
v2
v2
v2
Note that both Eqn. 10.77-a and Eqn. 10.77-b satisfy the essential (geometric) B.C.
L
=
0
v EIz v dx
= P2 v2
(10.78-a)
(10.78-b)
Solution 1:
12x
x
2
x
6
cos
v2
3 v2 EI1 1
dx
4L2
2L
L2
L
2L
=
0
3EI1
10 16
1
+ 2 v2 v2
3
2L
= P2 v2
=
(10.79-a)
which yields:
v2 =
P2 L3
2.648EI1
(10.80)
Solution 2:
L
=
0
x
4
x
cos2
v2 v2 EI1 1
dx
16L4
2L
2L
4 EI1
32L3
= P2 v2
=
3
1
v2 v2
+
4 2
(10.81-a)
which yields:
v2 =
P2 L3
2.57EI1
(10.82)
Solution 3:
L
=
0
12x
6
3
2
L
L
9EI
v2 v2
=
L3
= P2 v2
x
EI1 v2 v2 dx
2L
(10.83-a)
which yields:
v2 =
Victor Saouma
P2 L3
9EI
(10.84)
Draft
10.2.2
10.2.2.1
1017
Principle of Complementary Virtual Work

Derivation
61 Derivation of the principle of complementary virtual work starts from the assumption of a
kinematicaly admissible displacements and satisfaction of the essential boundary conditions.
62 Whereas we have previously used the vector notation for the principle of virtual work, we
will now use the indicial notation for this derivation.
63
The kinematic condition (strain-displacement) was given in Eq. 9.9.

1
(ui,j + uj,i )
2
ij =
64
(10.85)
The essential boundary conditions are expressed as
ui = u on u
(10.86)
Those two equations are rewritten as

ij ui,j
= 0
(10.87-a)
ui u = 0
(10.87-b)
We premultiply the rst equation by a virtual stress eld ij and integrate over the volume;
and we premultiply the second by corresponding virtual tractions ti and integrate over the
corresponding surface
(ij ui,j ) ij d
(ui u) ti d = 0
(10.88)
Note that since each term is equal to zero, the negative sign is introduced to maintain later
on consistency with previous results. Furthermore, according to the fundamental lemma of the
calculus of variation (Eq. 10.198), this equation is still equivalent to the kinematic conditions
10.85 and 10.86.
65 Since the arbitrary stresses must be statically admissible, it follows that they must satisfy
the equation of equilibrium
66
Next, we will focus our attention on
by
u
ui ti d which will be replaced (from Eq. 10.60-a)
ui ti d =
ui ti d
ui ti d
(10.89)
and note that the second term on the right hand side is zero since
ti = 0
on t
(10.90)
in order to satisfy the boundary conditions.

67 We now seek seek to convert the previous expression into a volume integral through Gauss
Theorem, Eq. 4.6 and 4.7.
ui ti d =
Victor Saouma
ui ti d =
ui ij,j d +
ui (ij nj )d
(10.91-a)
ui,j ij d
(10.91-b)
Draft
1018
However,the virtual stresses must be in equilibrium within , thus from Eq. 9.15, and in the
absence of body forces
ij,j = 0 in
(10.92)
thus
u
68
ui ti d =
ui,j ij d
(10.93)
Combinig this last equation with Eq. 10.88 leads to
ij ij d
ui,j ij d
uti d +
ui,j ij d = 0
(10.94)
which simplies into
ij ij d
uti d = 0
(10.95)
69 We note that each of the preceding term is a work expression, and that the rst one corresponds to the internal complementary virtualwork, and the scond to the external complementary
virtual work, Eq. 10.43
Wi We = 0
(10.96)
which is the expression of the principle of virtual complementary work (or more specically of
virtual force) which can be stated as
A deformable system satises all kinematical requirements if the sum of the external
complementary virtual work and the internal complementary virtual work is zero
for all statically admissible virtual stresses ij .
The major governing equations are summarized
ij ij d
(10.97)
ij,j = 0
in
(10.98)
ti = 0
on
(10.99)
Wi
ui ti d
We
Note that the principle is independent of material properties, and that the primary unknowns
are the stresses.
70
71
The principle of virtual forces leads to the exibility matrix.
Example 10-2: Tapered Cantilivered Beam; Virtual Force

Exact solution of previous problem using principle of virtual work with virtual force.
L
M
0
M
dx =
EIz
Internal
Victor Saouma
(10.100)
External
Draft
1019
P2
I=I1(1-x/2L)
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
00
11
x,u
y,v
Figure 10.6: Tapered Cantilevered Beam Analysed by the Virtual Force Method
Note: This represents the internal virtual strain energy and external virtual work written
in terms of forces and should be compared with the similar expression derived in Eq. 10.40
written in terms of displacements:
U =
L
0
EIz
d2 v d2 (v)
dx
dx2 dx2
(10.101)
M
Here: M and P are the virtual forces, and EIz and are the actual displacements. See
Fig. 10.6 If P = 1, then M = x and M = P2 x or:
L
(1) =
x
0
P2 x
x dx
EI1 (.5 + L )
P2
EI1
P2 2L
EI1
x2
dx
L+x
0 2L
L x2
0
L+x
dx
(10.102-a)
From Mathematica we note that:

1 1
x2
= 3
(a + bx)2 2a(a + bx) + a2 ln(a + bx)
b 2
0 a + bx
0
(10.103)
Thus substituting a = L and b = 1 into Eqn. 10.103, we obtain:

=
=
=
=
Victor Saouma
2P2 L
EI1
2P2 L
EI1
2P2 L
EI1
1
(L + x)2 2L(L + x) + L2 ln(L + x) |L
0
2
L2
+ 2L2 + L2 log L
2L2 4L2 + L2 ln 2L
2
1
L2 (ln 2 )
2
P2 L3
2.5887EI1
(10.104-a)
Draft
1020
This exact value should be compared with the approximate one obtained with the Virtual
P L3
Displacement method in which a displacement eld was assumed in Eq. 10.174 of 2.55EI1 .
Similarly:
L
M (1)
0 EI1 .5 +
x
L
2M L L 1
EI1 0 L + x
2M L
ln(L + x) |L
0
EI1
2M L
(ln 2L ln L)
EI1
2M L
ln 2
EI1
=
=
=
=
ML
.721EI1
(10.105-a)
Example 10-3: Three Hinged Semi-Circular Arch

We seek to determine the vertical deection of the crown of the three hinged statically
determined semi-circular arch under its own dead weight w. Fig. 10.7 We rst seek to determine
the analytical expression of the moment diagram. From statics, it can be shown that the vertical
and horizontal reactions are Rv = wR and Rh = 1 wR.
2
2
Next considering the free body diagram of the arch, and summing the forces in the radial
direction (FR = 0):
1 wR cos + wR sin
2
V = wR
wRd sin + V = 0
(10.106-a)
=0
1 cos +
sin
(10.106-b)
Similarly, if we consider the summation of forces in the axial direction (FT = 0):
1 wR sin + wR cos
2
N = wR
cos
wRd cos + N = 0
=0

2
(10.107-a)
1 sin
(10.107-b)
Now we can consider the third equation of equilibrium (M = 0):
1 wR R sin wR2 (1 cos ) +

2
=0
wRd R(cos cos ) + M = 0
(10.108-a)
sin ) +
(10.108-b)
M = wR2
2 (1
cos
The real curvature is obtained by deviding the moment by EI

=
Victor Saouma
wR2
cos
(1 sin ) +
EI 2
2
(10.109)
Draft
1021
dP=wRd
R sin
R cos
C =( /2-1)wR
x
R(1-cos )
R cos
C y /2 wR
=
R cos( )
Figure 10.7: Three Hinge Semi-Circular Arch

The virtual force P will be a unit vertical point in the direction of the desired deection,
causing a virtual internal moment
M =
R
[1 cos sin ]
2
(10.110)
Hence, application of the virtual work equation yields:

1 = 2
P
=0
R
wR2
cos [1 cos sin ] Rd
(1 sin ) +
EI 2
2
2
dx
M
=
=
wR4
16EI
7 2 18 12
4
.0337 wR
EI
(10.111-a)
Example 10-4: Cantilivered Semi-Circular Bow Girder

Considering the semi-circular cantilevered box girder shown in Fig. 10.8 subjected to its
own weight w, and with a rectangular cross-section of width b and height d = 2b and with
Poissons ratio = 0.3.
Victor Saouma
Draft
1022

x
x
wRd
d
V
r
C
A
A
y
Figure 10.8: Semi-Circular Cantilevered Box Girder

First, we determine the internal forces by applying the three applicable equations of equilibrium:
FZ = 0
wRd = 0
MR = 0 M
MT = 0 +
V = wr
M = wR2 (1 cos )
(wRd)(R sin ) = 0
0
(10.112)
(wRd)R(1 cos ) = 0 T = wR2 ( sin )
Noting that the member will be subjected to both exural and torsional deformations, we
seek to determine the two stinesses.
3
3
4
The exural stiness EI is given by EI = E bd = E b(2b) = 2Eb = .667Eb4 .
12
12
3
The torsional stiness of solid rectangular sections J = kb3 d where b is the shorter side of
E
E
the section, d the longer, and k a factor equal to .229 for d = 2. Hence G = 2(1+) = 2(1+.3) =
b
.385E, and GJ = (.385E)(.229b4 ) = .176Eb4 .
Considering both exural and torsional deformations, and replacing dx by rd:
P =
W
M
0
M
Rd +
EIz
T
0
T
Rd
GJ
(10.113)
torsional
exural
U
where the real moments were given above.

Assuming a unit virtual downward force P = 1, we have
M
T
= R sin
(10.114-a)
= R(1 cos )
(10.114-b)
Substituting these expression into Eq. 10.113

1 =
P
Victor Saouma
wR2
EI
(R sin ) (1 cos ) Rd +
M
wR2
GJ
( sin ) R(1 cos ) Rd

T
Draft
10.3 Potential Energy
=
=
1023
wR4
1
( cos sin + sin cos ) d
(sin sin cos ) +
EI 0
.265
wR4
( 2. + 18.56 )
EI
exure
torsion
4
20.56 wR
EI
(10.115-a)
10.3
Potential Energy
10.3.1
Derivation
72
From section 10.1.7.1, if U0 is a potential function, we take its dierential

U0
dij
ij
U0
dU0 =
dij
ij
dU0 =
73
(10.116-a)
(10.116-b)
However, from Eq. 10.4

U0 =
ij
0
ij dij
(10.117-a)
dU0 = ij dij
(10.117-b)
thus,
U0
ij
U0
ij
74
= ij
(10.118)
= ij
(10.119)
We now dene the variation of the strain energy density at a point1

U0 =
U
ij = ij ij
ij
The principle of virtual work, Eq. 10.73, ij ij d ui bi d

rewritten as
U0 d
ui bi d
ui ti d = 0
75
(10.120)
ui ti d = 0 can be
(10.121)
Note that the variation of strain energy density is, U0 = ij ij , and the variation of the strain energy itself
is U = U0 d.
Victor Saouma
Draft
1024
k= 500 lbf/in
111
000
mg= 100 lbf
Figure 10.9: Single DOF Example for Potential Energy

If nor the surface tractions, nor the body forces alter their magnitudes or directions during
deformation, the previous equation can be rewritten as
76
U0 d
ui bi d

ui ti d = 0
(10.122)
77
Hence, comparing this last equation, with Eq. 10.56 we obtain
(10.123)
def
U We
(10.124)
78
U0 d
u + uP
td
ubd +
(10.125)
We have thus derived the principle of stationary value of the potential energy:
Of all kinematically admissible deformations (displacements satisfying the essential
boundary conditions), the actual deformations (those which correspond to stresses
which satisfy equilibrium) are the ones for which the total potential energy assumes
a stationalry value.
79
For problems involving multiple degrees of freedom, it results from calculus that
=
1 +
2 + . . . +
n
1
2
n
(10.126)
which results in n equations with n unknowns.

80 It can be shown that the minimum potential energy yields a lower bound prediction of
displacements.
As an illustrative example, let us consider the single dof system shown in Fig. 10.9. The
strain energy U and potential of the external work W are given by
81
U
We
Victor Saouma
1
u(Ku) = 250u2
2
= mgu = 100u
(10.127-a)
(10.127-b)
Draft
1025
Potential Energy of Single DOF Structure
Total Potential Energy

Strain Energy
External Work
Energy [lbfin]
20.0
0.0
20.0
40.0
0.00
0.10
0.20
Displacement [in]
0.30
Figure 10.10: Graphical Representation of the Potential Energy

Thus the total potential energy is given by
= 250u2 100u
(10.128)
and will be stationary for

d
= 0 500u 100 = 0 u = 0.2 in
du
(10.129)
U = 250(0.2)2 = 10 lbf-in
W = 100(0.2) = 20 lbf-in
= 10 20
= 10 lbf-in
(10.130)
Substituting, this would yield
Fig. 10.10 illustrates the two components of the potential energy.
10.3.2
Euler Equations of the Potential Energy
A variational statement is obtained by taking the rst variation of the variational principle
and setting this scalar quantity equal to zero.
82
83 The variational statement for the general form of the potential energy functional (i.e. Equation 10.125) is
D d
D 0 d +
0 d
uT bd
uT = 0 (10.131)
td
t
which is the Principle of Virtual Work.
Victor Saouma
Draft
1026
84 Since the dierential operator L is linear, the variation of the strains

can be expressed in
terms of the variation of the displacements u
= (Lu) = Lu
(10.132)
This relationship is exploited to obtain a form of the variational statement in which only
variations of the displacements u are present
=
(Lu)T D d
(Lu)T D 0 d +
(Lu)T 0 d
uT bd
uT = 0
td
t
(10.133)
which is best suited for obtaining the corresponding Euler equations.

A form of the variational statement in which strain-displacement relationship (i.e. Equation
??) is substituted into Equation 10.133
85
(Lu)T D(Lu)d
uT bd
(Lu)T D 0 d +
(Lu)T 0 d
uT = 0
td
(10.134)
is better suited for obtaining the discrete system of equations.

To obtain the Euler equations for the general form of the potential energy variational principle the volume integrals dening the virtual strain energy U in Equation 10.133 must be
integrated by parts in order to convert the variation of the strains (Lu) into a variation of the
displacements u.
86
87
Integration by parts of these integrals using Greens theorem (Kreyszig 1988) yields
(Lu)T D d =
(Lu)T D 0 d =
(Lu)T 0 d =
uT G(D )d
uT LT (D )d
uT G(D 0 )d
uT G 0 d
uT LT (D 0 )d
(10.135)
u LT 0 d
where G is a transformation matrix containing the direction cosines for a unit normal vector
such that the surface tractions t are dened as t = G and the surface integrals are over the
entire surface of the body .
88
Substituting Equation 10.135 into Equation 10.133, the variational statement becomes
=
uT {LT [D(
uT {G[D(
0)
+ 0 ] + b}d
0)
+ 0] +
t}d = 0
(10.136)
89 Since u is arbitrary the expressions in the integrands within the braces must both be equal
to zero for to be equal to zero. Recognizing that the stress-strain relationship (i.e. Equation
10.8) appears in both the volume and surface integrals, the Euler equations are
LT + b = 0 on
G = 0 on t
t
Victor Saouma
(10.137)
Draft
1027
where the rst Euler equation is the equilibrium equation and the second Euler equation denes
the natural boundary conditions. The natural boundary conditions are dened on t rather
than because both the applied surface tractions and the matrix-vector product G are
t
identically zero outside t .
Starting from the Euler equations, it is possible to derive the total potential energy functional
by performing the operations just presented in reverse order.
90
91 Substituting Equations 10.10 and 10.54 into the expression for the total potential energy the
functional for the general form of the potential energy variational principle is obtained
1
2
D d
D 0 d +
0 d
uT bd
uT
td
(10.138)
Thus, it is important to notice that we have now gone full circle. We began with equation of
equilibrium, derived the principle of virtual work, then the principle of minimum total potential
energy. Finally, we just proved that the Euler equations of the Potential Energy are the very
same which were used to derive the principle of virtual work.
92
10.3.3
93
Castiglianos First Theorem
A global version of Eq. 10.119
U0
ij
= ij , is Castiglianos theorem.
94 Since we are now considering a general structure, we consider an arbitrary three dimensional

structure subjected to a set of external forces (or moments) P1 , P2 , , Pn with corresponding

unknown displacements 1 , 2 , , n . The total potential energy is given by
Pi i
= Wi + We = U +
(10.139)
i=1
The strain energy can also be expressed in terms of the displacements i thus the potential
energy will be dened in terms of generalized coordinates or generalized displacements.
95
96
For the solid to be in equilibrium, = 0 or

U
U
U
1
2
n
1
2
n
+P1 1 + P2 2 + Pn n = 0
(10.140-a)
or
U
U
U
+ P1 1 +
+ P2 2 + +
+ Pn n = 0
1
2
n
(10.141)
but since the variation i is arbitrary, then each factor within the parenthesis must be equal
to zero. Thus
U
(10.142)
= Pk
k
which is Castiglianos rst theorem:
Victor Saouma
Draft
1028
If the strain energy of a body is expressed in terms of displacement components in

the direction of the prescribed forces, then the rst partial derivative of the strain
energy with respect to a displacement, is equal to the corresponding force.
Example 10-5: Fixed End Beam, Variable I

Considering the beam shown in Fig. 10.11, we can assume the following solution:
1
0
1
0
1
0
1
0
1
0
EI/2
1
0
1
0
1
0
1
0
1
0
EI
L/4
L/4
L/4
L/4
Figure 10.11: Variable Cross Section Fixed Beam

v = a1 x3 + a2 x2 + a3 x + a4
(10.143)
1. First, this solution must satisfy the essential B.C.: v = v = 0 at x = 0; and v = vmax and
v = 0 at x = L . This will be enforced by determining the four parameters in terms of a
2
single unknown quantity (4 equations and 4 B.C.s):
@x = 0
v=0
a4 = 0
@x = 0
dv
dx = 0
a3 = 0
@x =
L
2
dv
3 a1 L2 + a2 L = 0
4
dx = 0
upon substitution, we obtain:
@x =
(10.144)
v = vmax vmax = a1 L + a2 L
8
4
L
2
v = 16x +
L3
12x2
L2
a2 = 3 a1 L
4
vmax
(10.145)
Hence, in this problem the solution is in terms of only one unknown variable vmax .
2. In order to apply the principle of Minimum Potential Energy we should evaluate:
M2
dx (Eq. 10.14);
2EIz
from above:
Internal Strain Energy U : for exural members is given by U =

recalling that
M
EIz
d
d
= dxv , thus we must evaluate dxv
2
2
dv
dx
Victor Saouma
48x2 24x
+ 2
L3
L
vmax
(10.146-a)
Draft
1029
d2 v
dx2
which yields
24
4x
1
vmax
L2
L
U = 2
1
2
d2 v
dx2
(10.146-b)
Iz dx
(10.147)
or:
U
2
+
U
L
4
E
2
E
2
4x 2 2 Iz
242
1
vmax dx
L4
L
2
L
2
2
2 24
4x
2
1
vmax Iz dx
L L4
L
4
72EIz 2
vmax
L3
(10.148-a)
Potential of the External Work W: For a point load, W = P vmax

3. Finally,
= 0
vmax
W
U
= 0
vmax vmax
144EIz
vmax = P
L3
P L3
144EIz
vmax =
(10.149-a)
(10.149-b)
(10.149-c)
(10.149-d)
4. Note, that had we applied Castiglianos theorem, then

U
vmax
144EIz
vmax
L3
def
(10.150-a)
(10.150-b)
vmax
P L3
144EIz
(10.150-c)
which is identical to the solution obrained through the principle of minimum potential
energy.
10.3.4
Rayleigh-Ritz Method
97 Continous systems have innite number of degrees of freedom, those are the displacements
at every point within the structure. Their behavior can be described by the Euler Equation, or
the partial dierential equation of equilibrium. However, only the simplest problems have an
exact solution which (satises equilibrium, and the boundary conditions).
Victor Saouma
Draft
1030
98 An approximate method of solution is the Rayleigh-Ritz method which is based on the

principle of virtual displacements. In this method we approximate the displacement eld by a
function
u1
c1 1 + 1
i i
0
(10.151-a)
c2 2 + 2
i i
0
(10.151-b)
c3 3 + 3
i i
0
(10.151-c)
i=1
n
u2
i=1
n
u3
i=1
where cj denote undetermined parameters, and are appropriate functions of positions.

i
99
should satisfy three conditions

1. Be continous.
2. Must be admissible, i.e. satisfy the essential boundary conditions (the natural boundary
conditions are included already in the variational statement. However, if also satisfy
them, then better results are achieved).
3. Must be independent and complete (which means that the exact displacement and their
derivatives that appear in can be arbitrary matched if enough terms are used. Furthermore, lowest order terms must also beincluded).
In general is a polynomial or trigonometric function.

We determine the parameters cj by requiring that the principle of virtual work for arbitrary
i
variations cj . or
i
100
(u1 , u2 , u3 ) =
i=1
1 2 3
c + 2 ci + 3 ci
c1 i
ci
ci
i
=0
(10.152)
for arbitrary and independent variations of c1 , c2 , and c3 , thus it follows that

i
i
i
cj
i
=0
i = 1, 2, , n; j = 1, 2, 3
(10.153)
Thus we obtain a total of 3n linearly independent simultaneous equations. From these displacements, we can then determine strains and stresses (or internal forces). Hence we have replaced
a problem with an innite number of d.o.f by one with a nite number.
101
Some general observations

1. cj can either be a set of coecients with no physical meanings, or variables associated
i
with nodal generalized displacements (such as deection or displacement).
2. If the coordinate functions satisfy the above requirements, then the solution converges
to the exact one if n increases.
3. For increasing values of n, the previously computed coecients remain unchanged.
Victor Saouma
Draft
1031
w
x
111
000
111
000
L
Figure 10.12: Uniformly Loaded Simply Supported Beam Analysed by the Rayleigh-Ritz
Method
4. Since the strains are computed from the approximate displacements, strains and stresses
are generally less accurate than the displacements.
5. The equilibrium equations of the problem are satised only in the energy sense = 0
and not in the dierential equation sense (i.e. in the weak form but not in the strong one).
Therefore the displacements obtained from the approximation generaly do not satisfy the
equations of equilibrium.
6. Since the continuous system is approximated by a nite number of coordinates (or d.o.f.),
then the approximate system is stier than the actual one, and the displacements obtained
from the Ritz method converge to the exact ones from below.
Example 10-6: Uniformly Loaded Simply Supported Beam; Polynomial Approximation
For the uniformly loaded beam shown in Fig. 10.12

let us assume a solution given by the following innite series:
v = a1 x(L x) + a2 x2 (L x)2 + . . .
(10.154)
for this particular solution, let us retain only the rst term:
v = a1 x(L x)
(10.155)
We observe that:
1. Contrarily to the previous example problem the geometric B.C. are immediately satised
at both x = 0 and x = L.
2. We can keep v in terms of a1 and take a1 = 0 (If we had left v in terms of a1 and a2 we
should then take both a1 = 0, and a2 = 0 ).
3. Or we can solve for a1 in terms of vmax (@x =
Victor Saouma
L
2)
and take
vmax
= 0.
Draft
1032
=U W =
Recalling that:
M
EIz
M2
dx
2EIz
wv(x)dx
(10.156)
d
= dxv , the above simplies to:
2
=
0
d2 v
dx2
EI
z
2
wv(x) dx
(10.157-a)
EIz
(2a1 )2 a1 wx(L x) dx
2
0
L3
L3
EIz 2
4a1 L a1 w
+ a1 w
=
2
2
3
a1 wL3
= 2a2 EIz L
1
6
=
If we now take
a1
(10.157-b)
= 0, we would obtain:
4a1 EIz l
wL3
6
= 0
a1 =
wL2
24EIz
(10.158-a)
Having solved the displacement eld in terms of a1 , we now determine vmax at
v =
wL4
24EIz
L
2:
x
x2
2
L L
a1
wL4
96EIz
(10.159-a)
4
5
wL
exact
This is to be compared with the exact value of vmax = 384 wLz = 76.8EIz which constitutes
EI
17% error.
wL2
w
Note: If two terms were retained, then we would have obtained: a1 = 24EIz & a2 = 24EIz
exact
and vmax would be equal to vmax . (Why?)
Example 10-7: Uniformly Loaded Simply Supported Beam; Fourrier Series

Let us consider again the problem of Fi. 10.12 but with a trigonometric series for the
continuous displacement eld:
nx
an sin
(10.160)
v=
n=1
L
we note that the B.C. are satised (v = 0 at x = 0 and x = L). The potential energy is given
by:
= U W
Victor Saouma
Draft
1033
=
0
d2 v
EIz
2
dx2
wv(x) dx
n2 2 an
L2
EIz
2
sin
nx
L
wan sin
nx
dx
L
= ...
1
n 4
L
nx L
EIz L 2
an cos
|
a2
+w
=
n
n=1
n=1 n
2
2
L
L 0
an
4 EIz
2wL
=
a2 n4
3
n=1 n
n=1,3,5 n
4L
Note that for n even, the second term vanishes.

We now take:
=0
= 0 ...
=0
a1
a2
an
which would yield:
4wL4
n =1,3,5
an =
EIz (n)5
or:
nx
1 5
4wL4
sin
v=
5
n=1,3,5 n
EIz
L
and for x =
L
2
(10.161-a)
(10.162)
(10.163)
(10.164)
we would get:
v = vmax =
4wL4
EIz 5
1
35
1
55
1
75
+ ...
(10.165)
Note that should we consider only the 1st term, then:

vmax =
wL4
exact
vmax
76.5EIz
(10.166)
Example 10-8: Tapered Beam; Fourrier Series

Revisiting the previous problem of a tapered beam subjected to a point load, Fig. 10.13 and
using the following approximation
v=
n=1,3,...
an 1 cos
nx
2L
(10.167)
we seek to solve for v2 and 2 , for n = 1 and n = 3.

Solution:
=
=
Victor Saouma
n
2L
an
1
2
L
0
EI1
2
cos
nx
2L
(10.168-a)
(v )2 EIz dx
(10.168-b)
L
0
n=1,3,
an
n
2L
cos
nx
2L
x
dx
L
(10.168-c)
Draft
1034
P2
I=I1(1-x/2L)
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
11
00
00
11
x,u
y,v
Figure 10.13: Example xx: External Virtual Work
However, we recall that:

L
0
L
x cos
0
0
nx
mx
cos
dx =
cos
L
2L
2L
m=n
(10.169-a)
m=n
2
0
nx
mx
cos
dx =
2L
2L
m=n
L2
8
L2
2n2 2
(10.169-b)
m=n
Thus combining Eqns. 10.168-c, 10.169-a, and 10.169-b, we obtain:

U=
4 EI1
64L3
1,3,5
1
3
+ 2 2 n4 a2
n
4 n
(10.170)
The potential of the external work W in turn is given by:

W = P2
an +
M2
L
(1)
v@x=l
n1
2
nan
(10.171)
@x=l
Finally, taking
U
W
=0
an
an an
Combining Eqns. 10.170, 10.171, and 10.172 we solve for an :
(10.172)
n1
n
32L3 P + 2L (1) 2 M
an = 4
3
1
4
EI1
4 + n2 2 n
Solving for v2 =
an we obtain:
v2 =
Victor Saouma
(10.173)
P L3
2.59EI1
P L3
2.55EI1
+
+
M L2
1.65EI1
M L2
1.65EI1
n=1
n=3
(10.174)
Draft
10.4 Complementary Potential Energy

Similarly we solve for 2 =
2L
n(1)
n1
2
1035
an
P L2
Ml
1.65EI + 1.05EI1
2
PL
ML
1.65EIz + 1.04EI1
n=1
n=3
10.4
Complementary Potential Energy
10.4.1
(10.175)
Derivation
102
Eq. 10.97,
ij ij d
ui ti d = 0 can be rewritten as
U0 d
ui ti d = 0
(10.176)
or
(10.177)
def
U We
(10.178)
which is the principle of stationary complementary energy which states that

Of all statically admissible states of stress (stresses satisfying the equation of equilibrium), the actual state of stress (the one which satisfy the kinematic conditions) are
the ones for which the total complementary potential energy assumes a stationalry
value.
10.4.2
Castiglianos Second Theorem
103 Considering again a three dimensional structure subjected to external displacements (or

rotations) (or moments) 1 , 2 , , n with corresponding unknown forces (or moments)

P1 , P2 , , Pn . The total complementary potential energy is given by
= Wi + We = Ui
i Pi
(10.179)
i=1
The complementary strain energy can also be expressed in terms of the forces Pi thus the
complementary potential energy will be dened in terms of generalized coordinates or generalized
forces.
104
105
For the solid to be in equilibrium, = 0 or

=
1 P1 2 P2 n Pn = 0
Victor Saouma
Wi
Wi
Wi
P1 +
P2 + +
Pn
P1
P2
Pn
(10.180-a)
Draft
1036
or
Wi
1 P1 +
P1
Wi
2 P2 + +
P2
Wi
n Pn
Pn
(10.181)
but since the variation Pi is arbitrary, then each factor within the parenthesis must be equal
to zero. Thus
Wi
(10.182)
= k
Pk
which is Castiglianos second theorem:
If the complementary strain energy of a body is expressed in terms of forces then
the rst partial derivative of the strain energy with respect to any one of the forces,
is equal to the corresponding displacement at the point where the force is located.
Example 10-9: Cantilivered beam

Solve for the displacement of the tip of a cantiliver loaded by a point load.
Solution:
From Eq. 10.14, the complementary strain energy is U =
the external work is W e
1
2
L M2
dx, and for a point load,

EI
= P thus the potential energy of the system is
= W e U
= P
1
2
M2
dx
0 EIz
L
However, for the point load, the moment is M = P x, substituting above

L P 2 x2
1
dx
2
0 EI
2 L3
P 1 PEI
6
2 L3
1 PEI = 0
3
1 P 2 L3
3 EI
= P
=
d =
dP
=
10.4.2.1
Distributed Loads
Castiglianos theorem can easily be applied to problems in which the structure is subjected
to point load or moments, and we seek the deection under these loads.
106
107 However when a structure is subjected to say a uniform load, and we wish to determine
the deection at a point where no point load is applied, then we must introduce a ctitious
corresponding force R and then write the complementary strain energy in terms of R and the
applied load, and ten set R = 0.
Victor Saouma
Draft
10.5 Comparison of Alternate Approximate Solutions
1037
Example 10-10: Deection of a Uniformly loaded Beam using Castiglianos second Theorem
Considering a simply supported uniformly loaded beam, we seek the midspan deection.
Solution:
We introduce a ctitious force R at midspan, and the moment is thus M (x) =
w x . The complementary strain energy is U = 2
2
2
=
=
=
10.5
U
R R=0
L
2
M
2
EI 0 M R dx
=
R=0
2
EI
L
2
M (x)
2EI dx
L
2
wL
2
wL
R
2 x+ 2x
and the displacement
R
2
x wx
2
x
2 dx
R=0
5wL4
384EI
Comparison of Alternate Approximate Solutions
108 While we were able to assess the accuracy of our approximate solutions with respect to the
exact one, (already known), in general this is not possible. (i.e., If an exact solution is known,
there is no need for an approximate one). Thus the question is, given two or more alternate
approximate solutions which one is the best?
This can be determined by evaluating the potential energy of each approximate solution
and identify the lowest one.
109
Example 10-11: Comparison of MPE Solutions

With reference to examples (simply supported uniformly loaded beams) we can determine
for each one its Potential Energy = U We :
Polynomial Solution: From Eq. 10.158-a and 10.157-b respectively, we had:
a1 =
wL2
24EIz
= 2a2 EIz L
1
wL2
= 2
24EIz
(10.183-a)
a1 wL3
6
2
EIz l
w2 L5
1
2
2
EIz 24
(6)(24)
2 L5
1 w
=
288 EIz
wL2
24EIz
wL3
6
Victor Saouma
(10.183-b)
Draft
1038
Trigonometric Solution: From Eq. 10.161-a and 10.163 respectively we had:

=
an =
2wL
4 EIz
a2 n4
3
n=1 n
4L
4wL4
n = 1, 3, 5
EIz (n)5
n=1,2,3
an
n
(10.184-a)
(10.184-b)
For n = 1:
a1 =
4wL4
EIz 5
4 EIz 4wL4
4L3
EIz 5
(10.185-a)
2
2wL 4wL4
EIz 5
w2 L5 16
8
EIz 4 6 6
4 w2 L5
= 6
EIz
1 w2 L5
=
240 EIz
=
(10.185-b)
We note that the Trigonometric solution has a lower potential energy than the polynomial
wL4
approximation and is thus more accurate (the exact displacement is vmax = 76.8EIz ) as shown
in Table 10.2.
Polynomial
Trigonometric
EIz
w2 L5
1
288
1
240
EI
vmax wLz
4
1
96
1
76.6
% error
17%
1%
Table 10.2: Comparison of 2 Alternative Approximate Solutions
10.6
110
Summary
A summary of the various methods introduced in this chapter is shown in Fig. 10.14.
The duality between the two variational principles is highlighted by Fig. 10.15, where
beginning with kinematically admissible displacements, the principle of virtual work provides
statically admissible solutions. Similarly, for statically admissible stresses, the principle of
complementary virtual work leads to kinematically admissible solutions.
111
Finally, Table 10.3 summarizes some of the major equations associated with one dimensional
rod elements.
112
Victor Saouma
Draft
10.6 Summary
1039
Displacement Based Formulation [k]

Natural B.C.
Essential B.C.
?
div + b = 0
t t = 0 t
def
U0 =
Du = 0
u = 0 u
Gauss
ij,j = 0
ti = 0 t
Gauss
T d
ij (ui,j + uj,i ) = 0
- u i u = 0 u
def

?
1
2
U0 =
Force Based Formulation [d]
t uT td = 0
Wi We = 0
u
T bd
Principle of Complementary
Virtual Work
ij ij d u ui ti d = 0
Wi We = 0
Principle of Stationary
Potential Energy
= 0
def
= U We
= U0 d ( ui bi d + t ui ti d)
?
Stationary Potential Energy
= 0
def
= Wi + We
=
U0 d + u ui ti d
?
Castiglianos
First Theorem
Castiglianos
Second Theorem
Wi
= Pk
k
Wi
= k
Pk
Rayleigh-Ritz
cj j + j
0
i i
uj
cj
i
i=1
=0
i = 1, 2, , n;
j = 1, 2, 3
Figure 10.14: Summary of Variational Methods
Victor Saouma
Draft
1040
Kinematically Admissible Displacements

Displacements satisfy the kinematic equations
and the the kinematic boundary conditions
6
Complementary Energy
Virtual Work
Potential Energy
Statically Admissible Stresses

Stresses satisfy the equilibrium conditions
and the static boundary conditions
Figure 10.15: Duality of Variational Principles
Victor Saouma
Draft
10.7 Variational Calculus; Preliminaries

U
Axial
P2
dx
0 AE
L
1
2
1041
Virtual Displacement U
uT LT + b d
+ t uT (t
t)d = 0
General
Linear
L
L
du d(u)
dx
E
Adx
dx dx
0
0
d
Shear
Flexure
...
M2
dx
0 EIz
L
1
2
V xy dx
...
0
L
M dx
0
EIz
Torsion
T2
dx
0 GJ
L
T dx
0
d2 v d2 (v)
dx
dx2 dx2
W
1
i 2 Pi i
1
i 2 Mi i
P
M
Virtual Displacement W
i Pi i
i Mi i
w(x)v(x)dx
0
L
...
L
M dx
0
M
0
T dx
0
M
dx
EIz
T
T
dx
GJ
Virtual Force W
i Pi i
i Mi i
L
w(x)v(x)dx
dx d(x )
GJ
dx
dx dx
V xy dx
1
2
Virtual Force U
(ij ui,j ) ij d
u (ui u) ti d = 0
General
Linear
L
L
P
dx
P
dx
AE
0
0
w(x)v(x)dx
0
Table 10.3: Summary of Variational Terms Associated with One Dimensional Elements
10.7
Variational Calculus; Preliminaries
113 The analysis of structures so far, has been achieved through the solution of specic boundaryvalue problems expressed in terms of partial dierential equations.
A dierent approach, exploiting minimum principles that characterize the equilibrium states
of elastic bodies, is developed in this chapter. It is an approach based on the use of direct
methods in the calculus of variations rst proposed by Lord Rayleigh and W. Ritz and extended
by R. Courant, B. Galerkin and others.
114
115 Broadly speaking, previous methods can be labelled as Newtonian, whereas methods based
on energy considerations (as will be the case in this chapter) are labelled as Lagrangian.
116
This chapter borrows heavily from [?] and [?].
10.7.1
117
Euler Equation
The fundamental problem of the calculus of variation2 is to nd a function u(x) such that
b
F (x, u, u )dx
(u) =
(10.186)
2
Dierential calculus involves a function of one or more variable, whereas variational calculus involves a
function of a function, or a functional.
Victor Saouma
Draft
1042

u, u
~
u(x)
B
u(x)
du
dx
x
x=a
x=c
x=b
Figure 10.16: Variational and Dierential Operators

is stationary. Or,
(10.187)
= 0
where indicates the variation
We dene u(x) to be a function of x in the interval (a, b), and F to be a known real function
(such as the energy density).
118
We dene the domain of a functional as the collection of admissible functions belonging to

a class of functions in function space rather than a region in coordinate space (as is the case
for a function).
119
120
We seek the function u(x) which extremizes .
Letting u to be a family of neighbouring paths of the extremizing function u(x) and we
assume that at the end points x = a, b they coincide. We dene u as the sum of the extremizing
path and some arbitrary variation, Fig. 10.16.
121
u(x, ) = u(x) + (x) = u(x) + u(x)
(10.188)
where is a small parameter, and u(x) is the variation of u(x)

u = u(x, ) u(x)
= (x)
(10.189-a)
(10.189-b)
and (x) is twice dierentiable, has undened amplitude, and (a) = (b) = 0. We note that
u coincides with u if = 0
122
It can be shown that the variation and derivation operators are commutative
d
dx (u) = u (x, ) u (x)
u = u (x, ) u (x)
d
du
dx (u) = dx
(10.190)
123 Furthermore, the variational operator and the dierential calculus operator d can be
similarly used, i.e.
(u )2 = 2u u
Victor Saouma
(10.191-a)
Draft
1043
(u + v) = u + v
udx
(10.191-b)
(u)dx
(10.191-c)
u
u
x +
y
x
y
u =
(10.191-d)
however, they have clearly dierent meanings. du is associated with a neighbouring point at
a distance dx, however u is a small arbitrary change in u for a given x (there is no associated
x).
124 For boundaries where u is specied, its variation must be zero, and it is arbitrary elsewhere.
The variation u of u is said to undergo a virtual change.
125
To solve the variational problem of extremizing , we consider

b
F (x, u + , u + )dx
(u + ) = () =
(10.192)
126
Since u u as 0, the necessary condition for to be an extremum is
d()
d
127
=0
(10.193)
=0
From Eq. 10.188 u = u + , and u = u + , and applying the chain rule
d()
=
d
b
a
u
F d F d
u
+
u d u d
dx =
F
F
+
u
dx
(10.194)
for = 0, u = u, thus
d()
d
128
=0
F
F
+
u
u
dx = 0
(10.195)
Integration by part (Eq. 4.1 and 4.1) of the second term leads to
b
129
F
u
dx =
F
u
d F
dx u
(x)
a
dx
(10.196)
Substituting,
d()
d
(x)
a
=0
F
d F
F
+ (x)
u
dx u
u
I (x [a, b])
=0
(10.197)
II (x = a, b)
We will force each one of the two terms to be equal to zero.

130 The fundamental lemma of the calculus of variation states that for continuous (x) in
a x b, and with arbitrary continuous function (x) which vanishes at a and b, then
b
a
Victor Saouma
(x)(x)dx = 0 (x) = 0
(10.198)
Draft
1044
Thus, part I in Eq. 10.197 yields

d F
F
=0
u
dx u
(10.199)
in a < x < b
131 This dierential equation is called the Euler-Lagrange equation associated with and
is a necessary condition for u(x) to extremize .
Generalizing for a functional which depends on two eld variables, u = u(x, y) and
v = v(x, y)
132
F (x, y, u, v, u,x , u,y , v,x , v,y , , v,yy )dxdy
(10.200)
There would be as many Euler equations as dependent eld variables
F
F
F
2 F
u x u,x y u,y + x2 u,xx
F
F
F
2 F
v x v,x y v,y + x2 v,xx
F
2
xy u,xy
F
2
+ xy v,xy
+
+
2 F
y 2 u,yy
2 F
y 2 v,yy
= 0
= 0
(10.201)
133 We note that the Functional and the corresponding Euler Equations, Eq. 10.186 and 10.199,
or Eq. 10.200 and 10.201 describe the same problem.
The Euler equations usually correspond to the governing dierential equation and are referred to as the strong form (or classical form).
134
135 The functional is referred to as the weak form (or generalized solution). This classication
stems from the fact that equilibrium is enforced in an average sense over the body (and the
eld variable is dierentiated m times in the weak form, and 2m times in the strong form).
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.
137 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
a F dx
=
a
F
F
u
u +
u
u
dx
(10.202)
As above, integration by parts of the second term yields

b
u
a
d F
F
u
dx u
dx
(10.203)
140 We have just shown that nding the stationary value of by setting = 0 is equivalent
equal to zero.
to nding the extremal value of by setting d()
d
=0
141 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.
Victor Saouma
Draft
10.7.2
142
1045
Boundary Conditions
Revisiting the second part of Eq. 10.197, we had

(x)
Ess.
F
u
=0
(10.204)
Nat.
Boundary Cond.
This can be achieved through the following combinations
(a) = 0
and
(a) = 0
and
F
(a) = 0
u
F
(a) = 0
u
(b) = 0
F
(b) = 0
u
(10.205-a)
(10.205-b)
and
(b) = 0
(10.205-c)
and
F
(b) = 0
u
(10.205-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.
Natural (or natural or static) if we left to be arbitrary, then it would be necessary to use
F
u = 0 at x = a or b. Natural boundary conditions, involve derivatives of order m
and up. This B.C. is implied by the satisfaction of the variational statement but not
explicitly stated in the functional itself. Mathematically, this corresponds to Neuman
boundary-value problems.
Mixed Boundary-Value problems, are those in which both essential and natural boundary
conditions are specied on complementary portions of the boundary (such as u and
Gammat ).
144
Table 10.4 illustrates the boundary conditions associated with some problems
Example 10-12: Extension of a Bar

The total potential energy of an axial member of length L, modulus of elasticity E, cross
sectional area A, xed at left end and subjected to an axial force P at the right one is given by
L
=
0
Victor Saouma
EA
2
du
dx
dx P u(L)
(10.206)
Draft
1046
Problem
Dierential Equation
m
Essential B.C. [0, m 1]
Natural B.C. [m, 2m 1]
Axial Member
Distributed load
d2
AE dxu + q = 0
2
1
u
du
dx
or xx = Eu,x
Flexural Member
Distributed load
d4 w
EI dx4 q = 0
2
dw
w, dx
2
d w
d w
dx2 and dx3
or M = EIw,xx and V = EIw,xxx
Table 10.4: Essential and Natural Boundary Conditions

where the rst term represents the strain energy sotred in the bar, and the second term denotes
the work done on the bar by the load P in displacing the end x = L through displacement
u(L).
Determine the Euler Equation by requiring that be a minimum.
Solution:
Solution I The rst variation of is given by
L
=
0
EA
du
du
dx P u(L)
2
2
dx
dx
(10.207)
Integrating by parts we obtain

d
du
du L
EA
udx + EA u P u(L) = 0
dx
dx
dx 0
0
L
d
du
du
u
EA
P u(L)
=
dx + EA
dx
dx
dx x=L
0
L
Euler Eq.
du
EA
dx
(10.208-a)
B.C.
u(0)
(10.208-b)
x=0
0
The last term is zero because of the specied essential boundary condition which implies
that u(0) = 0. Rcalling that in an arbitrary operator which can be assigned any
value, we set the coecients of u between (0, L) and those for u at x = L equal to zero
separately, and obtain
Euler Equation:
du
d
EA
dx
dx
=0
0<x<L
(10.209)
Natural Boundary Condition:

EA
du
P =0
dx
at x = L
Solution II We have
F (x, u, u ) =
Victor Saouma
EA
2
du
dx
(10.210)
(10.211)
Draft
1047
(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. 10.199, we evaluate
F
F
= EAu
=0 &
u
u
(10.212-a)
Thus, substituting, we obtain

F
d F
u
dx u
du
EA
dx
= 0
d
(EAu ) = 0 Euler Equation
dx
= 0 B.C.
(10.213-a)
(10.213-b)
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
1
(EIw )w pw dx
2
(10.214)
Derive the rst variational of .

Solution:
Extending Eq. 10.202, and integrating by part twice
L
0
L
=
0
F
F
w +
w dx
w
w
F dx =
(EIw w pw)dx
(EIw w )
= (EIw w )
Nat.
L
0
L
0
Nat.
Ess.
(10.215-b)
(EIw ) w pw dx
L
0 [(EIw
(10.215-a)
) w]|L +
0
(10.215-c)
(EIw ) + p wdx = 0 (10.215-d)

0
Ess.
Euler Eq.
BC
Or
(EIw ) = p
for all x
which is the governing dierential equation of beams and

Essential
w = 0
w = 0
or
or
Natural
EIw = M = 0
(EIw ) = V = 0
at x = 0 and x = L
Victor Saouma
Draft
1048
10.8
Homework
For the cantilivered beam shown below:

q
111111111111111111111111111111111
000000000000000000000000000000000
111111111111111111111111111111111
000000000000000000000000000000000
1
0
111111111111111111111111111111111
000000000000000000000000000000000
1
0
111111111111111111111111111111111
000000000000000000000000000000000
1
0
1
0
111111111111111111111111111111111
000000000000000000000000000000000
1
0
1
0
111
000
111111111111111111111111111111111
000000000000000000000000000000000
1 1 1
0 0 0
1
0
1
0
111
000
111111111111111111111111111111111
000000000000000000000000000000000
1 1 1
0 0 0
1
0
1
0
111
000
111111111111111111111111111111111
000000000000000000000000000000000
1
0
2
111
000 1
3
111
000
111
000
111
000
111
000
L/2
L/2
0
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
x
3x
)a1 + (1 cos
)a2
2L
2L
(10.216)
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.
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
Draft
Chapter 11
INTERPOLATION FUNCTIONS
11.1
Introduction
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 RayleighRitz method
y = a1 + a2 x + a3 x2 + a4 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 + N 2 2 + . . . + N n n
2 For simple problems both Eqn. 11.1 and Eqn. 11.2 can readily provide the exact solutions of
d4
q
the governing dierential equation (such as dxy = EI for exure), but for more complex ones,
4
one must use an approximate one.
11.2
Shape Functions
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, .
3
Ni (x)i = N(x) {}
(11.3)
i=1
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.
Draft
112
N1
N2
Figure 11.1: Axial Finite Element

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).
4
Shape functions should

1. Be continuous, of the type required by the variational principle.
2. Exhibit rigid body motion (i.e. v = a1 + . . .)
3. Exhibit constant strain.
Shape functions should be complete, and meet the same requirements as the coecients of
the Rayleigh Ritz method.
5
Shape functions can often be written in non-dimensional coordinates (i.e. = x ). This will
l
be exploited later by isoparametric elements.
We shall distinguish between two classes of problems, those involving displacements only, and
those involving displacement and their derivatives.
7
The rst class requires only continuity of displacement, and will be referred to as C 0 problems
(truss, torsion), whereas the second one requires continuity of slopes and will be referred to as
C 1 problems.
11.2.1
9
Axial/Torsional
With reference to Fig. 11.1 we start with:

u = N 1 u1 + N2 u2
x = N1 x1 + N2 x2
10
(11.4-a)
(11.4-b)
Since we have 2 d.o.fs, we will assume a linear deformation state

u = a1 x + a2
(11.5)
where u can be either or , and the B.C.s are given by: u = u1 at x = 0, and u = u2 at
x = L. Thus we have:
u 1 = a2
u2 = a1 L + a2
Victor Saouma
(11.6-a)
(11.6-b)
Draft
11.2 Shape Functions
11
113
Solving for a1 and a2 in terms of u1 and u2 we obtain:

u2 u1
L
L
= u1
a1 =
a2
12
(11.7-a)
(11.7-b)
Substituting and rearranging those expressions into Eq. 11.5 we obtain

u2 u1
)x + u1
L
L
x
x
= (1 ) u1 +
u2
L
L
u = (
N1
(11.8-a)
(11.8-b)
N2
or:
N1 = 1
x
N2 = L
11.2.2
13
x
L
(11.9)
Generalization
The previous derivation can be generalized by writing:

a1
a2
x 1
u = a1 x + a2 =
[p]
(11.10)
{a}
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
a1
a2
[L]
{a}
{}
(11.11)
following inversion of [L], this leads to

a1
a2
{a}
15
1 1
L 0
1
L
[L]1
u1
u2
(11.12)
{}
Substituting this last equation into Eq. 11.10, we obtain:

u=
x
(1 L )
[p][L]1
x
L
u1
u2
(11.13)
{}
[N]
Victor Saouma
Draft
114
y,v
V1
V2
1
2
L
Figure 11.2: Flexural Finite Element

16
Hence, the shape functions [N] can be directly obtained from

[N] = [p][L]1
11.2.3
(11.14)
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.
17
v = a1 x3 + a2 x2 + a3 x + a4
dv
= 3a1 x2 + 2a2 x + a3
=
dx
18
(11.15-a)
(11.15-b)
Note that v can be rewritten as:

v
dv
dx
x3 x2 x 1
3x2 2x 1 0
a1
a
2
a3
(11.16)
a4
{a}
19
We now apply the boundary conditions:

v = v1
v = v2
= 1 =
= 2 =
or:
v1
=
v2
2
{}
Victor Saouma
dv
dx
dv
dx
at
at
at
at
x=0
x=L
x=0
x=L
0
0 0 1
0
0 1 0
L3 L2 L 1
3L2 2L 1 0
[L]
a1
a3
(11.17)
a4
{a}
Draft
11.2 Shape Functions
115
N1
N2
N3
N4
=0
Ni Ni,x
1
0
0
1
0
0
0
0
Function
= (1 + 2 3 3 2 )
= (1 )2
= (3 2 2 3 )
= ( 2 )
=1
Ni Ni,x
0
0
0
0
1
0
0
1
Table 11.1: Characteristics of Beam Element Shape Functions

20
which when inverted yields:
a1
a3
1
L3
a4
v1
2
L
2
L
2 3L L2
3L 2L
0
L3
0
0 v2
L3
0
0
0
2
[L]1
{a}
21
(11.18)
{}
Combining Eq. 11.18 with Eq. 11.16, we obtain:
x3 x2 x 1
[p]
1
L3
N2
N3
N4
[p][L]1
v1
v2
(11.19-b)
{}
[N]
where =
(11.19-a)
{}
(1 + 2 3 3 2 ) x(1 )2 (3 2 2 3 ) x( 2 )
N1
v1
2
L
2
L
3L 2L2 3L L2 1
0
L3
0
0 v2
L3
0
0
0
2
[L]1
22
x
L.
Hence, the shape functions for the exural element are given by:
N1 = (1 + 2 3 3 2 )
(11.20)
N2 = x(1 )
(11.21)
N3 = (3 2 )
(11.22)
N4 = x( )
(11.23)
and are shown in Fig 11.3.

23
Table 11.1 illustrates the characteristics of those shape functions
11.2.4
24
Constant Strain Triangle Element
Next we consider a triangular element, Fig. 11.4 with bi-linear displacement eld (in both x
Victor Saouma
Draft
116
Shape Functions for Flexure

(v1; 1; v2; 2)
1.0
0.8
N1
N3
N2
N4
0.6
0.4
0.2
0.0
0.2
0.0
0.2
0.4
(x/L)
0.6
0.8
1.0
Figure 11.3: Shape Functions for Flexure of Uniform Beam Element.
Figure 11.4: Constant Strain Triangle Element
Victor Saouma
Draft
11.3 Interpolation Functions
117
and y):
u = a1 + a2 x + a3 y
(11.24-a)
v = a4 + a5 x + a6 y
a1
(11.24-b)
(11.24-c)
1 x y
2
a
3
[p]
{a}
As before, we rst seek the shape functions, and hence we apply the boundary conditions at
the nodes for the u displacements rst:
25
u1
1 0 0 a1
u2
= 1 x2 0 a2
1 x3 y3 a3
3
26
{a}
[L]
{}
(11.25)
We then multiply the inverse of [L] in Eq. 11.25 by [p] and obtain:
u = N1 u1 + N2 u2 + N3 u3
where
N1 =
N2 =
N3 =
1
x2 y3 (x2 y3 xy3
1
x2 y3 (xy3 x3 y)
y
y3
(11.26)
x2 y + x3 y)
(11.27)
We observe that each of the three shape functions is equal to 1 at the corrsponding node, and
equal to 0 at the other two.
27
The same shape functions can be derived for v:

v = N1 v 1 + N2 v 2 + N3 v 3
28
(11.28)
Hence, the displacement eld will be given by:
u
v
N1 0 N2 0 N3 0
0 N1 0 N 2 0 N 3
u1
v1
v2
u3
v3
(11.29)
The element is refered to as Constant Strain Triangle (CST) because it has a linear displacement eld, and hence a constant strain.
29
11.3
Interpolation Functions
30 Based on the preceding examples, we now seek to derive a general formula for shape
functions of polynomials of various orders.
Victor Saouma
Draft
118
11.3.1
31
C 0 : Lagrangian Interpolation Functions
In our earlier approach, the shape functions were obtained by:

1. Assumption of a polynomial function: = p {a}
2. Application of the boundary conditions {} = [L]{a}
3. Inversion of [L]
4. And nally [N] = [p][L]1
By following these operations, we have in eect dened the Lagrangian Interpolation Functions for problems with C 0 interelement continuity (i.e continuity of displacement only).
32
33 The Lagrangian interpolation denes the coecients ([N] in our case) of a polynomial series
representation of a function in terms of values dened at discrete points (nodes in our case).
For points along a line this would yield:
Ni =
34
(11.30)
If expanded, the preceding equation would yield:

N1 =
N2 =
Nm+1 =
35
m+1
(xxj )
j=1,j=i
m+1
(xi xj )
j=1,j=i
(xx2 )(xx3 )(xxm+1 )

(x1 x2 )(x1 x3 )(x1 xm+1 )
(xx1 )(xx3 )(xxm+1 )
(x2 x1 )(x2 x3 )(x2 xm+1 )
(xx1 )(xx2 )(xxm )
(xm+1 x1 )(xm+1 x2 )(xm+1 xm )
(11.31)
For the axial member, m = 1, x1 = 0, and x2 = L, the above equations will result in:
=
(x L)
x
x
x
1 + 2 = (1 ) 1 +
2
L
L
L
L
N1
(11.32)
N2
which is identical to Eq. 11.8-b.

11.3.1.1
Constant Strain Quadrilateral Element
Next we consider a quadrilateral element, Fig. 11.5 with bi-linear displacement eld (in both
x and y).
36
37 Using the Lagrangian interpolation function of Eq. 11.30, and starting with the u displacement, we perform two interpolations: the rst one along the bottom edge (1-2) and along the
top one (4-3).
38
From Eq. 11.30 with m = 1 we obtain:

u12 =
=
(11.33)
u43 =
=
39
x2 x
x1 x
x2 x1 u1 + x1 x2 u2
ax
x+a
2a u1 + 2a u2
x2 x
x1 x
x2 x1 u4 + x1 x2 u3
ax
x+a
2a u4 + 2a u3
(11.34)
Similarly
Victor Saouma
Draft
11.3 Interpolation Functions
119
Figure 11.5: Constant Strain Quadrilateral Element

Next, we interpolate in the y direction along 1-4 and 2-3 between u12 and u43 . Again, we
use Eq. 11.30 however this time we replace x by y:
40
u =
=
=
y2 y
y1 y
u12 +
u43
(11.35-a)
y2 y1
y1 y2
byx+a
y+bax
y+bx+a
byax
u1 +
u2 +
u4 +
u3
2b 2a
2b 2a
2b 2a
2b 2a
(a x)(b y)
(a + x)(b y)
(a + x)(b + y)
(a x)(b + y)
u1 +
u2 +
u3 +
u4
4ab
4ab
4ab
4ab
N1
N2
N3
N1
One can easily check that at each node i the corresponding Ni is equal to 1, and all others
to zero, and that at any point N1 + N2 + N3 + N4 = 1. Hence, the displacement eld will be
given by:
u1
v
1
u2
u
N1 0 N2 0 N 3 0 N4 0
2
=
(11.36)
v
0 N1 0 N2 0 N3 0 N 4 u3
v3
u
4
v4
41
11.3.1.2
Solid Rectangular Trilinear Element
42 By extension to the previous derivation, the shape functions of a solid rectangular trilinear
solid element, Fig. 11.6 will be given by:
Victor Saouma
Draft
1110
Figure 11.6: Solid Trilinear Rectangular Element
N1 0
0 N2 0
0 N3 0
0 N4 0
v
0 N2 0
0 N3 0
0 N4
= 0 N1 0
0 N2 0
0 N3 0
0
0
0 N1 0
where
Ni =
11.3.2
N4
u1
v1
w1
u2
v2
w2
u3
v3
w3
u4
v4
w4
(a x)(b y)(c z)
8abc
(11.37)
(11.38)
C 1 : Hermitian Interpolation Functions
For problems involving the rst derivative of the shape function, that is with C 1 interelement
continuity (i.e continuity of rst derivative or slope) such as for exure, Hermitian interpolation
functions rather than Lagrangian ones should be used.
43
44 Hermitian interpolation functions are piecewise cubic functions which satisfy the conditions
of displacement and slope (C 0 , C 1 ) continuities. They are exensively used in CAD as Bezier
curves.
11.4
Interpretation of Shape Functions in Terms of Polynomial

Series
45 A schematic interpretation of shape functions in terms of polynomial series terms is given in

Table 11.2.
46
Polynomial terms present in various element formulations is shown in Table 11.3
Victor Saouma
Draft
11.4 Interpretation of Shape Functions in Terms of Polynomial Series
Constant
Linear
Quadratic
Cubic
Quartic
a11 x4
1111
a1
3
a7 x
a4 x2
a12 x3 y
a2 x
a3 y
a5 xy
a8 x y
a13 x2 y 2
a9 xy
a6 y 2
a14 xy 3
a10 y 3
(11.39)
a15 y 4
Table 11.2: Interpretation of Shape Functions in Terms of Polynomial Series (1D & 2D)
Element
Terms
Linear
Quadratic
Bi-Linear (triangle)
Bi-Linear (quadrilateral)
Bi-Quadratic (Serendipity)
Bi-Quadratic (Lagrangian)
a1 ,
a1 ,
a1 ,
a1 ,
a1 ,
a1 ,
a2
a2 ,
a2 ,
a2 ,
a2 ,
a2 ,
a4
a3 ,
a3 , a5
a3 , a4 , a5 , a6 , a8 , a9
a3 , a4 , a5 , a6 , a8 , a9 , a13
# of Nodes
(terms)
2
3
3
4
8
9
Table 11.3: Polynomial Terms in Various Element Formulations (1D & 2D)
Victor Saouma
Draft
1112
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.
3 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
Strain Displacement Relations
4 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) {}
The strain is then dened as:
def
(x) = [B(x)]{}
(12.1)
(12.2)
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
u(x) =
(1
x
)
L
N1
x
L
N2
u1
u2
{}
(12.3-a)
Draft
122
FINITE ELEMENT FORMULATION
(x) = xx
1
du L
=
=
dx
1
L
N1
x
N2
x
[B]
12.1.2
5
u1
u2
(12.3-b)
{}
Flexural Members
Using the shape functions for exural elements previously derived in Eq. 11.23 we have:
y
d2 v
=y 2
dx
1
M
=
=
EI
2v
d
= y 2
dx
2
6
2 (2 1) (3 2)
L
= y L
=
2 N1
x2
2 N2
x2
(12.4-a)
(12.4-b)
(12.4-c)
v1
2
6
(2 + 1) (3 1)
1
L2
L
2N
2N
v2
3
4
x2
x2
[B]
12.2
(12.4-d)
{}
Virtual Displacement and Strains
In anticipation of the application of the principle of virtual displacement, we dene the vectors
of virtual displacements and strain in terms of nodal displacements and shape functions:
6
(x) = [N(x)]{}
(x) = [B(x)]{}
12.3
(12.5-a)
(12.5-b)
Element Stiness Matrix Formulation
In one dimensional elements with initial strain (temperature eect, support settlement, or
other) such that:
x
x =
+
0
(12.6)
x
E
due to load initial strain
7
thus:
x = Ex E0
x
{} = [D]{} [D]{0 }
(12.7)
(12.8)
Generalizing, and in matrix form:
Victor Saouma
Draft
12.3 Element Stiness Matrix Formulation
123
where [D] is the constitutive matrix which relates stress and strain vectors.
9
10
The element will be subjected to a load q(x) acting on its surface

Let us now apply the principle of virtual displacement and restate some known relations:
U
= W
(12.9-a)
{}d
(12.9-b)
{} = [D]{} [D]{0 }
(12.9-c)
{} = [B]{}
(12.9-d)
{} = [B]{}
[B]
(12.9-e)
(12.9-f)
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:
11
[B]T [D][B]{} d
{ }
{ }
{ }
[B] [D][B] d{}

T
12
[B]T [D]{0 } d
{ 0 }
(12.10)
[B] [D]{ }d
T
The virtual external work in turn is given by:

W =
{F}
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
(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]
{F} +
[B]T [D]{0 }d =
{F }
(12.14)
[N] q(x) dx
0
e
{F }
W
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
15
Victor Saouma
Draft
124
FINITE ELEMENT FORMULATION

e
[k], the external nodal force vector F , the distributed element force F
initial displacement.
16
, and the vector of
From this relation we dene:
The element stiness matrix:

[B]T [D][B]d
[k] =
(12.15)
[B]T [D]{0 }d
(12.16)
Element initial force vector:

0
{F } =
Element equivalent load vector:

e
{F } =
[N] q(x) dx
(12.17)
and the general equation of equilibrium can be written as:

0
[k]{} {F } = {F} + {F }
12.3.1
(12.18)
Stress Recovery
17 Whereas from the preceding section, we derived a general relationship in which the nodal
displacements are the primary unknowns, we next seek to determine the internal (generalized)
stresses which are most often needed for design.
18
Recalling that we have:

{} = [D]{}
(12.19-a)
{} = [B]{}
(12.19-b)
With the vector of nodal displacement {} known, those two equations would yield:
{} = [D] [B]{}
(12.20)
We note that the secondary variables (strain and stresses) are derivatives of the primary
variables (displacement), and as such may not always be determined with the same accuracy.
19
Victor Saouma
Draft
Chapter 13
SOME FINITE ELEMENTS

13.1
Introduction
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.
1
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:

du
dx
= [ dN1
dx
xx =
1
= [ L
dN2
dx
1
L ]
]
(13.2-a)
[B]
For the truss element, the constitutive matrix [D] reduces to the scalar E; Hence, substituting
[B]T [D][B]d and with d = Adx for element with constant cross
into Eq. 12.15 [k] =
sectional area we obtain:
[k] = A
0
1
L
1
L
1
L
1
L
dx
(13.3)
Draft
132

AE L
1 1
2
1 1
L
0
1 1
= AE
L
1 1
[k] =
dx
(13.4)
We observe that this stiness matrix is identical to the one earlier derived in Eq. 2.41.
13.3
Flexural Element
6 For a beam element, for which we have previously derived the shape functions in Eq. 11.23
and the [B] matrix in Eq. 12.4-d, substituting in Eq. 12.15:
[B]T [D][B] y 2 dA dx
[k] =
0
and noting that

A
y 2 dA = Iz Eq. 12.15 reduces to

L
[k] =
0
7
(13.5)
[B]T [D][B]Iz dx
(13.6)
For this simple case, we have: [D] = E, thus:

l
[k] = EIz
[B]T [B] dx
(13.7)
8 Using the shape function for the beam element from Eq. 11.23, and noting the change of
integration variable from dx to d, we obtain
1
[k] = EIz
6
L2 (2 1)
2
L (3 2)
6
L2 (2 + 1)
2
L (3 1)
6
L2 (2
1)
2
L (3 2)
6
L2 (2
2
+ 1) L (3 1)
Ld (13.8)
dx
or
[k] =
v1
V1 12EIz
L3
M1 6EIz
L2
V2 12EIz
L3
M2 6EIz
L2
6EIz
L2
4EIz
L
6EIz
L2
2EIz
L
v2
12EIz
L3
6EIz
L2
12EIz
L3
6EIz
L2
6EIz
L2
2EIz
L
6EIz
L2
4EIz
L
(13.9)
Which is identical to the beam stiness matrix derived in Eq. 2.41 from equilibrium relations.
13.4
Triangular Element
9 Having retrieved the stiness matrices of simple one dimensional elements using the principle
of virtual displacement, we next consider two dimensional continuum elements starting with
Victor Saouma
Draft
13.4 Triangular Element
133
the triangular element of constant thickness t made out of isotropic linear elastic material. The
element will have two d.o.fs at each node:
=
13.4.1
u1 u2 u3 v 1 v 2 v 3
(13.10)
Strain-Displacement Relations
10
The strain displacement relations is required to determine [B]
11
For the 2D plane elasticity problem, the strain vector {} is given by:
{} =
xx yy xy
(13.11)
hence we can rewrite the strains in terms of the derivatives of the shape functions through the
matrix [B]:
0
xx
u
yy
= 0 N
(13.12)
y
N
N
xy
y
x
[B]
We note that because we have 3 u and 3 v displacements, the size of [B] and [] are 3 6
and 6 1 respectively.
12
13
Dierentiating the shape functions from Eq. 11.27 we obtain:
1
x2
N1
xx
0
yy
=

xy
x3 x2
{}
x2 y3
N1
y
1
x2
N2
x
x3
x2 y3
N1
y
N3
x
x3 x2
x2 y3
N1
y
1
y3
1
x2
N1
x
N3
x
x3
x2 y3
N2
y
1
x2
N2
x
1
y3
N3
y
0
N3
x
u1
u2
u3
v1
v2
v3
(13.13)
[B]
13.4.2
Stiness Matrix
14 With the constitutive matrix [D] given by Eq. ??, the strain-displacement relation [B] by
Eq. 13.13, we can substitute those two quantities into the general equation for stiness matrix,
Eq. 12.15:
[k]
[B]T [D][B]d
Victor Saouma
Draft
134
1
x2
1
x2
0
0
0
0
x3 x2
x2 y3
x
x2 y3
3
1
y3
1
x2
1
x2
0
0
0
x3 x2
x2 y3
x
x2 y3
3
1
y3
1 2
1
0
1
x2
1
x2
0
0
x3 x2
x2 y3
x
x2 y3
3
1
y3
1
)
2
x3 x2
x2 y3
1
x2
x
x2 y3
3
1
x2
1
y3
tdxdy
(13.14-a)
dvol
[B]
2
y3 + x2
32
2
y3 x3 x32
x2 x32
=
y3 x32
x3 y3 + y3 x32
x2 y3
where =
0
0
[D]
[B]T
13.4.3
1
2 ,
1+
2 ,
2
y3 x3 x32
2
y3 + x2
3
x2 x3
y3 x32 + x3 y3
x3 y3
x2 y3
ET
,
2(1 2 )x2 y3
x2 x32
x2 x3
x2
2
x2 y3
x2 y3
0
y3 x32
y3 x32 + x3 y3
x2 y3
2
y3 + x2
32
2
y3 x3 x32
x2 x32
x2 y3
x2 y3
0
x2 x32
x2 x3
x2
2
x3 y3 + y3 x32
x3 y3
x2 y3
2
y3 x3 x32
2
y3 + x2
3
x2 x3
x32 = x3 x2 , and y32 = y3 y2 .
Internal Stresses
Recall from Eq. 12.20 that {} = [D] [B]{} hence for this particular element we will
have:
15
xx
yy

xy
1
E
1
1 2
0 0
{ }
0
0
1
2 )
1
x2
0
1
x2
0
0
x3 x2
x 2 y3
x
x2 y3
3
1
y3
[D]
0
x3 x2
x 2 y3
1
x2
0
x
x2 y3
3
1
x2
[B]
1
y3
u1
u2
u3
v1
v2
v3
y3
y3
0 x32 x3
y3
0
x32
x3
= y3
x32 x3 x2 y3 y3
where =
x2
x2
u1
u2
u3
v1
v2
v3
(13.15-a)
E
(1 2 )x2 y3
We should note that for this element the stress is independent of x and y because a linear
displacement relation was assumed resulting in a constant strain and stress (for linear elastic
material).
16
13.4.4
17
Observations
For this element we should note that:
Victor Saouma
Draft
13.5 Quadrilateral Element
135
1. Both and are constants

2. Interelement equilibrium conditions are not satised
3. Interelement continuity of displacement is satised
13.5
Quadrilateral Element
Victor Saouma
Draft
136
13.6
Homework
1. Establish the exact displacement expression and the shape functions for an axial member
of length L, and variable cross section given by
A1
F
,u
x, u
,u
2x
3L
A = A1 1
(13.16)
Formulate the relevant stiness matrix. (McGuire and Gallagher, Prob. 9.10).
2. Formulate, by use of the work equivalent load approach, the ctitious nodal forces (F1 , M1 , F2 , M2 )
in a beam for the following load in the interval 1-2.
q = qn sin2
nx
L
(13.17)
3. A three noded axial element, of length 2L, has the following displacement eld:
u=
3x
2L
x2
2L2
2x
L
x2
L2
x
2L +
x2
2L2
u1
2
u
3
(13.18)
is subjected to a temperature change given by:

T = 1
x
x
T3
T1 +
2L
2L
(13.19)
where T1 and T3 are the temperature change at points 1 and 3. Calculate the relevant
initial force vector.
Note: Strongly advised to use Mathematica
Victor Saouma
Draft
Chapter 14
DYNAMIC ANALYSIS
14.1
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.
1
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.
2
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:
3
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)
Draft
142
DYNAMIC ANALYSIS
where m is the mass density matrix equal to I, and b is the vector of body forces. The
Dierential operator L is

0
0
x
0
0

0
0 z
L=
(14.2)
y x 0

z
0 x
0 z y
6
For linear elastic material

= De
(14.3)
for nonlinear material, the constitutive equations can be written as
= Di
(14.4)
where Di is the tangent stiness matrix.

7 Whereas Eq. 14.1 describes the body motion in a strong sense, a weak formulation is obtained
by the principle of minimum complementary virtual work (or Weighted Residual/Galerkin)
uT [LT + b m ]d = 0
u
(14.5)
Applying Gauss divergence theorem, Eq. 4.10:

div qdA =
A
qT nds
( )T qdA
(14.6)
uT nd
(14.7)
uT td = 0
(14.8)
and recalling that Lu = ,
uT LT d =
T d +
thus, Eq. 14.5 transforms into
[uT (m b) + T ]d
u
so far no assumption has been made with regard to material behavior.

Next we will seek the spatial discretization of the virtual work equation (Eq. 14.8). From the
previous chapters, we have the following relations
8
(14.9-a)
u = Nu
T
= u N
u = Nu
B = LN
Victor Saouma
(14.9-c)
(14.9-d)
= Bu
T
(14.9-b)
(14.9-e)
= u B
= Bu
(14.9-f)
(14.9-g)
Draft
14.2 Variational Formulation
14.2.1
9
143
Explicit Time Integration; Central Dierence
For explicit integration, we consider the virtual work (Eq. 14.8) at time t
uT md t =
u
uT bt d +
uT tt d
T t d
Substituting
(14.10)
t
N mNd u
T
T t
N b d +
N t d +
(14.11)
t
fi
t
fe
B d = 0
Since this identity must hold for any admissible u, we conclude that
t
Mu = fe fit
(14.12)
Which represents the semi-discrete linear equation of motion in the explicit time integration in
terms of the mass matrix
NT mNd
M=
(14.13)
the vector of external forces

t
fe =
NT bt d +
NT tt d
(14.14)
and the vector of internal forces

BT t d
(14.15)
BT DBdu
fit =
(14.16)
But since = D = DB, then

fit =
10 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.
11
For linear elastic systems, and if damping was considered

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
Victor Saouma
(14.18)
Draft
144
14.2.2
DYNAMIC ANALYSIS
Implicit Time Integration
14.2.2.1
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
13
T t+t d =
uT tt+t d +
uT bt+t d
(14.19)
For linear problems

t+t = De But+t
14
(14.20)
Again, upon proper substitution, this would yield
t+t +
N mNd u
B De Bd u
t+t
t+t
N b
T t+t
d +
N t
d = 0
t+t
fe
(14.21)
or
t+t
Mu
t+t
+ Kut+t = fe
(14.22)
Which represents the semi-discrete linear equation of motion in the implicit time integration.
14.2.2.2
NonLinear Case
In nonlinear problems, the constitutive matrix D is no longer constant and we must use the
tangential relation which is linearized as
15
= Di
(14.23)
t+t = t +
(14.24)
with the stress at time t + t:
16
Substituting into the virtual work equation (Eq. 14.19)

uT md t+t +
u
T d =
uT tt+t d+
uT bt+t d
T t d (14.25)
new term
17
Again, combining with

= Di Bu
(14.26)
and substituting:
t+t +
NT mNd u
uT [
BT Di Bd u(
Kt
NT bt+t d +
NT tt+t d)
t+t
fe
BT t d] = 0
t
fi
(14.27)
Victor Saouma
Draft
14.3 Time Discretization
145
or
Mu
t+t
t+t
+ Kt u = fe
fit
(14.28)
Which represents the semi-discrete nonlinear equation of motion in the implicit time integration.
14.3
Time Discretization
18 Mathematically Equation ?? represents a system of linear dierential equation of second

order, and in principle, the solution of the equations can be obtained by standard procedures
for the solution of dierential equations with constant coecients.
19 However, the procedures proposed for the solution of general systems of dierential equations
can become very expensive if the order of the matrices is large - unless specic advantage is
taken of the special characteristics of the coecient matrices M, C and K.
20 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
Explicit Time Integration
22 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)
(14.29)
fe (t)
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.
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.
23
14.3.1.1
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
24
Victor Saouma
Draft
146
DYNAMIC ANALYSIS
the velocity and the acceleration at time t:

t
u =
t
u =
ut+t utt
2t
tt 2ut + ut+t
u
t2
(14.30-a)
(14.30-b)
where t is the time step.

t
25
Substituting into Eq. 14.17 (Mu + Cu + Ke u = fe ) we obtain
1
t+t
1
2
1
t
= f e K
M ut 2 M
C utt
t2 M + 2t C u
t
2
t
2t
a0
a1
a2
a0
(14.31)
a1
t
fe
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:
1. Form the stiness matrix K, mass matrix M, and damping matrix C.
0 0
2. Initialise u0 , u , u at time t = 0
3. Select the time step t tcr
4. Determine the constants a0 =
1
,
t2
a1 =
1
2t ,
a2 = 2a0 , a3 =
1
a2 .
5. Select time step t smaller than the critical one.

6. Compute
0
0
ut = u0 tu + a3 u
(14.32)
7. Compute the eective mass matrix M = a0 M + a1 C,

8. Triangularize M (not necessary if lumped mass matrix is used):
M = LDLT
(14.33)
9. For each step:

(a) Determine the eective loads at time t:
t
t
fe = fe (K a2 M)ut (a0 M a1 C)utt
(14.34)
(b) Solve for the displacements at time t + t

t
LDLT ut+t = fe
Victor Saouma
(14.35)
Draft
147
(c) If required, solve for velocities and accelerations, strains and stresses at time t
t
u = a0 (utt 2ut + ut+t )
t
u = a1 (utt + ut+t )
t+t
= Bu
t+t
= De
t+t
(14.36-a)
(14.36-b)
(14.36-c)
t+t
(14.36-d)
(d) Update next time t = t + t and go back to a

28 An explicit time integration has many advantages, the major one is that it does not require
factorisation of the stiness matrix.
If we adopt a lumped, instead of consistent mass, matrix, and there is no damping, then no
factorization is necessary, and Eq. 14.31 reduces to:
29
1
t
M ut+t = fe
t2
where
t
t
fe = fe K
2
M ut
t2
1
M utt
t2
(14.37)
(14.38)
hence, if the mass is diagonal, there is no need to factorize.

Furthermore, if the structure stiness matrix need not be triangularized, then it is not
necessary to assemble the global stiness matrix. It can be shown that
30
Kut =
K(e) ut =
Ft
(e)
(14.39)
which means that K(i) ut needed in Eq. 14.38 can be evaluated on the element level by summing
the contributions from each element to the eective load vector. Thus
t
t
fe = f e
Ft
(i)
i
1
M(utt 2ut )
t2
(14.40)
31 Since no structural stiness matrix needs to be assembled, very large problems can be effectively 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...
32
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
=
(14.42)
T
If this condition is violated, numerical instability occurs.
Victor Saouma
Draft
148
DYNAMIC ANALYSIS
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)
(14.43)
max < max
(e)
where max is the maximum frequency of a single nite element (usually the smallest one). It
can in turn be determined from
e
det(K(e) (max )2 Me ) = 0
(14.44)
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);
% Initialize the constants
a_0=1/(Delta_t^2);a_1=1/(2*Delta_t);a_2=2*a_0;a_3=1/a_2;
% Determine displacement at t-Delta_t;
u_tmdt=u_t-Delta_t*du_t+a_3*ddu_t;
% Determine the effective mass matrix
Mhat=a_0*M+a_1*C;
MhatInv=inv(Mhat);
u(1:2,1:13)=0;
Victor Saouma
Draft
149
d(1:2,1)=u_t;
%===============================================================
% I T E R A T I O N S
%===============================================================
for i=2:13
% Effective Load
Phat=P-(K-a_2*M)*u_t-(a_0*M-a_1*C)*u_tmdt;
% Solve for the displacement
u_tpdt=MhatInv*Phat;
u_tmdt=u_t;
u_t=u_tpdt;
d(1:2,i)=u_tpdt;
end
14.3.1.2
NonLinear Systems
In the nonlinear systems the incremental displacements u, the incremental strains and
stresses and the stiness matrix are updated according to
36
u = ut+t ut
(14.46-a)
= Bu
(14.46-b)
t+t
Dt+t
i
t
(14.46-c)
= +
Kt+t =
i
(14.46-d)
t+t
BT D i
Bd
(14.46-e)
In the algorithm shown above, we must evaluate Ki at each time step.
14.3.2
14.3.2.1
Implicit Case
Newmark Method; Forward dierence
37 Consider the Taylor series expansions of the displacement and velocity terms about the values
at t
ut
2 ut t2 3 ut+t t3
ut+t = ut +
t +
+
(14.47)
t
t2 2!
t3
3!
2 t
3 t+t t2
t+t = ut + u t + u
(14.48)
u
t2
t3
2!
The above two equations represent the approximate displacement and velocity ( ut+t and
u
) by a truncated Taylor series. Looking at the remainder term (last term above),
38
t+t
R1 =
Victor Saouma
3 ut+t t3
t3
3!
2 ut+t
t2
2 ut
t2
t3
3!
2
t+t ut ) t
(u
3!
t+t
t
u )t2
(u
(14.49)
Draft
1410
DYNAMIC ANALYSIS
Note the presence of t + t because of the forward dierence nature of the algorithm.
39
Similarly,
R2 =
2 ut+t
t2
2 t
tu t2
2
t
2!
t+t
t
u )t
(u
3 ut+t t2
t3
2!
(14.50)
and are parameters representing numerical approximations.

Those paramters will account for R1 and R2 plus additional terms which were dropped from
our Taylor series approximation.
40
41
Substituting Equations 14.49 and 14.50 into Equations and respectively, we obtain,
t t
ut+t = ut + u t + u
t+t
+ (u
u )t2
t+t = ut + ut t + (ut+t ut )t
u
42
(14.51)
(14.52)
By rearrangement, we obtain,
t+t = ut + [(1 )ut + ut+t ]t
u
t
t
t+t ]t2
ut+t = ut + u t + [(1/2 )u + u
t+t
t+t
= Mu
fe
t+t
+ Cu
+ Kut+t
(14.53)
(14.54)
(14.55)
Where the last equation is the equation of equilibrium, Eq. 14.22 expressed at time t + t.
43 It can be shown that Newmarks forward dierence assumes constant average acceleration
over the time step.
44 and are parameters that can be determined to obtain integration accuracy and stability.
If = 1/2 and = 1/6, then we recover the linear acceleration method.
The scheme is explicit when is 0. When is 1/2, this explicit form has the same numerical
properties as the central dierence method.
45
46
It can be shown that the Newmark method is

1. unconditionally stable if

1
2
(14.56)
(14.57)
2. conditionally stable if

<
1
2
(14.58)
(14.59)
with the following stability limit:

tcrit =
( 1/2) + [/2 + 2 ( 1/2)2 ]1/2

(/2 )
(14.60)
where is the damping parameter.

Victor Saouma
Draft
14.4 Free Vibration

14.3.2.2
1411
Linear Case
t+t in terms of ut+t and then substituting for ut+t into Eq.
Solving from Eq. 14.54 for u

t+t
t+t
14.53, we obtain equations for u

and u
each in terms of the unknown displacements
t+t only. These two equations for ut+t and ut+t are then substituted in Eq. 14.55 to solve
u
t+t , and ut+t can be determined.
for ut+t , after which, using Eq. 14.53 and 14.54, u

47
48
This leads to the following algorithm:

1. Form the stiness matrix K, mass matrix M, and damping matrix C.
0 0
2. Initialise u0 , u , u at time t = 0
3. Select the time step t and parameters and .
4. Determine the constants a0 =
a5 =
t
2
1
,
t2
a1 =
t ,
a2 =
1
t ,
a3 =
1
2
1, a4 =
1,
2 , a6 = t(1 ), a7 = t.
5. Form the eective stiness matrix K: K = K + a0 M + a1 C

6. Triangularize K = LDLT
7. For each time step:
(a) Determine the eective load at time t + t
t
t+t
t+t
fe
= fe
+ M(a0 ut + a2 u + a3 u ) + C(a1 ut + a4 u + a5 u )
(14.61)
(b) Solve for the displacement at time t + t

t+t
LDLT ut+t = fe
(14.62)
(c) Compute the accelerations and velocities
t+t = a0 (ut+t ut ) a2 ut a3 ut
u
t+t = ut + a6 ut + a7 ut+t
(14.63-a)
(14.63-b)
(d) If necessary solve for the stresses

t+t = f (ut+t )
(14.64)
(e) Increase time step t = t + t

14.3.2.3
14.4
49
NonLinear Case
Free Vibration
The governing equation for the free (undamped) vibration of a structure is

M + Ku = 0
u
(14.65)
where the motion is referred to being free, since there are no applied loads.
50
By assuming a harmonic motion

u = sin t
Victor Saouma
(14.66)
Draft
1412
DYNAMIC ANALYSIS
the natural frequencies and the corresponding mode shapes can be determined from the
generalized eigenvalue problem
2 M = K
(14.67)
or
(K 2 M) = 0
|K 2 M| = 0
(14.69)
|K M| = 0
Since is nontrivial
(14.68)
(14.70)
or with 2 =
which is the characteristic equation, and is called the eigenvalue of the equation, and the
structure is said to respond in the mode corresponding to a particular frequency.
For computational purposes, if we premultiply each side of the preceding equation by M1 ,
then
51
(M1 K I) = 0
(14.71)
It should be noted that a zero eigenvalue is obtained for each possible rigid body motion of a
structure that is not completely supported.
52 Depending on which mass matrix is adopted, slightly dierent results are obtained.
In
general, lumped mass matrices approach the exact value (consistent mass matrix) from below.
53 The mode shapes are shapes , anfd give a relative magnitude of the DOF, not the
absolute values (since they are the solution to a set of homogeneous equations).
54 The natural frequencies and mode shapes provide a fundamental description of the vibrating
structure.
Victor Saouma
Draft
14.5 Homework
14.5
1413
Homework
The mass matrices, as well as the problem statement are taken from Pilkey and Wunderlich,
Mechanics of Structures, Variational and Computational Methods.
1) Show that the consistent mass matrix for a beam element is given by
NT Ndx =
AL
420
156
22L
54
13L
13L 3L2
22L
4L2
54
13L
156 22L
13L 3L2 22L 4L2
(14.72)
If axial motion of the beam is considered (u1 , v1 , 1 , u2 , v2 , 2 ) the consistent mass matrix is
140
0
0
70
0
0
0
156
22L
0
54
13L
2
0
22L
4L
0
13L 3L2
70
0
0
140
0
0
0
54
13L
0
156 22L
0 13L 3L2 0 22L 4L2
AL
T
N Ndx =
420
0
(14.73)
If rotary inertia is included, the additional mass matrix is:
30L
2
Ary
0
0
0
0 36
3L
0 3L 4L2
0
0
0
0 36 3L
0 3L L2
0
0
0
0 36 3L
0 3L L2
0
0
0
0 36 3L
0 3L 4L2
(14.74)
2) Perform two time history analysis of the following frame:

B
P(t)
P(t) [N]
2
10
3.0 m
Y
1111
0000
1111
0000
1111
0000
D 0000
1111
1111
0000
1111
0000
t [sec]
1.732 m
3.0 m
using
1. Explicit central dierence scheme,
2. Implicit Newmark method
The properties of the elements are as follows
Member Iy [cm4 ]
1&2
2,356
3
5245
Victor Saouma
A [cm2 ]
32
66
Draft
1414
DYNAMIC ANALYSIS
E = 200 GN/m2 , and = 7, 800 Kg/m3 . Assume the damping matrix C to be proportional
to the mass matrix, 0.000625M. Take T = 0.5 103 .
Plot the displacements at node C from time t = 0 to t = 2.0 sec.
Note that the displacements at time t = 0 correspond to the static solution (and not zero).
Victor Saouma
Draft
Chapter 15
GEOMETRIC NONLINEARITY
15.1
1
Introduction
With refernce to Fig. 15.1, we distinguish dierent levels of analysis:
Load
First order
elastic analysis
Stiffening
Bifurcation
Elastic Critical Load

Elastic Stability Limit
Bifurcation
Softening
Bifurcation
Second order
elastic analysis
Inelastic Critical Load

Plastic Limit Load
First order inelastic analysis
Inelastic Stability Limit

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.
Draft
152
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.
This chapter will focus on elastic critical load determination. We will begin by reviewing the
derivation of some of the fundamental equations in stability analysis. We will examine both
the strong form, and the weak formulation.
2
15.1.1
3
Strong Form
Column buckling theory originated with Leonhard Euler in 1744.
An initially straight member is concentrically loaded, and all bers remain elastic until buckling occur. For buckling to occur, it must be assumed that the column is slightly bent as shown
in Fig. 15.2. Note, in reality no column is either perfectly straight, and in all cases a minor
4
x and y are
principal axes
Slightly bent position

L
Figure 15.2: Euler Column

imperfection is present.
15.1.1.1
Lower Order Dierential Equation; Essential B.C.
At any location x along the column, the imperfection in the column compounded by the
concentric load P , gives rise to a moment
5
Mz = P y
(15.1)
Note that the value or sign of y is irrelevant.

6
Recalling that
Mz
d2 y
=
2
dx
EI
upon substitution, we obtain the following dierential equation
P
d2 y
y=0
dx2 EI
7
Letting k 2 =
P
EI ,
(15.3)
the solution to this second-order linear dierential equation is

y = A sin kx B cos kx
(15.2)
(15.4)
The two constants are determined by applying the boundary conditions
Victor Saouma
Draft
15.1 Introduction
153
1. y = 0 at x = 0, thus B = 0
2. y = 0 at x = L, thus
A sin kL = 0
(15.5)
9 This last equation can be satised if: 1) A = 0, that is there is no deection; 2) kL = 0, that
is no applied load; or 3)
kL = n
(15.6)
Thus buckling will occur if
P
EI
n 2
L
or
P =
n2 2 EI
L2
The fundamental buckling mode, i.e. a single curvature deection, will occur for n = 1; Thus
Euler critical load for a pinned column is
10
Pcr =
11
2 EI
L2
(15.7)
The corresponding critical stress is

cr =
2 E
L
r
(15.8)
where I = Ar2 .
12
Note that buckling will take place with respect to the weakest of the two axis.
15.1.1.2
Higher Order Dierential Equation; Essential and Natural B.C.
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.
Considering a beam-column subjected to axial and shear forces as well as a moment, Fig.
dv
15.3, taking the moment about i for the beam segment and assuming the angle dx between the
axis of the beam and the horizontal axis is small, leads to
15
M M+
dM
(dx)2
dV
dx + w
+ V +
dx
2
dx
dx P
dv
dx = 0
dx
(15.9)
Neglecting the terms in dx2 which are small, and then dierentiating each term with respect
to x, we obtain
dV
d2 v
d2 M
(15.10)
P 2 =0
dx2
dx
dx
16
Victor Saouma
Draft
154
w
w(x)
1111
0000
1111
0000
y, u
1111
0000
1111
0000
dx
V+ V dx
x
P
v
x
P
dx
P
M+ M dx
x
i
j
Figure 15.3: Simply Supported Beam Column; Dierential Segment; Eect of Axial Force P
17
However, considering equilibrium in the y direction gives

dV
= w
dx
18
(15.11)
From beam theory, neglecting axial and shear deformations, we have

M = EI
d2 v
dx2
(15.12)
19 Substituting Eq. 15.11 and 15.12 into 15.10, and assuming a beam of uniform cross section,
we obtain
d4 v
d2 v
(15.13)
EI 4 P 2 = w
dx
dx
P
Introdcing k 2 = EI , the general solution of this fourth order dierential equation to any set
of boundary conditions is
20
v = C1 sin kx + C2 cos kx + C3 x + C4
21
(15.14)
If we consider again the stability of a hinged-hinged column, the boundary conditions are
Essential
Natural
v = 0,
v,xx = 0 at x = 0
v = 0,
v,xx = 0 at x = L
(15.15)
substitution of the two conditions at x = 0 leads to C2 = C4 = 0. From the remaining

conditions, we obtain
C1 sin kL + C3 L = 0
(15.16-a)
C1 k sin kl = 0
(15.16-b)
these relations are satised either if C1 = C3 = 0 or if sin kl = C3 = 0. The rst alternative

leads to the trivial solution of equilibrium at all loads, and the second to kL = n for n =
1, 2, 3 . For n = 1, the critical load is
Pcr =
2 EI
L2
(15.17)
which was derived earlier using the lower order dierential equation.
Victor Saouma
Draft
15.1 Introduction
155
22 Next we consider a column with one end xed (at x = 0), and one end hinged (at x = L).
The boundary conditions are
v = 0, v,xx = 0 at x = 0
v = 0, v,x = 0 at x = L
(15.18)
These boundary conditions will yield C2 = C4 = 0, and

sin kL kL cos kL = 0
(15.19)
But since cos kL can not possibly be equal to zero, the preceding equation can be reduced to
tan kL = kL
(15.20)
which is a transcendental algebraic equation and can only be solved numerically. We are
essentially looking at the intersection of y = x and y = tan x, Fig. 15.4 and the smallest
10.0
8.0
6.0
4.0
2.0
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
Figure 15.4: Solution of the Tanscendental Equation for the Buckling Load of a Fixed-Hinged
Column
positive root is kL = 4.4934, since k2 =
Pcr =
P
EI ,
the smallest critical load is
(4.4934)2
2
EI =
EI
L2
(0.699L)2
(15.21)
Note that if we were to solve for x such that v,xx = 0 (i.e. an inection point), then x = 0.699L.
23 We observe that in using the higher order dierential equation, we can account for both
natural and essential boundary conditions.
Victor Saouma
10.0
Draft
156
15.1.2
Weak Form
15.1.2.1
Strain Energy
Considering a uniform section prismatic element, Fig. ??, subjected to axial and exural
deformation (no shear), the Lagrangian nite strain-displacement relation is given by ??
24
1
2
2
xx = u,x + (u2 + v,x + w,x )
2 ,x
25
(15.22)
Recalling from Strength of Materials

d2 v =
dx 2
=
M
EI
My
I
=y
d2 v
dx2
(15.23)
thus, the total strain would be

d2 v
dx2
du
y
dx
xx =
Axial
1
2
dv
dx
(15.24)
Large Deformation
Flexure
26 We note that the rst and second terms are the familiar components of axial and exural
strains respectively, and the third one (which is nonlinear) is obtained from large-deection
strain-displacement.
27
The Strain energy of the element is given by

1
2
Ue =
28
E2 d
xx
(15.25)
Substituting Eq. 15.24 into U e we obtain
Ue
1
2
du
dx
dv
dx
d
y dxv
2
29
+y
d2 v
dx2
+ du
dx
dv
dx
1
4
dv
dx
2y
d2 v
dx2
du
dx
(15.26)
EdAdx
Noting that
dA = A;
y 2 dA = I
ydA = 0;
(15.27)
for y measured from the centroid, U e reduces to
1
Ue =
2
du
E A
dx
L
+I
We discard the highest order term

linear instability formulation.
Victor Saouma
d2 v
dx2
A
4
dv
dx
A
+
4
dv
dx
du
+A
dx
dv
dx
dx
(15.28)
in order to transform the above equation into a
Draft
15.1 Introduction
157
30 Under the assumption of an independent prebuckling analysis for axial loading, the axial
load Px is
du
Px
du
A
=
(15.29)
Px = EA
dx
dx
E
Thus Eq. 15.28 reduces to
du
EA
dx
L
1
Ue =
2
d2 v
dx2
+ EI
dv
dx
+ Px
dx
(15.30)
31 We can thus decouple the strain energy into two components, one associated with axial and
the other with exural deformations
e
e
U e = U a + Uf
e
Ua =
e
Uf =
15.1.2.2
1
2
1
2
EA
L
(15.31-a)
du
dx
EI
L
dx
(15.31-b)
d2 v
dx2
dv
dx
+ Px
dx
(15.31-c)
Euler Equation
Recall, from Eq. 10.200 that for a functional in terms of two eld variables (u and v) with
higher order derivatives of the form
32
F (x, y, u, v, u,x , u,y , v,x , v,y , , v,yy )dxdy
(15.32)
there would be as many Euler equations as dependent eld variables, Eq. 10.201
F
y u,y
F
2
xy u,xy
2 F
y 2 u,yy
= 0
33
F
u
F
v
F
x v,x
F
y v,y
F
+ x2 v,xx +
F
2
xy v,xy
2 F
y 2 v,yy
= 0
F
x u,x
2 F
x2 u,xx
2
(15.33)
For the problem at hand, those two equations reduce to

e
Uf =
1
2
2
EIv,xx + Px v,x dx
2
(15.34)
and the corresponding Euler equation will be
F
2 F
+ 2
=0
x v,x x v,xx
(15.35)
The terms of the Euler Equation are given by

F
v,x
F
v,xx
Victor Saouma
= Px v,x
(15.36-a)
= EIv,xx
(15.36-b)
Draft
158
34
Substituting into the Euler equation, and assuming constant Px , and EI, we obtain
EI
d4 v
d2 v
Px 2 = 0
dx4
dx
(15.37)
which is identical to Eq. 15.13
15.2
Finite Element Discretization
Assuming a functional representation of the transverse displacements in terms of the four

joint displacements
35
v = Nu
dv
= N,x u
dx
d2 v
= N,xx u
dx2
36
(15.38-a)
(15.38-b)
(15.38-c)
Substituting this last equation into Eq. 15.31-c, the element potential energy is given by
e
e = Uf + W e
1
1
ue [ke ] {ue } + ue [kg ] {ue } u {P}
=
2
2
(15.39-a)
(15.39-b)
where
[ke ] =
EI {N,xx } N,xx dx
(15.40)
and
[kg ] = P
{N,x } N,x dx
(15.41)
where [ke ] is the conventional element exural stiness matrix.

37 [kg ] introduces the considerations related to elastic instability. We note that its terms solely
depend on geometric parameters (length), therefore this matrix is often referred to as the
geometric stiness matrix.
38 Using the shape functions for exural elements, Eq. 11.23, and substituting into Eq. 15.40
and Eq. 15.41 we obtain
u1
EA
L
0
ke = EA
0
Victor Saouma
v1
0
1
0
12EI
L3
6EI
L2
6EI
L2
4EI
L
12EI
L3
6EI
L2
6EI
L2
2EI
L
u2
EA
L
0
0
EA
L
v2
0
12EI
L3
6EI
L2
0
0
6EI
L2
2EI
L
(15.42)
12EI
0
6EI
L3
L2
6EI
4EI
0
L2
L
Draft
15.3 Elastic Instability; Bifurcation Analysis
159
which is the same element stiness matrix derived earlier in Eq. 13.9.
39
The geometric stiness matrix is given by

u1
0
0
P 0
kg =
L 0
0
0
40
v1
0
1
0
6
5
L
10
L
10
2 2
15 L
0
6
5
L
10
u2
0
0
0
0
0
0
0
L
10
L2
30
v2
0
6
5
L
10
0
6
5
L
10
0
L
10
2
L
30
0
L
10
2 2
15 L
(15.43)
The equilibrium relation is thus

ku = P
(15.44)
where the element stiness matrix is expressed in terms of both the elastic and geometric
components)
k = k e + kg
K = K e + Kg
41
(15.45)
(15.46)
In a global formulation,we would have
we note that the structure becomes stier for tensile load P applied through Kg , and weaker
in compression.
We assume that conservative loading is applied, that is the direction of the load does not
follow the deected direction of the member upon which it acts.
42
15.3
Elastic Instability; Bifurcation Analysis
43 In elastic instability, the intensity of the axial load system to cause buckling is yet unknown,
the incremental stiness matrix must rst be numerically evaluated using an arbitrary chosen
load intensity (since Kg is itself a function of P ).
For buckling to occur, the intensity of the axial load system must be times the initially
arbitrarily chosen intensity of the force. Note that for a structure, the initial distribution of P
must be obtained from a linear elastic analysis. Hence, the buckling load, P is given by
44
P = P
(15.47)
45 Since the geometric stiness matrix is proportional to the internal forces at the start, it
follows that
Kg = K
(15.48)
g
where K corresponds to the geometric stiness matrix for unit values of the applied loading
g
(P = 1).
Victor Saouma
Draft
1510
46
The elastic stiness matrix Ke remains a constant, hence we can write
(Ke + K )u P = 0
g
P
47
The displacements are in turn given by

u = (Ke + K )1 P
g
and for the displacements to tend toward innity (i.e buckling/bifurcation/instability), then
|Ke + K | = 0
g
(15.49)
|K1 Ke + I| = 0
g
(15.50)
which can also be expressed as
This is an eigenvalue problem.

Alternatively, it can simply be argued that there is no unique solution (bifurcation condition)
to u.
48
The lowest value of , crit will give the buckling load for the structure and the buckling
loads will be given by
49
Pcrit = crit P
(15.51)
50 Finally, the corresponding deformed shape is directly obtained from the corresponding eigenvector.
Example 15-1: Column Stability

Determine the buckling load of the following column.
P
1
0
1
0
1
0
1
3
2
L
4
L
11
00
8
7
Victor Saouma
Draft
1511
Solution:
1. The following elastic stiness matrices are obtained
k1 =
e
3
0
12EI
L3
6EI
L2
6EI
L2
4EI
L
12EI
L3
6EI
L2
6EI
L2
5
0
6
0
12EI
L3
6EI
L2
6EI
L2
4EI
L
12EI
L3
6EI
L2
EA
L
EA
k2 =
e
2
0
EA
L
EA
5
0
6
0
12EI
L3
6EI
L2
0
6EI
L2
2EI
L
0
0
12EI
L3
6EI
L2
6EI
L2
8
0
9
0
12EI
L3
6EI
L2
0
6EI
L2
2EI
L
12EI
L3
6EI
L2
6EI
L2
EA
L
0
0
EA
L
2EI
L
6EI
L2
EA
L
0
0
EA
L
0
0
2EI
L
(15.52-a)
4EI
L
(15.52-b)
4EI
L
2. Similarly, the geometric stiness matrices are given by

1
0
0
P 0
L 0
0
0
k1 =
g
4
0
0
P
0
L
0
k2 =
g
2
0
3
0
6
5
L
10
L
10
2 2
15 L
0
6
5
0
L
10
2
L
30
L
10
6
5
L
10
L
10
2 2
15 L
0
6
5
0
L
10
2
L
30
L
10
4
0
0
0
0
0
0
5
0
6
5
L
10
0
7
0
0
0
0
0
8
6
5
L
10
0
6
5
L
10
6
5
L
10
0
L
10
2
L
30
0
L
10
2 2
15 L
9
L
10
2
L
30
0
L
10
2 2
15 L
(15.53-a)
(15.53-b)
3. The structures stiness matrices Ke and Kg can now be assembled from the element
stinesses. Eliminating rows and columns 2, 7, 8, 9 corresponding to zero displacements
in the column, we obtain
Ke =
Victor Saouma
AL2
I
AL2
I
EI
L3
0
0
0
4
2
AL
I
2
2 AL
I
0
0
0
3
5
6
0
0
0
0
0
0
4L2 6L 2L2
6L 24
0
0
8L2
2L2
(15.54)
Draft
1512
and
1
0
0
P
0
Kg =
L 0
4
0
0
0
0
0
3
0
0
2 2
15 L
L
10
L2
30
5
0
0
L
10
12
5
0
0
L2
30
0
4 2
15 L
(15.55)
4. Noting that in this case K = Kg for P = 1, the determinant |Ke + K | = 0 leads to

g
g
1
1
4
3
5
6
AL2
I
2
AL
I
2
AL
I
2
2 AL
I
0
0
0
0
0
0
5. Introducing =
AL2
I
3
0
0
2 2 L4
4L
15 EI
3
1
6L + 10 L
EI
4
1
2L2 + 30 L
EI
and =
1
1
4
3 0
5 0
6 0
L2
EI ,
5
0
0
3
1
6L + 10 L
EI
2
24 12 L
5 EI
0
6
0
0
2 + 1 L4 = 0
2L
30 EI
0
4
4
8L2 15 L
EI
(15.56)
the determinant becomes
4
3
0
2
0
0 2 2 15
0 6L + 10
0
2 + 30
5
0
0
6L +
12 2
0
10
6
0
0
2 + 30
0
4 2 15
=0
(15.57)
6. Expanding the determinant, we obtain the cubic equation in

33 2202 + 3, 840 14, 400 = 0
(15.58)
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)
and thus, the numerical value is about 2.6 percent higher than the exact one.
51
The mathematica code for this operation is:
(* 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
,
}
ke1=N[ke[e,a,l,i]]
ke2=N[ke[e,a,l,i]]
Victor Saouma
*)
-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
},
},
},
},
},
}
Draft
1513
(* Assemble structure elastic stiffness matrices *)

ke={
{ke1[[3,3]], ke1[[3,5]]
, ke1[[3,6]]
},
{ ke1[[5,3]], ke1[[5,5]]+ke2[[2,2]], ke1[[5,6]]+ke2[[2,3]]},
{ ke1[[6,3]], ke1[[6,5]]+ke2[[3,2]], ke1[[6,6]]+ke2[[3,3]]}
}
WriteString["mat.out",MatrixForm[ke1]]
WriteString["mat.out",MatrixForm[ke]]
(* Define geometric stiffness matrices *)
kg[p_,l_]:=p/l{
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , 6/5 , l/10
, 0 , - 6/5 , l/10
},
{0 , l/10 , 2 l^2/15 , 0 , - l/10 , - l^2/30 },
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , -6/5 , - l/10
, 0 , 6/5
, - l/10
},
{0 , l/10 , - l^2/30 , 0 , - l/10 , 2 l^2/15 }
}
kg1=kg[p,l]
kg2=kg[p,l]
(* Assemble structure geometric stiffness matrices *)
kg={
{kg1[[3,3]], kg1[[3,5]]
, kg1[[3,6]]
},
{ kg1[[5,3]], kg1[[5,5]]+kg2[[2,2]], kg1[[5,6]]+kg2[[2,3]]},
{ kg1[[6,3]], kg1[[6,5]]+kg2[[3,2]], kg1[[6,6]]+kg2[[3,3]]}
}
(* Determine critical loads in terms of p (note p=1) *)
p=1
keigen= l^2 (Inverse[kg] . ke)/( e i)
pcrit=N[Eigenvalues[keigen]]
(* Alternatively*)
knew =ke - x kg
pcrit2=NSolve[Det[knew]==0,x]
Example 15-2: Frame Stability

Determine the buckling load for the following frame. Neglect axial deformation.
2
3
u1
u1
I=200
I=50
10
I=100
111
000
Victor Saouma
111
000
15
Draft
1514
Solution:
The element stiness matrices are given by
u1
20
1, 208
k1 =
e
u1
0.01
0.10
= P

k1
g
2
1, 208
96, 667
2
0.10
16.00
0
2
128, 890
64, 440
0
3
64, 440

128, 890
k2 =
e
u1
47
1, 678
k3 =
e
u1
0.01667
0.1
= P

k3
g
3
1, 678
80, 556
3
0.1
9.6
(15.60-a)
(15.60-b)
(15.60-c)
(15.60-d)
(15.60-e)
The global equilibrium relation can now be written as

(Ke P Kg ) = 0
u1
(66.75) P (0.026666)
(1, 208.33) P (0.1)
(1, 678.24) P (0.1)
2
3
(1, 208.33) P (0.1) (1, 678.24) P (0.1)
(225, 556.) P (16.) (64, 444.4) P (0) = 0
(64, 444.) P (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
Victor Saouma
Draft
15.4 Second-Order Elastic Analysis; Geometric Non-Linearity
1515
e1=29000
e2=e1
e3=e1
(* Define elastic stiffness matrices *)
ke[e_,a_,l_,i_]:={
{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
, 4 e i/l
, 0
, -6 e i/l^2 , 2 e i/l
},
{-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
, 6e i/l^2
, 2 e i/l
, 0
, -6 e i/l^2 , 4 e i/l
}
}
ke1=ke[e1,a1,l1,i1]
ke2=ke[e2,a2,l2,i2]
ke3=ke[e3,a3,l3,i3]
kg[l_,p_]:=p/l{
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , 6/5 , l/10
, 0 , - 6/5 , l/10
},
{0 , l/10 , 2 l^2/15 , 0 , - l/10 , - l^2/30 },
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , -6/5 , - l/10
, 0 , 6/5
, - l/10
},
{0 , l/10 , - l^2/30 , 0 , - l/10 , 2 l^2/15 }
}
kg1=kg[l1,1]
kg3=kg[l3,1]
(* Assemble structure elastic and geometric stiffness matrices *)
ke={
{ ke1[[2,2]]+ke3[[2,2]] , ke1[[2,3]]
, ke3[[2,3]]
},
{ ke1[[3,2]]
, ke1[[3,3]]+ke2[[3,3]] , ke2[[3,6]]
},
{ ke3[[3,2]]
, ke2[[6,3]]
, ke2[[6,6]]+ke3[[3,3]] }
}
kg={
{ kg1[[2,2]]+kg3[[2,2]] , kg1[[2,3]] , kg3[[2,3]] },
{ kg1[[3,2]]
, kg1[[3,3]] , 0
},
{ kg3[[3,2]]
, 0
, kg3[[3,3]] }
}
(* Determine critical loads in terms of p (note p=1) *)
p=1
keigen=Inverse[kg] . ke
pcrit=N[Eigenvalues[keigen]]
modshap=N[Eigensystems[keigen]]
15.4
Second-Order Elastic Analysis; Geometric Non-Linearity
From Eq. 15.45 it is evident that since kg depends on the magnitude of Px , which itself may
be an unknown in a framework, then we do have a geometrically non-linear problem.
51
52
We rewrite Eq. 15.46 as

[Ke + Kg ] u = P
(15.63)
but since Kg depends on the axial laod P , the preoblem is nonlinear.

Victor Saouma
Draft
1516
53 A simple way to solve this nonlinear equation is to use a step-by-step incremental procedure. The linearized incremental formulation can be obtained by applying an incremental
operator
= [Ke + Kg ]i1 {u}i
{u}i = [Ke + Kg ]1 P
i1
(15.64)
i
(15.65)
Example 15-3: Eect of Axial Load on Flexural Deformation

Determine the midspan displacement and member end forces for the beam-column shown
below in terms of Px ; The concentrated force is 50 kN applied at midspan, E=2 109 kN/m2
and I=2 103 m4 .
50 kN
1
0
80,000 kN
1
0
1
0
1
0
6m
6m
1
0
1
0
1
0
1
0
80,000 kN
Solution:
Using two elements for the beam column, the only degrees of freedom are the deection and
rotation at midspan (we neglect the axial deformation).
The element stiness and geometric matrices are given by
0
0
0
0
v1
2
0
0
0
0
0
0
0
222, 222.
666, 666.
0 222, 222. 666, 666.
0
666, 666. 2, 666, 666 0. 666, 666. 1, 333, 333
[K1 ] =
e
0
0
0
0
0
0
0 222, 222. 666, 666. 0

222, 222. 666, 666.
0 666, 666. 1, 333, 333 0. 666, 666. 2, 666, 666
0
v1
2
0
0
0
0
0
0
0
0
0
0
222, 222.
666, 666.
0 222, 222. 666, 666.
0
666, 666. 2, 666, 666 0. 666, 666. 1, 333, 333
2
[Ke ] =
0
0
0
0
0
0 222, 222. 666, 666. 0

222, 222. 666, 666.
0 666, 666. 1, 333, 333 0. 666, 666. 2, 666, 666
(15.66)
0
0
0
0
0 16, 000
0 8, 000
1
[Kg ] =
0
0
0
16, 000
0 8, 000
Victor Saouma
0
0
8, 000
64, 000
0
8, 000
16, 000
(15.67)
0
v1
2
0
0
0
0
16, 000
8, 000
0
8, 000
16, 000
(15.68)
0
0
0
0 16, 000
8, 000
0.
8, 000
64, 000
Draft

0
v1
0
0
0 16, 000
0 8, 000
2
[Kg ] =
0
0
0
16, 000
0 8, 000
2
0
8, 000
64, 000
0
8, 000
16, 000
0
0
0
0
0
16, 000
0
8, 000
0
0
0 16, 000
0.
8, 000
0
8, 000
16, 000
8, 000
64, 000
1517
(15.69)
Assembling the stiness and geometric matrices we get

v1
412, 444.
0.
[K] =
2
0.
5, 205, 330
(15.70)
and the displacements would be

v1
2
0.00012123
0
(15.71)
and the member end forces for element 1 are given by
Plf t
V
lf t
Prgt
Vrgt
lf t
Mrgt
[K1 ] + [K1 ]
e
g
ulf t
v
lf t
urgt
vrgt
lf t
rgt
0
0
0
0
0 206, 222.
0 658, 667.
0
0
0 206, 222
0 658, 667.
25.
79.8491
=
0.
25.
79.8491
0
0
658, 667.
260, 2670
0
658, 667.
1, 349, 330
0
v1
2
0
0
,0
0
0 206, 222. 658, 667.

0
0 658, 667. 1, 349, 330

0
0
0
0
0
0.00012123
0 206, 222. 658, 667.
0 658, 667. 2, 602, 670

0
(15.72-a)
Note that had we not accounted for the axial forces, then
v1
2
Plf t
V
lf t
Prgt
Vrgt
lf t
Victor Saouma
Mrgt
0.0001125
0
25.
75.
(15.73-a)
(15.73-b)
0.
25.
75.
Draft
1518
Alternatively, if instead of having a compressive force, we had a tensile force, then

v1
2
Plf t
V
lf t
Mlf t
Prgt
Vrgt
Mrgt
0.000104944
0
(15.74-a)
0
25.
70.8022
0.
25.
70.8022
(15.74-b)
We observe that the compressive force increased the displacements and the end moments,
whereas a tensile one stiens the structure by reducing them.
The Mathematica to solve this problem follows
OpenWrite["mat.out"]
a1=1; a2=1;
i1=1 l^3/12;i2=i1;
l1=12;l2=12;
e1=200000;e2=e1;e3=e1;
theta1=N[Pi/8];theta2=Pi-theta1;
load
i1=i
i2=i1
l=6
l1=l
l2=6
p=-80000 (* negative compression *)
load={-50,0}
(* Define elastic stiffness matrices *)
ke[e_,a_,l_,i_]:={
{e a/l , 0
, 0
, -e a/l , 0
{0
, 12 e i/l^3 , 6 e i/l^2 , 0
, -12 e i/l^3
{0
, 6 e i/l^2
, 4 e i/l
, 0
, -6 e i/l^2
{-e a/l , 0
, 0
, e a/l , 0
{ 0
, -12 e i/l^3 , -6 e i/l^2 , 0
, 12 e i/l^3
{ 0
, 6e i/l^2
, 2 e i/l
, 0
, -6 e i/l^2
}
ke1=N[ke[e,a1,l1,i1]]
ke2=N[ke[e,a2,l2,i2]]
(* Assemble structure elastic stiffness matrices *)
ke=N[{
{ ke1[[5,5]]+ke2[[2,2]], ke1[[5,6]]+ke2[[2,3]]},
{ ke1[[6,5]]+ke2[[3,2]], ke1[[6,6]]+ke2[[3,3]]}
}]
WriteString["mat.out",MatrixForm[ke]]
kg[p_,l_]:=p/l {
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , 6/5 , l/10
, 0 , - 6/5 , l/10
},
{0 , l/10 , 2 l^2/15 , 0 , - l/10 , - l^2/30 },
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , -6/5 , - l/10
, 0 , 6/5
, - l/10
},
{0 , l/10 , - l^2/30 , 0 , - l/10 , 2 l^2/15 }
Victor Saouma
,
,
,
,
,
,
0
6 e i/l^2
2 e i/l
0
-6 e i/l^2
4 e i/l
},
},
},
},
},
}
Draft
1519
}
kg1=N[kg[p,l1]]
kg2=N[kg[p,l2]]
(* Assemble structure geometric stiffness matrices *)
kg=N[{
{ kg1[[5,5]]+kg2[[2,2]], kg1[[5,6]]+kg2[[2,3]]},
{ kg1[[6,5]]+kg2[[3,2]], kg1[[6,6]]+kg2[[3,3]]}
}]
(* Determine critical loads and normalize wrt p *)
keigen=Inverse[kg] . ke
pcrit=N[Eigenvalues[keigen] p]
(* Note that this gives lowest pcrit=1.11 10^6, exact value is 1.095 10^6 *)
(* Add elastic to geometric structure stiffness matrices *)
k=ke+kg
(* Invert stiffness matrix and solve for displacements *)
km1=Inverse[k]
dis=N[km1 . load]
(* Displacements of element 1*)
dis1={0, 0, 0, 0, dis[[1]], dis[[2]]}
k1=ke1+kg1
(* Member end forces for element 1 with axial forces *)
endfrc1=N[k1 . dis1]
(* Member end forces for element 1 without axial forces
knopm1=Inverse[ke]
disnop=N[knopm1 . load]
disnop1={0, 0, 0, 0, disnop[[1]], disnop[[2]]}
(* Displacements of element 1*)
endfrcnop1=N[ke1 . disnop1]
*)
Example 15-4: Bifurcation

Analyse the stability of the following structure. Compare the axial force caused by the
coupled membrane/exural eects with the case where there is no interaction.
1,000
12
1
0
1
0
/8
1
0
1
0
1
0
1
0
1
0
12
/8
1
0
1
0
Solution:
In the following solution, we will rst determine the axial forces based on the elastic stiness
matrix only. Then, on the basis of those axial forces, we shall determine the geometric stiness
matrix, and solve for the displacements. Because of the non-linearity of the problem, we may
Victor Saouma
Draft
1520
have to iterate in order to reach convergence. Following each analysis, we shall recompute the
geometric stiness matrix on the basis of the axial loads detemined from the previous iteration.
Note that convergence will be reached only for stable problems. If the method fails to
converge, it implies possible biurcation which could be caused by elastic displacements approaching L sin , due to either being too small, or E being too small (i.e not sti enough).
NEEDS SOME CORRECTION
a1 = 1
a2 = 1
i1 = 1 1^3/12
i2 = i1
l1 = 12
l2 = 12
e1 = 200000
e2 = e1
e3 = e1
theta1 =N[Pi/8]
theta2 = Pi-theta1
load ={0, -1000, 0}
normold = 0
epsilon = 0.01
puncpl = load[[2]] / (Sin[theta1] 2)
(*
Define elastic stiffness matrices
*)
ke[e_,a_,l_,i_] := {
{e a/l , 0
, 0
, -e a/l , 0
{0
, 12 e i/l^3 , 6 e i/l^2 , 0
, -12 e i/l^3
{0
, 6 e i/l^2
, 4 e i/l
, 0
, -6 e i/l^2
{-e a/l , 0
, 0
, e a/l , 0
{ 0
, -12 e i/l^3 , -6 e i/l^2 , 0
, 12 e i/l^3
{ 0
, 6e i/l^2
, 2 e i/l
, 0
, -6 e i/l^2
}
(*
Define geometric stiffness matrix
*)
kg[l_,p_] := p/l {
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , 6/5 , l/10
, 0 , - 6/5 , l/10
},
{0 , l/10 , 2 l^2/15 , 0 , - l/10 , - l^2/30 },
{0 , 0
, 0
, 0 , 0
, 0
},
{0 , -6/5 , - l/10
, 0 , 6/5
, - l/10
},
{0 , l/10 , - l^2/30 , 0 , - l/10 , 2 l^2/15 }
}
(*
Define Transformation matrix and its transpose
*)
gam[theta_] := {
{ Cos[theta] , Sin[theta], 0 , 0
, 0
,
{ -Sin[theta], Cos[theta], 0 , 0
, 0
,
{ 0
, 0
, 1 , 0
, 0
,
{ 0
, 0
, 0 , Cos[theta] , Sin[theta] ,
{ 0
, 0
, 0 , -Sin[theta] , Cos[theta] ,
{ 0
, 0
, 0 , 0
, 0
,
}
gamt[theta_] := {
Victor Saouma
,
,
,
,
,
,
0
6 e i/l^2
2 e i/l
0
-6 e i/l^2
4 e i/l
0
0
0
0
0
1
},
},
},
},
},
}
},
},
},
},
},
}
Draft
1521
{ Cos[theta] , -Sin[theta], 0 , 0
, 0
, 0 },
{ Sin[theta] , Cos[theta] , 0 , 0
, 0
, 0 },
{ 0
, 0
, 1 , 0
, 0
, 0 },
{ 0
, 0
, 0 , Cos[theta] , -Sin[theta] , 0 },
{ 0
, 0
, 0 , Sin[theta] , Cos[theta] , 0 },
{ 0
, 0
, 0 , 0
, 0
, 1 }
}
(*
Define functions for local displacments and loads
*)
u[theta_,v1_,v2_] := Cos[theta] v1 + Sin[theta] v2
(*
Transformation and transpose matrices
*)
gam1 = gam[theta1]
gam2 = gam[theta2]
gam1t = gamt[theta1]
gam2t = gamt[theta2]
(*
Element elastic stiffness matrices
*)
ke1 = ke[e1, a1, l1, i1]
ke2 = ke[e2, a2, l2, i2]
Ke1 = gam1t . ke1 . gam1
Ke2 = gam2t . ke2 . gam2
(*
Structures global stiffness matrix
*)
Ke={
{ Ke1[[4,4]] + Ke2[[1,1]] , Ke1[[4,5]] + Ke2[[1,2]] , Ke1[[4,6]] + Ke2[[1,3]] },
{ Ke1[[5,4]] + Ke2[[2,1]] , Ke1[[5,5]] + Ke2[[2,2]] , Ke1[[5,6]] + Ke2[[2,3]] },
{ Ke1[[6,4]] + Ke2[[3,1]] , Ke1[[6,5]] + Ke2[[3,2]] , Ke1[[6,6]] + Ke2[[3,3]] }
}
(*
======= uncoupled analysis ==========
*)
dise=Inverse[Ke].load
u[theta_,diseg1_,diseg2_] := Cos[theta] diseg1 + Sin[theta] diseg2
uu1 = u[ theta1, dise[[1]], dise[[2]] ]
uu2 = u[ theta2, dise[[1]], dise[[2]] ]
up1 = a1 e1 uu1/l1
up2 = a2 e2 uu2/l2
(*
========== Coupled Nonlinear Analysis ==============
Start Iteration
*)
diseg = N[dise]
For[ iter = 1 , iter <= 100, ++iter,
(* displacements in local coordinates *)
disloc={ 0,0,0,
u[ theta1, diseg[[1]], diseg[[2]] ],
u[ theta2, diseg[[1]], diseg[[2]] ],
0};
(* local force *)
ploc = ke1 . disloc;
p1 = ploc[[4]];
p2 = p1;
kg1 = kg[ l1 , p1 ];
kg2 = kg[ l2 , p2 ];
Victor Saouma
Draft
1522
Kg1 = gam1t . kg1 . gam1;

Kg2 = gam2t . kg2 . gam2;
Kg={
{ Kg1[[4,4]] + Kg2[[1,1]] , Kg1[[4,5]] + Kg2[[1,2]] , Kg1[[4,6]] + Kg2[[1,3]] },
{ Kg1[[5,4]] + Kg2[[2,1]] , Kg1[[5,5]] + Kg2[[2,2]] , Kg1[[5,6]] + Kg2[[2,3]] },
{ Kg1[[6,4]] + Kg2[[3,1]] , Kg1[[6,5]] + Kg2[[3,2]] , Kg1[[6,6]] + Kg2[[3,3]] }
};
(*
Solve
*)
Ks = Ke + Kg;
diseg = Inverse[Ks] . load;
normnew = Sqrt[ diseg . diseg ];
ratio = ( normnew-normold ) / normnew;
Print["Iteration ",N[iter],"; u1 ",N[u1],"; p1 ",N[p1]," ratio ",N[ratio]];
normold = normnew;
If[ Abs[ ratio ] < epsilon, Break[] ]
]
Print[" p1 ",N[p1]," up1 ",N[up1]," p1/up1 ",N[p1/up1]," ratio ",N[ratio]]
15.5
Summary
Victor Saouma
Draft
15.5 Summary
1523
STRONG FORM
WEAK FORM
2nd Order D.E.

2 B.C.
4th Order D.E.

4 B.C.
d2
x = du y dxv
2
dx
1
2
dv
dx
d2 y P = 0
dx2 EI
v = A sin kx B cos kx
U=
1
2
2 E d
?
4
d
d
EI dxv P dxv = w
4
2
v = C1 sin kx + C2 cos kx + C3 x + C4
K=
Kg
Ke
P = (Ke + Kg )u
|Ke + Kg | = 0
Figure 15.5: Summary of Stability Solutions
Victor Saouma
Draft
1524
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
Draft
2
REFERENCES
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
2. Cook, Malkus, and Plesha, Concepts and Applications of Finite Element Analysis,
John Wiley & Sons, 1989 (Third Edition)
3. Gallagher, R.H., Finite Element Analysis Fundamentals, Prentice Hall, Inc., Englewood Clis, N.J., 1979
4. Hinton and Owen, An Introduction to Finite Element Computation, Pineridge Press,
Swansea U.K., 1978
5. Hughes, T.R., The Finite Element Method, Linear Static and Dynamic Finite Element
Analysis, Prentice Hall, 1987
6. Zienkiewicz, O., and Taylor, R., The Finite Element Method, Vol. 1 Basic Formulation
and Linear Problems, 4th Ed., McGraw-Hill, 1989
Energy Methods :
1. Pilkey and Wunderlich, Mechanics of Structures, Variational and Computational Methods, CRC Press, 1994
2. Langhaar, H., Energy Methods in Applied Mechanics, John Wiley and Sons, N.Y.,
1962
3. Reddy, J.N., Energy and Variational Methods in Applied Mechanics, John Wiley and
Sons, 1984.
Numerical Techniques :
1. Jennings, A., Matrix Computations for Engineers and Scientists, John Wiley and
Sons, N.Y., 1977.
2. Hilderbrand, F.B., An Introduction to Numerical Analysis, McGraw-Hill, N.Y., 1974
3. Press, W., et. al., Numerical Recipes, The Art of Scientic Computing, Cambridge
University Press, 1987
Journals :
1. Journal of Structural Engineering, American Society of Civil Engineering
2. Computers and Structures
3. International Journal for Numerical Methods in Engineering
Victor Saouma
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
Denitions
Matrix:
A11
A21
.
.
.
[A] =
Ai1
.
.
.
A12
A22
.
.
.
...
...
..
.
Ai2
.
.
.
. . . Aij
.
..
.
.
.
. . . Amj
Am1 Am2
A1j
A2j
.
.
.
...
...
..
.
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:
{X} =
B1
B
2
.
.
Bi
.
.
Bm
(A.2)
A row vector would be

C =
B1 B2 . . . 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
Draft
A2
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
0,
1,
Iij =
if i = j
if i = j
(A.4)
Diagonal matrix: is a square matrix with all its entries equal to zero except the diagonal
terms which are dierent from zero. It is often denoted as [D], and
Dij =
0,
if i = j
= 0, if i = j
(A.5)
Upper Triangular matrix: is a square matrix with all its entries equal to zero, except those
along and above the diagonal. It is often denoted as [U], and
Uij =
0,
if i > j
= 0, if i j
(A.6)
Lower Triangular matrix: is a square matrix with all its entries equal to zero except those
along and below the diagonal. It is often denoted as [L], and
Lij =
0,
if i < j
= 0, if i j
(A.7)
Orthogonal matrices: [A]mn and [B]mn are said to be orthogonal if [A]T [B] = [B]T [A] =
[I]
A square matrix [C]mm is orthogonal if [C]T [C] = [C] [C]T = [I]
Trace of a matrix: tr(A) =
n
i=1 Aii
Submatrices: are matrices within a matrix, for example
5 3 1
[A] = 4 6 2 =
10 3 4
[A11 ] [A12 ]
[A21 ] [A22 ]
(1.8-a)
1 5
[B] = 2 4 =
3 2
[B1 ]
[B2 ]
(1.8-b)
[A11 ] [B1 ] + [A12 ] [B2 ]

[A21 ] [B1 ] + [A22 ] [B2 ]
[A11 ] [B1 ] =
(1.8-c)
5 3
4 6
[A] [B] =
(1.8-d)
1 5
2 4
[A22 ] [B2 ] = [4] [3 2] = [12 8]
14 34
[A] [B] = 22 48
28 70
11 37
16 44
(1.8-e)
(1.8-f)
Operate on submatrices just as if we were operating on individual matrix elements.
Victor Saouma
Draft
A.2 Elementary Matrix Operations
A.2
A3
Elementary Matrix Operations
Transpose: of a matrix [A]mn is another matrix [B] = [B]T

nm such that Bij = Aji Note
that
([A] [B])T = [B]T [A]T
(A.9)
Addition (subtraction):
[A]mn = [B]mn + [C]mn
Aij
= Bij + Cij
(1.10-a)
(1.10-b)
(1.10-c)
Scalar Multiplication:
[B] = k [A]
Bij
(1.11-a)
= kAij
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 = [B]mp [C]pn
Aij
Bi
1p
(1.12-a)
{Cj }p1
11
p
Bir Crj
(1.12-b)
r=1
(1.12-c)
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
The Determinant of a matrix [A]nn , denoted as det A or |A|, is recursively dened as

n
(1)1+j a1j det A1j
det A =
(A.13)
j=1
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
a21 a22
Victor Saouma
= a11 a22 a12 a21
(A.14)
Draft
A4
For a 3 3 matrix
a11 a12 a13
a21 a22 a23
a31 a32 a33
= a11
a22 a23
a32 a33
a12
a21 a23
a31 a33
+ a13
a21 a22
a31 a32
= 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
(1.15-a)
(1.15-b)
(1.15-c)
(1.15-d)
(1.15-e)
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:
1. The determinant of the transpose of a matrix is equal to the determinant of the matrix
| A |=| AT |
(1.16)
2. If at least one row or one column is a linear combination of the other rows or columns,
then the determinant is zero. The inverse is also true, if the determinant is equal to zero,
then at least one row or one column is a linear combination of of other rows or columns.
3. If there is linear dependancy between rows, then there is also one between columns and
vice-versa.
4. The determinant of an upper or lower triangular matrix is equal to the product of the
main diagonal terms.
5. The determinant of the product of two square matrices is equal to the product of the
individual determinants
| AB |=| A || B |
(1.17)
A.4
Singularity and Rank
If the deteminant of a matrix [A]nn is zero, than the matrix is said to be singular. As we
have seen earlier, this means that there is at least one row or one column which is a linear
combinations of the others. Should we remove this row and column, we can repeat the test for
singularity until the size of the submatrix is r r. Then we refer to r as the rank of the matrix
or rank(A) = r. We deduce that the rank of a nonsigular n n matrix is n. If the rank of a
matrix r is less than its size n, we say that it has n r rank deciency.
If n is the size of the global stiness matrix of a structure in which the boundary conditions
have not been accounted for (n = is equal to the total number of nodes times the total number
of degrees of freedom per node) would have a rank r equal to n minus the number of possible
rigid body motions (3 and 6 in two and three dimensional respectively).
Victor Saouma
Draft
A.5 Inversion
A.5
A5
Inversion
The inverse of a square (nonsingular) matrix [A] is denoted by [A]1 and is such that
[A] [A]1 = [A]1 [A] = [I]
(1.18)
Some observations
1. The inverse of the transpose of a matrix is equal to the transpose of the inverse
AT
= A1
(1.19)
2. The inverse of a matrix product is the reverse product of the inverses

([A] [B])1 = [B]1 [A]1
(1.20)
3. The inverse of a symmetric matrix is also symmetric

4. The inverse of a diagonal matrix is another diagonal one with entries equal to the inverse
of the entries of the original matrix.
5. The inverse of a triangular matrix is a triangular matrix.
6. It is computationally more ecient to decompose a matrix ([A] = [L] [D] [U]) using upper
and lower decomposition or Gauss elimination) than to invert a matrix.
A.6
Eigenvalues and Eigenvectors
A special form of the system of linear equation
[A] =
A11 A12 . . . A1n

A21 A22 . . . A2n
.
.
.
..
.
.
.
.
.
.
.
Ai1 Ai2 . . . Ann
x1
x2
.
.
.
xn
B1
B2
.
.
.
Bn
(1.21)
is one in which the right hand side is a multiple of the solution:

[A] {x} = {x}
(1.22)
[A I] {x} = 0
(1.23)
which can be rewritten as

A nontrivial solution to this system of equations is possible if and only if [A I] is singular
or
|A I| = 0
(1.24)
or
[A] =
Victor Saouma
A11
A12
A21
A22
.
.
.
.
.
.
Ai1
Ai2
...
A1n
...
A2n
.
..
.
.
.
. . . Ann
=0
(1.25)
Draft
A6
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 }
(1.26)
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.
Victor Saouma
Draft
Appendix B
SOLUTIONS OF LINEAR
EQUATIONS
Note this chapter is incomplete
B.1
Introduction
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}
There are two approaches for this operation:
Direct inversion using Cramers rule where [A]1 = [adjA] . However, this approach is compu[A]
tationally 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
Draft
B2
SOLUTIONS OF LINEAR EQUATIONS
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}.
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.
Example B-1: Gauss Elimination

In this rst example we simply seek to solve for the unknown vector {x} given:
+10x1
+x2 5x3 = 1.
20x1 +3x2 +20x3 = 2.
+5x
+3x2 +5x3 = 6.
1
(2.1-a)
Solution:
1. Add 20 times the rst equation to the second one will elliminate the x1 coecient from
10
the second equation.
5
2. Substract 10 times the rst equation from the third one will elliminate the x1 coecient
from the third equation.
10.x1
+x2
5.x3 = 1.
+5.x2 +10.x3 = 4.
+2.5x2 +7.5x3 = 5.5
3. Substract 2.5 times the second equation from the third one will elliminate the x2 coecient
5
from the last equation
10.x1
+x2
5.x3 = 1.
+5.x2 +10x3 = 4.
+2.5x3 = 3.5
4. Now we can backsubstitute and solve from the bottom up:

x3 =
x2 =
x1 =
Victor Saouma
3.5
= 1.4
2.5
4. 10.x3
= 2.
5.
1. x2 + 5.x3
= 1.
10.
(2.4-a)
(2.4-b)
(2.4-c)
Draft
B.2 Direct Methods
B3
Example B-2: Gauss-Jordan Elimination

In this second example we will determine both {x} and the matrix inverse [A]1 .
Solution:
The operation is identical to the rst, however we augment the matrix [A] by [I]: [A|I], and
operate simultaneously on the two submatrices.
1. Initial matrix
10
1 5 1 0 0 1
2
20 3 20 0 1 0
0 0 1 6
5
3 5
(2.5-a)
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 0.1

0 0 0.1
2
1 0
0 5
10
4
0 2.5 7.5
0.5 0 1 5.5
(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 0.7 0.06 0.02 0 0.02
0.4
0.2
0
0.8
0 1 2
0 0 2.5
1.5 0.5 1 3.5
(2.7-a)
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
1 0 0 0.36 0.16 0.28
0.6
0.8 2
0 1 0 1.6
1.4
0 0 1 0.6 0.2 0.4
(2.8-a)
{x}
This last equation is [I|A1 ]
Victor Saouma
Draft
B4
B.2.1.1
Algorithm
4 Based on the preceding numerical examples, we dene a two step algorithm for the Gaussian
ellimination.
Dening ak to be the coecient of the ith row & j th column at the k th reduction step with
ij
i k & j k:
Reduction:
k<in
ak+1 = 0
ik
ak+1
ij
= ak
ij
bk+1
ij
= bk
ij
ak ak
ik kj
ak
kk
a k bk
ik kj
ak
kk
Backsubstitution:
xij =
bi
ij
k < i n; k < j n
(2.9)
k < i n; 1 < j m
n
ai x
k=i+1 ik kj
ai
ii
(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
3
operations than Gauss or LU decomposition ( N ).
3
B.2.2
LU Decomposition
6 In the previous decomposition method, the right hand side ({b} must have been known before
decomposition (unless we want to detemine the inverse of the matrix which is computationaly
more expensive).
7 In some applications it may be desirable to decompose the matrix without having the RHS
completed. For instance, in the direct stiness method we may have multiple load cases yet we
would like to invert only once the stiness matrix.
This will be achieved through the following decomposition:
[A] = [L][U]
(2.11)
It can be shown that:

1. Both decompositions are equivalent.
2. Count on number of operation show that the 2 methods yield the same number of operations. Number of operations in LU decomposition is equal to the one in Gauss elimination.
8
The solution consists in:
Decomposition: of the matrix independently of the right hand side vector

[A] = [L] [U]
[L] [U] {x} = {b}
(2.12-a)
(2.12-b)
{y}
Backsubstitution: for each right hand side vector

Victor Saouma
Draft
B.2 Direct Methods
B5
1. Solve for {y} from [L]{y} = {b} starting from top

2. Solve for {x} from [U]{x} = {y} starting from bottom
The vector {y} is the same as the one to which {b} was reduced to in the Gauss Elimination.
B.2.2.1
Algorithm
1. Given:
a11
a21
.
.
.
a12
a22
.
.
.
an1 an2
a1n
a2n
.
.
.
.
.
.
ann
1
l21
.
.
.
..
.
1
.
.
.
u11 u12 u1n

u22 u2n
..
.
ln1 ln2 1
(2.13-a)
unn
2. solve:
a11 = u11
a21 = l21 u11
.
.
.
a12 = u12
a22 = l21 u12 + u22
a1n = u1n
a2n = l21 u1n + u2n
n1
an1 = ln1 u11 an2 = ln1 u12 + ln2 u22 ann =

3. let:
[A]F =
u11 u12
l21 u22
.
.
.
ln1
ln2
l u
k=1 nk kn
u1n
u2n
.
.
.
unn
(2.14-a)
+ unn
(2.15-a)
4. Take row by row or column by column
lij =
aij
uij = aij
lii = 1
j1
l u
k=1 ik kj
ujj
i1
l u
k=1 ik kj
i>j
ij
(2.16)
Note:
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
l u
k=1 ik kj
ij
(2.17)
Example B-3: Example

Victor Saouma
Draft
B6
Given:
A=
7 9 1 2
4 5 2 7
1 6 3 4
3 2 1 5
(2.18-a)
Solution:
Following the above procedure, it can be decomposed into:
Row 1: u11 = a11 = 7; u12 = a12 = 9; u13 = a13 = 1; u14 = a14 = 2
Row 2:
l21
u22
u23
u24
a
= u21
11
= a22 l21 u12
= a23 l21 u13
= a24 l21 u14
=4
7
= 5 4 9
7
= 2 + 41
7
= 7 4 2
7
= 10.1429
= 2.5714
= 10.1429
Row 3:
l31
l32
u33
u34
a
= u31
11
31
= a32 l22 u12
u
= a33 l31 u13 l32 u23
= a34 l31 u14 l32 u24
=1
7
= 6(0.1429)(9)
= 0.4647
10.1429
= 3 (0.1429)(1) (0.4647)(2.5714) = 1.6622
= 4 (0.1429)(2) (0.4647)(8.1429) = 8.0698
Row 4:
l41
l42
l43
u44
or
a
= u41
11
41
= a42 l22 u12
u
= a43 l41 u13 l42 u23
u33
= a44 l41 u14 l42 u24 l43 u34
1
0
0
.5714
1
0
.1429 .4647
1
.4286 .5775 1.2371
0
0
0
1
=3
7
= 2(0.4286)(9)
= 0.
10.1429
1(0.4286)(1)(0.5775)(2.5714)
=
= 1.23
1.6622
= 5 (0.4286)(2) (0.5775)(8.1429) (1.2371)(8.0698) = 8.82
7
9
1
2
0 10.1429 2.571 8.143
0
0
1.662 8.069
0
0
0
8.8285
[L]
(2.22-a)
[U]
[A]
B.2.3
10
Choleskys Decomposition
If [A] is symmetric [A]F is not. For example:
16 4
8
1
16 4 8
5 4 = .25
1
4 6
(2.23-a)
4
8 4 22
.5 1.5 1
9
Victor Saouma
Draft
B.2 Direct Methods
11
B7
In the most general case, we will have:

[A] = [L ][D][U ]T
12
(2.24)
For aa symmetric [A] matrix, [U ] should be the transpose of [L ] or

[A] = [L ][D][L ]T
(2.25)
Furthermore, the diagonal matrix [D] can be factored as as the product of two matrices:
1
1
[D] = [D] 2 [D] 2 Thus:
1
1
(2.26)
[A] = [L ][D] 2 [D] 2 [L ]T
13
[L]
14
This algorithm can be summarized as:

aii
lii =
lij =
15
[L]T
aij
ljj
i1 2
l
k=1 ik
j1
k=1
(2.27)
i>j
Note:
1. Decomposition takes place by columns
2. lij will occupy same space as aij
Example B-4: Choleskys Decomposition

Given:
A=
4 6 10 4
6 13 13 6
10 13 27 2
4 6 2 72
(2.28-a)
Solution:
Column 1:
l11
l21
l31
l41
Victor Saouma
=
=
=
=
a11 =
a21
=
l11
a31
=
l11
a41
=
l11
4 =2
=3
=5
=2
6
2
10
2
4
2
Draft
B8
Column 2:
2
a22 l21 =
l22 =
l32 =
l42 =
a32 l31 l21

l22
a42 l41 l21
l22
13 32 = 2
13(5)(3)
2
6(2)(3)
2
=
=
= 1
=0
Column 3:
l33 =
l43 =
2
2
a33 l31 l32 =
a43 l41 l31 l42 l32

l33
27 52 (1)2 = 1
2(2)(5)(0)(1)
1
= 8
Column 4:
l44 =
2
2
2
a44 l41 l42 l43 =
or
2 0
0
3 2
0
5 1 1
2 0 8
0
0
0
2
72 (2)2 (0)2 (8)2 = 2
2
0
0
0
[L]
3 5
2
2 1 0
0 1 8
0 0
2
(2.33-a)
[U]
[A]
B.2.4
B.3
16
Pivoting
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.
B.3.1
Gauss Seidel
c11 x1 + c12 x2 + c13 x3 = r1
c21 x1 + c22 x2 + c23 x3 = r2
c31 x1 + c32 x2 + c33 x3 = r3
(2.34-a)
solve 1st equation for x1 using initial guess for x2 , x3 .

x1 =
Victor Saouma
r1 c12 x2 c13 x3
c11
(2.35)
Draft
B.4 Ill Conditioning
B9
solve 2nd equation for x2 using the computed value of x1 & initial guess of x3
x2 =
so on & so forth
Note:
r2 c21 x1 c23 x3
c22
(2.36)
1. The iterative process can be considered to have converged if:

|
xk xk1
|
xlk
(2.37)
2. The convergence can be accelerated by relaxation

k1
xk = xk + (1 )xi
i
i
(2.38)
where is a weight factor between 0. and 2. For values below 1 we have underrelaxation,
and for values greater than 1 we have overrelaxation. The former is used for nonconvergent
systems, whereas the later is used to accelerate convergence of converging ones. optimum
for frame analysis is around 1.8.
B.4
Ill Conditioning
An ill condition system of linear equations is one in which a small perturbation of the
coecient aij results in large variation in the results x. Such a system arises in attempting
to solve for the intersection of two lines which are nearly parallel, or the decomposition of a
structure stiness matrix in which very sti elements are used next to very soft ones.
18
B.4.1
19
Condition Number
Ill conditioning can be detected by determining the condition number of the matrix.
max
(2.39)
min
and min are the maximum and minimum eigenvalues of the coecient matrix.
=
where max
20 In the decomposition of a matrix, truncation errors may result in a loss of precision which
has been quantied by:
s = p log
(2.40)
where p is the number of decimal places to which the coecient matrix is represented in the
computer, and s is the number of correct decimal places in the solution.
21 Note that because the formula involves log , the eigenvalues need only be approximately
evaluated.
B.4.2
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)
22
However there are no general rules for selecting [D1 ] and [D2 ].
Victor Saouma
Draft
B10
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.
In general 3-D space the number of components of a tensor is 3n where n is the order of
the tensor.
For example, force and a stress are tensors of order 1 and 2 respectively.
To express tensors, there are three distinct notations which can be used: 1) Engineering; 2)
indicial; or 3) Dyadic.
3
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
Dyadic/Vector Notation
Uses bold face characters for tensors of order one and higher, , . This notation is independent of coordinate systems.
5
Draft
C2
6
TENSOR NOTATION
Since scalar operations are in general not applicable to vectors, we dene

A+B = B+A
(3.1-a)
AB = BA
(3.1-b)
A = Ax i + Ay j + Az k
(3.1-c)
AB = |A||B| cos(A, B)
= Ax Bx + Ay By + Az Bz
AB =
grad A =
A =
div A =
A =
=
Laplacian
(3.1-d)
i
j
k
Ax Ay Az
Bx By Bz
A
A
A
+j
+k
i
x
y
z
+j
+k
i
(iAx + jAy + kAz )
x
y
z
Ax Ay
Az
+
+
x
y
z
2A
2A
2A
+
+
x2
y 2
z 2
(3.1-e)
(3.1-f)
(3.1-g)
(3.1-h)
(3.1-i)
C.3
Indicial/Tensorial Notation
This notation uses letter appended indices (sub or super scripts) to the letter representing the
tensor quantity of interest. i.e. ai ; ij ; ij , where the number of indices is the rank of the tensor
(see sect. A.4).
7
The following rules dene tensorial notation:

1. If there is one letter index, that index goes from i to n. For instance:
ai = ai =
a1
a1 a2 a3
2
a
3
i = 1, 3
(3.2)
assuming that n = 3.
2. A repeated index will take on all the values of its range, and the resulting tensors summed.
For instance:
(3.3)
a1i xi = a11 x1 + a12 x2 + a13 x3
3. Tensors order:
First order tensor (such as force) has only one free index:
ai = ai =
Victor Saouma
a1 a2 a3
(3.4)
Draft
C.3 Indicial/Tensorial Notation
C3
Second order tensor (such as stress or strain) will have two free indeces.
D11 D22 D13
Dij D21 D22 D23

D31 D32 D33
(3.5)
A fourth order tensor (such as Elastic constants) will have four free indeces.
4. Derivatives of tensor with respect to xi is written as , i. For example:
xi
= ,i
vi
xi
= vi,i
vi
xj
= vi,j
Ti,j
xk
= Ti,j,k
(3.6)
Usefulness of the indicial notation is in presenting systems of equations in compact form.

For instance:
(3.7)
xi = cij zj
this simple compacted equation (expressed as x = cz in dyadic notation), when expanded would
yield:
x1 = c11 z1 + c12 z2 + c13 z3
x2 = c21 z1 + c22 z2 + c23 z3
(3.8-a)
x3 = c31 z1 + c32 z2 + c33 z3

Similarly:
Aij = Bip Cjq Dpq
(3.9)
A11 = B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
A12 = B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
A21 = B21 C11 D11 + B21 C12 D12 + B22 C11 D21 + B22 C12 D22
A22 = B21 C21 D11 + B21 C22 D12 + B22 C21 D21 + B22 C22 D22
Victor Saouma
(3.10-a)
Draft
C4
Victor Saouma
TENSOR NOTATION
Draft
Appendix D
INTEGRAL THEOREMS
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.
1
D.1
Integration by Parts
The integration by part formula is

b
a
u(x)v (x)dx = u(x)v(x)|b

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

divvd
v.nd =
(4.4)
vi,i d
(4.5)
or
vi ni d =
Draft
D2
3
INTEGRAL THEOREMS
In 2D-3D Gauss integral theorem is

qT .ndS
div qdV =
V
(4.6)
or
V
vi,i dV =
(4.7)
vi ni dS
Alternatively
div qdV =
V
qT .ndS
( )T qdV
(4.8)
For 2D-1D transformations, we have

qT nds
div qdA =
A
(4.9)
or
div qdA =
A
Victor Saouma
qT nds
( )T qdA
(4.10)

V, Saouma FEM I PDF

Uploaded by

Copyright:

Available Formats

V, Saouma FEM I PDF

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

V, Saouma FEM I PDF

Uploaded by

Copyright:

Available Formats

Draft

Dept. of Civil Environmental and Architectural Engineering

Finite Element I; Framed Structures

Matrix Structural Analysis of Framed Structures

2 ELEMENT STIFFNESS MATRIX

3 STIFFNESS METHOD; Part I: ORTHOGONAL STRUCTURES

5 STIFFNESS METHOD; Part II

Finite Element I; Framed Structures

6 EQUATIONS OF STATICS and KINEMATICS

Finite Element I; Framed Structures

Stiness Matrix of a Curved Element . . . . . . . . . . . . . . . . . . . . . . . . . 79

8 SPECIAL ANALYSIS PROCEDURES

Introduction to Finite Elements

10 VARIATIONAL AND ENERGY METHODS

Finite Element I; Framed Structures

10.1.7.1 Potential Functions . . . . . . . . . . . . . . . . . . . . . . . . . 1011

Finite Element I; Framed Structures

11.3.2 C 1 : Hermitian Interpolation Functions . . . . . . . . . . . . . . . . . . . . 1110

13 SOME FINITE ELEMENTS

Finite Element I; Framed Structures

15.1.2.1 Strain Energy . . . . . . . . . . . . . .

B SOLUTIONS OF LINEAR EQUATIONS

Finite Element I; Framed Structures

Finite Element I; Framed Structures

Global Coordinate System . . . . . . . . . . . . . . . . .

Example for Flexibility Method . . . . . . . . . . . . . . . . . . . . .

Problem with 2 Global d.o.f. 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . 32

Arbitrary 3D Vector Transformation . . . . . . . . . . .

of [B] Matrix for a Statically Determinate Truss .

Stable and Statically Determinate Element . . . . . . . . . . . . . . . . . . . . . 75

Flexible, Rigid, and Semi-Rigid Beams . . . . . . . . . . . . . . . . . . . . . . . . 81

Stress Components on an Innitesimal Element

10.1 Strain Energy and Complementary Strain Energy . . . . . . . . . . . . . . . . . . 102

Axial Finite Element . . . . . . . . . . . . . . . . . . . .

Finite Element I; Framed Structures

11.6 Solid Trilinear Rectangular Element . . . . . . . . . . . . . . . . . . . . . . . . . 1110

Finite Element I; Framed Structures

Finite Element I; Framed Structures

Example of Nodal Denition .

Examples of Inuence Coecients . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3D Transformations of Linear Elements . . . . . . . . . . . . . . . . . . . . . . . 47

Internal Element Force Denition for the Statics Matrix . . . . . . . . . . . . . . 62

Possible Combinations of Real and Hypothetical Formulations . .

11.1 Characteristics of Beam Element Shape Functions . . . . . . . . . . . . . . . . . 115

Finite Element I; Framed Structures

Vector of coeccients in assumed displacement eld

lc: Local Coordinate system

Finite Element I; Framed Structures

Why Matrix Structural Analysis?

Those students who want to pursue a specialization in structural engineering/mechanics, do

4 Matrix Structural Analysis, or Advanced Structural Analysis, or Introduction to Structural

and three dimensional continuum are essential in civil engineering to model

encountered in continuum nite elements.

Overview of Structural Analysis

1D framework (truss, beam, columns)

Finite Element I; Framed Structures

1.3 Structural Idealization

(d) Sectional properties: