Draft: Finite Element I

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

2.1 Introduction . . . . . . . . . . . . . . . .
2.2 Influence Coefficients . . . . . . . . . . .
2.3 Flexibility Matrix (Review) . . . . . . .
2.4 Stiffness Coefficients . . . . . . . . . . .
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 Stiffness Matrices .
2.8 Homework . . . . . . . . . . . . . . . . .
3 STIFFNESS METHOD;
3.1 Introduction . . . . . .
3.2 The Stiffness Method .
3.3 Examples . . . . . . .
E 3-1 Beam . . . . .
E 3-2 Frame . . . . .
E 3-3 Grid . . . . . .
3.4 Observations . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Part I: ORTHOGONAL
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
17
17
18
19
20
21
21
22
25
27
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
29
29
30
31
32
32
32
35
36
39
39
40
40
40
42
42
43
43
45
STRUCTURES
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
47
47
47
49
49
51
54
56
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
111
111
112
113
115
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 . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
117
117
117
117
119
120
120
122
124
124
125
125
126
126
127
127
128
130
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 Stiffness Flexibility Relations . . . . . . . . . . . . . . . .
7.3.1 From Stiffness to Flexibility . . . . . . . . . . . . .
E 7-2 Flexibility Matrix . . . . . . . . . . . . . . . . . .
7.3.2 From Flexibility to Stiffness . . . . . . . . . . . . .
E 7-3 Flexibility to Stiffness . . . . . . . . . . . . . . . .
7.4 Stiffness Matrix of a Curved Element . . . . . . . . . . . .
7.5 Duality between the Flexibility and the Stiffness Methods
7.6 Homework . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
133
133
133
134
135
135
135
137
137
137
138
139
140
141
142
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 . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
143
143
145
145
145
145
145
145
5.7
5.6.2.11
5.6.2.12
5.6.2.13
5.6.2.14
Homework . . .
Victor Saouma
Nodal Displacements
Reactions . . . . . . .
Internal Forces . . . .
Sample Output File .
. . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Finite Element I; Framed Structures
Draft
CONTENTS
E 10-8 Tapered Beam; Fourrier Series . . . . . . . . . . . . . . .

10.4 Complementary Potential Energy . . . . . . . . . . . . . . . . .
10.4.1 Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4.2 Castiglianos Second Theorem . . . . . . . . . . . . . . . .
E 10-9 Cantilivered beam . . . . . . . . . . . . . . . . . . . . . .
10.4.2.1 Distributed Loads . . . . . . . . . . . . . . . . .
E 10-10Deflection of a Uniformly loaded Beam using Castiglianos
10.5 Comparison of Alternate Approximate Solutions . . . . . . . . .
E 10-11Comparison of MPE Solutions . . . . . . . . . . . . . . .
10.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.7 Variational Calculus; Preliminaries . . . . . . . . . . . . . . . . .
10.7.1 Euler Equation . . . . . . . . . . . . . . . . . . . . . . . .
10.7.2 Boundary Conditions . . . . . . . . . . . . . . . . . . . .
E 10-12Extension of a Bar . . . . . . . . . . . . . . . . . . . . . .
E 10-13Flexure of a Beam . . . . . . . . . . . . . . . . . . . . . .
10.8 Homework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
second Theorem
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
186
188
188
188
189
189
189
190
190
191
191
194
197
198
199
201
11 INTERPOLATION FUNCTIONS
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2 Shape Functions . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2.1 Axial/Torsional . . . . . . . . . . . . . . . . . . . . . . . .
11.2.2 Generalization . . . . . . . . . . . . . . . . . . . . . . . .
11.2.3 Flexural . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2.4 Constant Strain Triangle Element . . . . . . . . . . . . .
11.3 Interpolation Functions . . . . . . . . . . . . . . . . . . . . . . .
11.3.1 C 0 : Lagrangian Interpolation Functions . . . . . . . . . .
11.3.1.1 Constant Strain Quadrilateral Element . . . . .
11.3.1.2 Solid Rectangular Trilinear Element . . . . . . .
11.3.2 C 1 : Hermitian Interpolation Functions . . . . . . . . . . .
11.4 Interpretation of Shape Functions in Terms of Polynomial Series
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
203
203
203
204
205
205
207
209
209
210
211
211
212
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 Stiffness Matrix Formulation
12.3.1 Stress Recovery . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
213
213
213
214
214
214
216
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 Stiffness Matrix . . . . . . . . .
13.4.3 Internal Stresses . . . . . . . .
13.4.4 Observations . . . . . . . . . .
13.5 Quadrilateral Element . . . . . . . . .
13.6 Homework . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
217
217
217
218
218
218
219
220
220
220
221
Victor Saouma
Draft
CONTENTS
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
9
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
269
270
270
271
271
272
272
272
273
273
C TENSOR NOTATION
275
C.1 Engineering Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
C.2 Dyadic/Vector Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
C.3 Indicial/Tensorial Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
D INTEGRAL THEOREMS
279
D.1 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
D.2 Green-Gradient Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
D.3 Gauss-Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Victor Saouma
Draft
List of Figures
1.1
1.2
1.3
1.4
1.5
1.6
1.7
Global Coordinate System . . . . . . . . . . . . . . . . .

Local Coordinate Systems . . . . . . . . . . . . . . . . .
Sign Convention, Design and Analysis . . . . . . . . . .
Total Degrees of Freedom for various Type of Elements
Independent Displacements . . . . . . . . . . . . . . . .
Examples of Global Degrees of Freedom . . . . . . . . .
Organization of the Course . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
21
22
22
23
24
26
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
Example for Flexibility Method . . . . . . . . . . . . . . . . . . . . .

Definition of Element Stiffness Coefficients . . . . . . . . . . . . . . .
Stiffness Coefficients for One Dimensional Elements . . . . . . . . . .
Flexural Problem Formulation . . . . . . . . . . . . . . . . . . . . . .
Torsion Rotation Relations . . . . . . . . . . . . . . . . . . . . . . .
Deformation of an Infinitesimal Element Due to Shear . . . . . . . .
Effect of Flexure and Shear Deformation on Translation at One End
Effect of Flexure and Shear Deformation on Rotation at One End . .
Coordinate System for Element Stiffness Matrices . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
30
33
34
34
35
37
38
39
40
3.1
3.2
3.3
Problem with 2 Global d.o.f. 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Frame Example (correct K23 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Grid Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
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 . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
60
61
62
63
64
65
66
67
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
Frame Example . . . . . . . . . . . . . . . . . . . .
Example for [ID] Matrix Determination . . . . . .
Simple Frame Analyzed with the MATLAB Code .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
Simple Frame Analyzed with the MATLAB Code .
Stiffness Analysis of one Element Structure . . . .
Example of Bandwidth . . . . . . . . . . . . . . . .
Numbering Schemes for Simple Structure . . . . .
Program Flowchart . . . . . . . . . . . . . . . . . .
Programs Tree Structure . . . . . . . . . . . . . .
Flowchart for the Skyline Height Determination . .
Flowchart for the Global Stiffness Matrix Assembly
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
69
72
73
75
80
82
86
87
91
92
94
96
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
List of Tables
1.1
1.2
1.3
1.4
Example of Nodal Definition .

Example of Element Definition
Example of Group Number . .
Degrees of Freedom of Different
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
Structure Types Systems
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2.1
Examples of Influence Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1
3D Transformations of Linear Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.1
6.2
Internal Element Force Definition for the Statics Matrix . . . . . . . . . . . . . . . . . . . 118

Conditions for Static Determinacy, and Kinematic Instability . . . . . . . . . . . . . . . . 124
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
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
20
20
21
23
167
191
194
198
11.1 Characteristics of Beam Element Shape Functions . . . . . . . . . . . . . . . . . . . . . . 208

11.2 Interpretation of Shape Functions in Terms of Polynomial Series (1D & 2D) . . . . . . . . 212
11.3 Polynomial Terms in Various Element Formulations (1D & 2D) . . . . . . . . . . . . . . . 212
Draft
LIST OF TABLES
a
A
A
b
B
[B0 ]
[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
b
t
u
u
b (x)
u
u, v, w
U
Victor Saouma
15
NOTATION
Vector of coefficcients in assumed displacement field
Area
Kinematics Matrix
Body force vector
Statics Matrix, relating external nodal forces to internal forces
Statics Matrix relating nodal load to internal forces p = [B0 ]P
Matrix relating assumed displacement fields parameters to joint displacements
Cosine
Matrices derived from the statics matrix
Element flexibility matrix (lc)
Structure flexibility 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 field parameters
to joint displacements
Idendity matrix
Matrix relating nodal dof to structure dof
St Venants torsional constant
Element stiffness matrix (lc)
Matrix of coefficients of a polynomial series
Geometric element stiffness matrix (lc)
Rotational stiffness matrix ( [d] inverse )
Structure stiffness matrix (GC)
Structures geometric stiffness matrix (GC)
Length
Linear differential 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
Specified tractions along t
Displacement vector
Neighbour function to u(x)
Specified displacements along u
Translational displacements along the x, y, and z directions
Strain energy
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, deflections, and internal forces) of both statically determinate
and indeterminate structures.
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.
2
3 Those students who want to pursue a specialization in structural engineering/mechanics, do take more
advanced courses such as Matrix Structural Analysis and/or Finite Element Analysis.
Matrix Structural Analysis, or Advanced Structural Analysis, or Introduction to Structural Engineering Finite Element, builds on the introductory analysis course to focus on those methods which lend
themselves to computer implementation. In doing so, we will place equal emphasis on both two and three
dimensional structures, and develop a thorough understanding of computer aided analysis of structures.
4
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 specific tasks,
and most importantly to understand their limitations.
6 With the recently placed emphasis on the finite element method in most graduate schools, many
students have been tempted to skip a course such as this one and rush into a finite element one. Hence
it is important that you understand the connection and role of those two courses. The Finite Element
Method addresses the analysis of two or three dimensional continuum. As such, the primary unknowns
is u the nodal displacements, and internal forces are usually restricted to stress . The only analogous
one dimensional structure is the truss.
7 Whereas two and three dimensional continuum are essential in civil engineering to model structures
such as dams, shells, and foundation, the majority of Civil engineering structures are constituted by rod
one-dimensional elements such as beams, girders, or columns. For those elements, displacements and
internal forces are somehow more complex than those encountered in continuum finite elements.
8 Hence, contrarily to continuum finite 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 flexural and/or torsional rotation. Similarly, internal forces are not
stresses, but rather axial and shear forces, and/or flexural or torsional moments. Those concepts are far
Draft
1.3 Structural Idealization
19
3. Type of solution:
(a) Continuum, analytical, Partial Differential Equation
(b) Discrete, numerical, Finite ELement, Finite Difference, Boundary Element
13
Structural design must satisfy:

1. Strength ( < f )
2. Stiffness (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
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:
15
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 flexural stiffness?
3. Rigid or semi-rigid connections (most important in steel structures)
4. Rigid supports or elastic foundations (are the foundations over solid rock, or over clay which may
consolidate over time)
5. Include or not secondary members (such as diagonal braces in a three dimensional analysis).
6. Include or not axial deformation (can we neglect the axial stiffness 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 amplified by the product of the axial load times the lateral deformation, P
effects).
11. Time dependent effects (such as creep, which is extremely important in prestressed concrete, or
cable stayed concrete bridges).
12. Partial collapse or local yielding (would the failure of a single element trigger the failure of the
entire structure?).
Victor Saouma
Draft
1.3 Structural Idealization
21
Group
No.
1
2
3
Element
Type
1
2
1
Material
Group
1
1
2
Table 1.3: Example of Group Number
X
Z
2D TRUSS
FRAME
BEAM
3D TRUSS
GRID & FRAME
Figure 1.1: Global Coordinate System
1.3.2
22
Coordinate Systems
We should differentiate between 2 coordinate systems:
Global: to describe the structure nodal coordinates. This system can be arbitrarily selected provided
it is a Right Hand Side (RHS) one, and we will associate with it upper case axis labels, X, Y, Z,
Fig. 1.1 or 1,2,3 (running indeces within a computer program).
Local: system is associated with each element and is used to describe the element internal forces. We
will associate with it lower case axis labels, x, y, z (or 1,2,3), Fig. 1.2.
23 The x-axis is assumed to be along the member, and the direction is chosen such that it points from
the 1st node to the 2nd node, Fig. 1.2.
24
Two dimensional structures will be defined in the X-Y plane.
1.3.3
Sign Convention
25 The sign convention in structural analysis is completely different than the one previously adopted in
structural analysis/design, Fig. 1.3 (where we focused mostly on flexure and defined a positive moment
as one causing tension below. This would be awkward to program!).
y, 2
x, 1
BEAM, TRUSS
x, 1
z, 3
GRID, FRAME
Figure 1.2: Local Coordinate Systems

Victor Saouma
Draft
1.4 Degrees of Freedom
23
Figure 1.5: Independent Displacements

various types of structures made up of one dimensional rod elements, Table 1.4.
34
This table shows the degree of freedoms and the corresponding generalized forces.
Type
Node 1
Node 2
Fy1 , Mz2
{}
v1 , 2
{p}
Fx1
Truss
u1
Fx1 , Fy2 , Mz3
u2
Fx4 , Fy5 , Mz6
Frame
{}
{p}
u 1 , v2 , 3
Tx1 , Fy2 , Mz3
u 4 , v5 , 6
Tx4 , Fy5 , Mz6
Grid
{}
{p}
{}
1 , v2 , 3
{p}
Fx1 ,
{}
{p}
u1 ,
Fx1 , Fy2 , Fy3 ,
Tx4 My5 , Mz6
u2
Fx7 , Fy8 , Fy9 ,
Tx10 My11 , Mz12
{}
u 1 , v2 , w 3 ,
4 , 5 6
u 7 , v8 , w 9 ,
10 , 11 12
Truss
[K]
(Global)
44
44
22
44
66
66
66
66
22
66
12 12
12 12
1 Dimensional
Fy3 , Mz4
{p}
Beam
[k]
(Local)
v3 , 4
2 Dimensional
Fx2
4 , v5 , 6
3 Dimensional
Fx2
Frame
Table 1.4: Degrees of Freedom of Different Structure Types Systems

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

Framed Structures
Draft
Chapter 2
ELEMENT STIFFNESS MATRIX

2.1
Introduction
1 In this chapter, we shall derive the element stiffness matrix [k] of various one dimensional elements.
Only after this important step is well understood, we could expand the theory and introduce the structure
stiffness matrix [K] in its global coordinate system.
2 As will be seen later, there are two fundamentally different approaches to derive the stiffness matrix of
one dimensional element. The first 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
stiffness matrix, we will be reviewing concepts earlier seen.
The other approach, based on energy consideration through the use of assumed shape functions, will
be examined in chapter 12. This second approach, exclusively used in the finite element method, will
also be extended to two and three dimensional continuum elements.
3
2.2
Influence Coefficients
4 In structural analysis an influence coefficient C ij can be defined as the effect 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 Influence Coefficients are
shown in Table 2.1.
Influence Line
Influence Line
Influence Line
Flexibility Coefficient
Stiffness Coefficient
Unit Action
Load
Load
Load
Load
Displacement
Effect on
Shear
Moment
Deflection
Displacement
Load
Table 2.1: Examples of Influence Coefficients

5 It should be recalled that influence lines are associated with the analysis of structures subjected to
moving loads (such as bridges), and that the flexibility and stiffness coefficients are components of
matrices used in structural analysis.
Draft
2.4 Stiffness Coefficients
31
Virtual Force:
Z
x x dvol
M xy
Z l
x =
I
M
My
x
dx
M
U =
x = E = EI
Z
EI
0
y 2 dA = I
dvol = dAdx
W = P
U = W
U
Hence:
EI |{z}
1 d11 =
|{z}
Similarly, we would obtain:

EId22
EId12
=
=
L
0
L
0
L
0
x 2
M
0
M
dx = P
EI
L
x 2
dx =
1
L
3
{z
}
(2.3)
(2.4)
M M
dx =
L
3
(2.5-a)
x x
L
dx =
L L
6
EId21
(2.5-b)
Those results can be summarized in a matrix form as:

[d] =
L
6EIz
2
1
1
2
(2.6)
The flexibility method will be covered in more detailed, in chapter 7.
2.4
Stiffness Coefficients
In the flexibility method, we have applied a unit force at a time and determined all the induced
displacements in the statically determinate structure.
10
In the stiffness method, we

1. Constrain all the degrees of freedom
2. Apply a unit displacement at each d.o.f. (while restraining all others to be zero)
3. Determine the reactions associated with all the d.o.f.
{p} = [k]{}
(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
Victor Saouma
Draft
2.5 Force-Displacement Relations
33
Figure 2.2: Definition of Element Stiffness Coefficients
Victor Saouma
Draft
2.7 Remarks on Element Stiffness Matrices
43
Upon substitution, the grid element stiffness matrix is given by

1x
x
T1x GI
L
V1y
0
M1z
0
Gi
x
T2x
L
V2y
0
M2z 0
[kg ] =
u1y
0
1z
0
12EIz
L3
6EIz
L2
6EIz
L2
4EIz
L
z
12EI
L3
z
6EI
L2
6EIz
L2
2x
x
GI
L
0
0
z
12EI
L3
6EIz
L2
0
GIx
L
12EIz
L3
z
6EI
L2
0
0
2EIz
L3
2z
0
6EIz
L2
2EIz
z
6EI
L2
4EIz
u2y
0
(2.46)
Note that if shear deformations must be accounted for, the entries corresponding to shear and flexure
must be modified in accordance with Eq. 2.42
52
2.6.5
3D Frame Element
u1
t
Px1 k11
Vy1 0
V z1
0
Tx1
0
My1
0
Mz1
0t
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
For [k3D
11 ] and with we obtain:
Px1
Vz1
0
Tx1
0
0
My1
0
Mz1
Px2 EA
L
Vy2
0
Vz2
0
Tx2
0
0
My2
Vy1
[k
3df r
] =
u1
EA
l
0
Mz2
v1
w1
x1
y1
z1
u2
v2
w2
x2
y2
z2
0
12EIz
L3
0
6EIy
L2
0
12EIz
L3
EA
L
0
0
12EIy
L3
0
6EIz
L2
0
12EIy
L3
0
EIy
6
L2
0
6EIz
L2
0
0
0
6EIz
L2
0
12EIz
L3
0
0
0
6EIz
L2
0
6EIy
L2
0
GIx
L
0
0
4EIy
L
0
12EIy
L3
0
EIy
6
L2
0
6EIy
L2
0
0
GIx
L
0
0
0
2EIy
L
0
0
4EIz
L
0
6EIz
L2
0
0
EA
L
0
0
6EIz
L2
0
12EIz
L3
0
2EIz
L
0
6EIz
L2
0
GIx
L
0
0
2EIy
L
0
12EIy
L3
0
6EIy
L2
0
6EIy
L2
0
6EIy
L2
0
0
GIx
L
0
0
0
4EIy
L
0
0
0
0
2EIz
L
0
6EIz
L2
0
0
0
4EIz
L
(2.47)
(2.48)
Note that if shear deformations must be accounted for, the entries corresponding to shear and flexure
must be modified in accordance with Eq. 2.42
53
2.7
Remarks on Element Stiffness Matrices
Singularity: All the derived stiffness matrices are singular, that is there is at least one row and one
column which is a linear combination of others. For example in the beam element, row 4 = row
Victor Saouma
Draft
2.8 Homework
2.8
45
Homework
Using the virtual force method, derive the flexibility 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 flexibility matrix with equilibrium relations to
derive the element stiffness matrix.
x
z
Victor Saouma
Draft
Chapter 3
STIFFNESS METHOD; Part I:

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

{}
Note that we use the same matrix

one).
6
= [k(e) ]{}
= [K(e) ]{}
(e)
{p}
=
=
[(e) ]{}
[
(e)
]{P}
(4.3)
(4.4)
since both {} and {p} are vector quantities (or tensors of order
Substituting Eqn. 4.3 and Eqn. 4.4 into Eqn. 4.1 we obtain
[(e) ]{P} = [k(e) ][(e) ]{}
(4.5)
{P} = [(e) ]1 [k(e) ][(e) ]{}
(4.6)
premultiplying by [(e) ]1
But since the rotation matrix is orthogonal, we have [(e) ]1 = [(e) ]T and
{P} = [(e) ]T [k(e) ][(e) ]{}
{z
}
|
(4.7)
[K(e) ] = [(e) ]T [k(e) ][(e) ]
(4.8)
[K(e) ]
which is the general relationship between element stiffness matrix in local and global coordinates.
Draft
4.1 Preliminaries
61
where lij is the direction cosine of axis i with respect to axis j, and thus the rows of the matrix correspond
to the rotated vectors with respect to the original ones corresponding to the columns.
11
With respect to Fig. 4.2, lxX = cos ; lxY = cos , and lxZ = cos or
VY
VX
VZ
Z
Figure 4.2: 3D Vector Transformation

Vx
12
VX lxX + VY lxY + VZ lxZ
(4.14-a)
VX cos + VY cos + VZ cos
(4.14-b)
Direction cosines are unit orthogonal vectors satisfying the following relations:
3
X
lij lij = 1
i = 1, 2, 3
(4.15)
j=1
i.e:
2
2
2
= 1 or cos2 + cos2 + cos2 = 1 = 11
+ l13
+ l12
l11
and
3
X
lij lkj
j=1
l11 l21 + l12 l22 + l13 l23
(4.16)
i = 1, 2, 3
k = 1, 2, 3
= 0
i 6= k
(4.17-a)
= 0 = 12
(4.17-b)
By direct multiplication of [] and [] it can be shown that: [] [] = [I] [] = []

an orthogonal matrix.
13
14
[] is
The reverse transformation (from local to global) would be

T
{V} = [] {v}
or
Victor Saouma
(4.18)
lxX lyX lzX Vx

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

5.1
Direct Stiffness Method
5.1.1
Global Stiffness Matrix
1 The physical interpretation of the global stiffness matrix K is analogous to the one of the element, i.e.
If all degrees of freedom are restrained, then Kij corresponds to the force along global degree of freedom
i due to a unit positive displacement (or rotation) along global degree of freedom j.
For instance, with reference to Fig. 5.1, we have three global degrees of freedom, 1 , 2 , and 3 . and
0/0
/
EI
M
B
P/2
@?@
L/2
)#)#))**
<<
A
B
>#= >=
&%&
A
B
J#I JI
$#" C$#" $"
B#B#B 1 D#C D#C DC

A#A#A
K21
77
K22
K31
,#+,#+ ,+,+
775#6#5 6#5 65
K32
K23
43
4343
K12
B
K11
F#EC FE
L/2
<<'#(#' (#' ('
21
2121
9#8 98
Figure 5.1: Frame Example
-#.#- -.#. .
G#H#GC HG
;#: ;#: ;:
1
K 33
K13
Draft
5.2 Logistics
71
(e)
at the element level where pint is the six by six array of internal forces, k(e) the element stiffness 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 first 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
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.
10
11
To obtain the global degree of freedom for a given node, we need to define 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:

0
if unrestrained d.o.f.
ID(idof, inod) =
1
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 file may contain:
[ID]T
000
110
000
100
Node No.
1
2
3
4
2. At this stage, the [ID] matrix is equal to:
0
ID = 0
0
1
1
0
0
0
0
1
0
0
3. After we determined the equation numbers, we would have:
1 10 5 12
8
ID = 2 11 6
3
4
7
9
Victor Saouma
(5.7)
(5.8)
Draft
5.2 Logistics
73
[KS ](N EQAN EQA ) to zero, and then loop through all the elements, and then through each entry of
(e)
the respective element stiffness matrix Kij .

(e)
The assignment of the element stiffness matrix term Kij (note that e, i, and j are all known since
we are looping on e from 1 to the number of elements, and then looping on the rows and columns of the
S
element stiffness matrix i, j) into the global stiffness matrix Kkl
is made through the LM vector (note
that it is k and l which must be determined).
20
21 Since the global stiffness 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 stiffness matrix.
Example 5-1: Assembly of the Global Stiffness Matrix
As an example, let us consider the frame shown in Fig. 5.3.
50kN
8m
KLKLMLKMLK MLKMLK MKMK

KLMLK MLK MK
OON
4 kN/m
3m
7.416 m
8m
Figure 5.3: Simple Frame Analyzed with the MATLAB Code

The ID matrix is initially set to:
1
[ID] = 1
1
0
0
0
1
1
1
(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 defined from node 1 to node
2, and element 2 from node 2 to node 3):

4 5 6 1
2
3
[LM ] =
(5.11)
1
2
3 7 8 9
Let us simplify the operation by designating the element stiffness matrices in global coordinates as
follows:
K (1)
Victor Saouma
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
(5.12-a)
Draft
5.2 Logistics
75
5. Backsubstitute and obtain nodal displacements in global coordinate system.
6. Solve for the reactions, Eq. 5.3.

7. For each element, transform its nodal displacement from global to local coordinates {} =
[(e) ]{}, and determine the internal forces [p] = [k]{}.
24
Some of the prescribed steps are further discussed in the next sections.
Example 5-2: Direct Stiffness Analysis of a Truss
Using the direct stiffness method, analyze the truss shown in Fig. 5.4.
4
5
1
6
8
50k
12
100k
16
16
Figure 5.4:
Solution:
1. Determine the structure ID matrix
Node #
Bound.
X
0
0
1
0
0
1
2
3
4
5
ID
0
1
0
0
1
1
N ode 1

1
8
0
0
2
2
3
Cond.
Y
1
0
1
0
0
0
0
3
9
10
(5.14-a)
4 5

4 6
5 7
(5.14-b)
2. The LM vector of each element is evaluated next

bLM c1
bLM c2
bLM c3
bLM c4
Victor Saouma
b 1 8 4 5 c
= b 1 8 2 3 c
= b 2 3 4 5 c
= b 4 5 6 7 c
=
(5.15-a)
(5.15-b)
(5.15-c)
(5.15-d)
Draft
5.2 Logistics
81
% FEA_Rest for all the restrained nodes

FEA_Rest=[0,0,0,FEA(4:6,2)];
% Assemble the load vector for the unrestrained node
P(1)=50*3/8;P(2)=-50*7.416/8-FEA(2,2);P(3)=-FEA(3,2);
% Solve for the Displacements in meters and radians
Displacements=inv(Ktt)*P
% Extract Kut
Kut=Kaug(4:9,1:3);
% Compute the Reactions and do not forget to add fixed end actions
Reactions=Kut*Displacements+FEA_Rest
% Solve for the internal forces and do not forget to include the fixed end actions
dis_global(:,:,1)=[0,0,0,Displacements(1:3)];
dis_global(:,:,2)=[Displacements(1:3),0,0,0]; for elem=1:2
dis_local=Gamma(:,:,elem)*dis_global(:,:,elem);
int_forces=k(:,:,elem)*dis_local+fea(1:6,elem)
end
function [k,K,Gamma]=stiff(EE,II,A,i,j)
% Determine the length
L=sqrt((j(2)-i(2))^2+(j(1)-i(1))^2);
% Compute the angle theta (careful with vertical members!)
if(j(1)-i(1))~=0
alpha=atan((j(2)-i(2))/(j(1)-i(1)));
else
alpha=-pi/2;
end
% form rotation matrix Gamma
Gamma=[ cos(alpha) sin(alpha) 0 0
0
0; -sin(alpha)
cos(alpha) 0 0
0
0; 0
0
1 0
0; 0
0
0 cos(alpha) sin(alpha) 0; 0
0
0 -sin(alpha) cos(alpha) 0; 0
0
0 0
0
1];
% form element stiffness matrix in local coordinate system
EI=EE*II; EA=EE*A; k=[EA/L,
0,
0, -EA/L,
0,
0,
12*EI/L^3, 6*EI/L^2,
0, -12*EI/L^3, 6*EI/L^2;
0,
6*EI/L^2,
4*EI/L,
0, -6*EI/L^2,
2*EI/L;
-EA/L,
0,
0, EA/L,
0,
0;
0, -12*EI/L^3, -6*EI/L^2,
0, 12*EI/L^3, -6*EI/L^2;
0,
6*EI/L^2,
2*EI/L,
0, -6*EI/L^2,
4*EI/L];
% Element stiffness matrix in global coordinate system
K=Gamma*k*Gamma;
0;
This simple proigram will produce the following results:

Displacements =
0.0010
-0.0050
-0.0005
Reactions =
130.4973
55.6766
13.3742
Victor Saouma
Draft
5.6 Computer Implementation with MATLAB
105
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
%************************************************************************************************
%SCRIPTFILE NAME:
IDRASMBL.M
%
%MAIN FILE
:
CASAP
%
%Description
:
This file re-assambles the ID matrix such that the restrained
%
degrees of freedom are given negative values and the unrestrained
%
degrees of freedom are given incremental values beginning with one
%
and ending with the total number of unrestrained degrees of freedom.
%
%
%************************************************************************************************
%
TAKE CARE OF SOME INITIAL BUSINESS: TRANSPOSE THE PNODS ARRAY
Pnods=Pnods.;
%
SET THE COUNTER TO ZERO
count=1;
negcount=-1;
%
REASSEMBLE THE ID MATRIX
if istrtp==3
ndofpn=3;
nterm=6;
else
error(Incorrect structure type specified)
end
%
SET THE ORIGINAL ID MATRIX TO TEMP MATRIX

orig_ID=ID;
%
%
REASSEMBLE THE ID MATRIX, SUBSTITUTING RESTRAINED DEGREES OF FREEDOM WITH NEGATIVES,

AND NUMBERING GLOBAL DEGREES OF FREEDOM
for inode=1:npoin
XXXXXXXXXXXXXXXXXXXXXX COMPLETE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
%
END OF IDRASMBL.M SCRIPTFILE
5.6.2.3
Element Nodal Coordinates
%**********************************************************************************************
Victor Saouma
Draft
6.1 Statics Matrix [B]
119
the unknown forces and reactions

0 0
F1
0 0

0 0
F3

0 0
F4
=
1 0
R
x1

0 1
y1
0 0
Rx4
Ry4
0 0
| {z } |
{F}
can be determined through inversion of [B]:

Py2
0
1
0
0 0 0
0
P
1 0
0
0 0 0
0
x2

1
1
Px2
0
0
0
0
P
x2
C
C
C

S
S
S
0 C
1 0 0
C
Px2
C
Py2
=
0
1
0
1
0 0 0
P
x2
S
S
S
1
0 0 0
C
C
C Px2 + Py2
0
0
0
0
0 1 0
0

S
S
S
0
0 C
1 0 1
C
C
Px2
{z
} | {z }
{P}
[B]1
(6.3)
We observe that the matrix [B] is totally independent of the external load, and once inverted can be
used for multiple load cases with minimal computational efforts.
Example 6-2: Beam Statics Matrix

Considering the beam shown in Fig. 6.2, we have 3 elements, each with 2 internalunknowns (v and
m) plus two unknown reactions, for a total of 8 unknowns. To solve for those unknowns we have 2
equations of equilibrium at each of the 4 nodes. Note that in this problem we have selected as primary
unknowns the shear and moment at the right end of each element. The left components can be recovered
from equilibrium. From equilibrium we thus have:
Figure 6.2: Example of [B] Matrix for a Statically Determinate Beam
P1
M1
P2
M2
P3
P4
M4
| {z
{P}
Victor Saouma
} |
1
8
1
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
1
2
1
0
0
0
0
0
0
0
0
0
1 0
0 1
1 3
0
1
0
0
{z
[B]
0
0
0
0
0
1
0
1
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
}|
v1
m1
v2
m2
v3
m3
R1
R2
{z
{F}
(6.4)
Draft
121
Px2
Px2
f1
Py3
Py2
Py2
Rx1
5
X
f5
f3
3
f4
f5 f
4
Py4
Py1 f1
f2
f2
f3
Px1
Rx4
Px3
Px4
4
R y4
R y1
Figure 6.3: Example of [B] Matrix for a Statically Indeterminate Truss
We can solve for the internal forces in terms of the (still unknown) redundant force
{F0 }
F1
F2
F3
F4
Rx1
Ry1
Rx4
Ry4
| {z
{F0 }
Or using the
[B0 ]1 {P} [B0 ]1 [Bx ] {Fx }
0
0
0
0
1
0
0
0
0
0
0 1
1
0
C
S
0 C
0
1
S
1
C
0
0
S
0 C
following relations

0 0
F1
0 0
F2

0 0
F3

0 0
F4
=
1 0
Rx1
0 1
Ry1
0 0
Rx4
Ry4
0 0
{z } |
{F0 }
1
0
0
0
0
1
0
0
0
0
1
C
S
C
S
C
{z
0
S
C
[B0 ]1 [C1 ]
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
{z
|
}
{P}
{F5 } (6.10-b)
0
| {z }
{F }
x
0
} | {z }
[Bx ]
[B0 ]1 [C1 ] and [B0 ]1 [Bx ] [C2 ] we obtain
0
1
0
0 0 0
0
S
1 0
0
0 0 0
0
C
1
1
0
0
0
0
P
1
x2
C
C
S
S
C 0 C 1 0 0
Py2
S
+
{F5 }
1
0
1
0 0 0
C
0
| {z }
S
S
{Fx }
0
0
1
0 0 0
C
C
C
0
0
0
0
0 1 0
S
S
0
0
0 C
1 0 1
C
{z
} | {z } | {z }
[C1 ]
(6.10-a)
{P}
(6.11)
[C2 ]
Note, that this equation is not sufficient to solve for the unknown forces, as {F x } must be obtained
through force displacement relations ([D] or [K]).
10 ddagWhereas the identification of redundant forces was done by mere inspection of the structure in
hand based analysis of structure, this identification process can be automated.
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}

F
B I 2n(4n+r)
= {0}
P
Victor Saouma
(6.12-a)
Draft
123
2. Interchange columns
A
B
G
H
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
F2
0
0
1
0
1
0
0
0
F1
0
1
0
1
0
0
0
0
F3
C
S
0
0
C
S
0
0
F4 F5
0
0
0
0
0 C
0
S
0
0
1
0
0
C
1 S
Rx4
0
0
0
0
0
0
1
0
Ry4
0
0
0
0
0
0
0
1
Px2
0
0
1
0
0
0
0
0
Py2
0
0
0
0
(6.17)
3. Operate as indicates
A
B 0 = B + DB
E 0 = E+C
C
G
H
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
F2
0
0
1
0
1
0
0
0
F1
0
0
0
1
0
0
0
0
F3
C
S
0
0
1
S
0
0
F4 F5
0
0
0
S
0 C
0
S
0
1
1
0
0
C
1 S
Rx4
0
0
0
0
0
0
1
0
Ry4
0
0
0
0
0
0
0
1
Px2
0
0
1
0
1/C
0
0
0
Py2
0
1
0
0
(6.18)
F4
0
0
0
0
0
1
0
0
Rx4
0
0
0
0
0
0
1
0
Ry4
0
0
0
0
0
0
0
1
Px2
1
S/C
1
0
1/C
S/C
0
S/C
Py2
0
1
0
0
(6.19)
4. Operate as indicated
A0 = A + CE 0
0
B 00 = B 0 + SE
C = C
E0
F 0 = F SE 0
G
H0 = H + F 0
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
F2
0
0
1
0
0
0
0
0
F1
0
0
0
1
0
0
0
0
F3
0
0
0
0
1
0
0
0
F5
C
0
C
S
1
S
C
0
5. Interchange columns and observe that F5 is the selected redundant.
A0
B 00
C 0
E 0
F 0
G
H0
Victor Saouma
Rx1
1
0
0
0
0
0
0
0
Ry1
0
1
0
0
0
0
0
0
F2
0
0
1
0
0
0
0
0
F1
0
0
0
1
0
0
0
0
F3
0
0
0
0
1
0
0
0
F4
0
0
0
0
0
1
0
0
Rx4
0
0
0
0
0
0
1
0
Ry4
0
0
0
0
0
0
0
1
F5
C
0
C
S
1
S
C
0
Px2
1
S/C
1
0
1/C
S/C
0
S/C
Py2
0
1
0
0
(6.20)
Draft
6.3 Statics-Kinematics Matrix Relationship
125
[A] is a rectangular matrix which number of rows is equal to the number of the element internal
displacements, and the number of columns is equal to the number of nodal displacements. Contrarily
to the rotation matrix introduced earlier and which transforms the displacements from global to local
coordinate for one single element, the kinematics matrix applies to the entire structure.
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
(6.23)
where is the angle between the element and the X axis. whereas for flexural members:
v21
z21
v2 v1 z1 L
(6.24-a)
z2 z1
(6.24-b)
Example 6-5: Kinematics Matrix of a Truss

Considering again the statically indeterminate truss of
will be given by:
e
0
1
0
1 0
1e
0
1
0 1
C S
0
0
C
3
e

0
0
0
0
0
4

e5
0
0
C
S
0
=
1
0
0
0 0
u1

0
1
0
0 0
v
0
0
0
0 0
u
v4
0
0
0
0 0
{z
|
the previous example, the kinematic matrix

0
0
S
1
0
0
0
0
0
0
0
0
0
C
0
0
1
0
0
0
0
1
S
0
0
0
1
[A]
Applying the constraints: u1 = 0; v1 = 0; u4 = 0; and v4 = 0 we obtain:

e
0
1
0
1 0 0 0
0
1
0
0
1
0
1
0
0
0
C S
e3
0
0
C
S
0
0
0
e4
0
0
0 0 1 0 1
0
0
C
S
0 0 C S
e5
=
0
0
0
0
0 0 0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0 0 0 0
1
{z
|
[A]
u1
v1
u2
v2
u3
v3
u4
v4
u1
v1
u2
v2
u3
v3
u4
v4
(6.25)
(6.26)
We should observe that [A] is indeed the transpose of the [B] matrix in Eq. 6.15
6.3
Statics-Kinematics Matrix Relationship
Having defined 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.
19
Victor Saouma
Draft
6.4 Kinematic Relations through Inverse of Statics Matrix
6.4
127
Kinematic Relations through Inverse of Statics Matrix
28 We now seek to derive some additional relations between the displacements through the inverse of
the statics matrix. Those relations will be used later in the flexibility methods, and have no immediate
applications.
29
Rewriting Eq. 6.33 as

i1
it
h
h
t
1
{} = [A0 ]1 {0 } = [B0 ]
{0 } = [B0 ]
{0 }
we can solve for {F0 } from Eq. 6.7
{F0 } = [B0 ] {P} [B0 ] [Bx ] {Fx }

| {z }
| {z }
(6.37)
{} = [C1 ]t {0 }
(6.38)
[C1 ]
30
31
(6.36)
[C2 ]
Combining this equation with [B0 ]1 = [C1 ] from Eq. 6.37, and with Eq. 6.36 we obtain
Similarly, we can revisit Eq. 6.33 and write

{x } = [Ax ] {}
(6.39)
When the previous equation is combined with the rightmost side of Eq. 6.36 and 6.35 we obtain
h
it
t
1
{x } = [Bx ] [B0 ]
{0 }
(6.40)
32
Thus, with [B0 ]1 [Bx ] = [C2 ] from Eq. 6.37

{x } = [C2 ]t {0 }
(6.41)
This equation relates the unknown relative displacements to the relative known ones.
6.5
Congruent Transformation Approach to [K]
Note: This section is largely based on section 3.3 of Gallagher, Finite Element Analysys, Prentice Hall.
33 For an arbitrary structure composed of n elements, we can define the unconnected nodal load and
displacement vectors in global coordinate as
{Pe }
e
{ }
=
=
b bP1 c
1
b b c
bP2 c
...
b c
...
bPn c cT
n
b c c
(6.42-a)
(6.42-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.
Similarly, we can define the unconnected (or unassembled) global stiffness matrix of the structure as
[Ke ]:
34
{F}
e
[K ]
Victor Saouma
[Ke ] {}
[K1 ]
[K2 ]
[K ]
..
.
[Kn ]
(6.43-a)
(6.43-b)
Draft
6.5 Congruent Transformation Approach to [K]
129
Figure 6.4: Example 1, Congruent Transfer

We shall determine the global stiffness matrix using the two approaches:
Direct Stiffness
[ID]
{LM1 }
{LM }
[K]
0
0
0
b 0
b 1
1
2
3
0
2
0
0
0
(6.47-a)
2
0
3 cT
0 c
(6.47-b)
(6.47-c)
(7.692 + 1 105 )
(0. + 0.)
(18.75 + 0.)
(0. + 0.)
(.4 105 + 14.423)
(12. + 0.)
(0. + 18.75)
(12. + 0.)
(.0048 + .00469)
1 105
0.
18.75
0.
.4 105
12.
18.75
12.
.00949
(6.48-a)
Congruent Transformation
1. The unassembled stiffness matrix [Ke ], for node 2, is given by:
e
{F}
Mx1
My1
Fz1
Mx2
My2
Fz2
{}
7.692
sym
0
.4 105
0
12.
.0048
0.
1 10
0.
0.
14.423
sym
18.75
0.
.00469
1x
1y
Wz1
2x
2y
Wz2
(6.49-a)
element 1
(6.49-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 reflected by this matrix.
2. The kinematics matrix is given by:
{}
Victor Saouma
[A] {}
(6.50-a)
Draft
6.5 Congruent Transformation Approach to [K]

[]BC
131
[I]
(6.52-b)
The element stiffness matrices in global coordinates will then be given by:
[K]AB
[]TAB [k]AB []AB
.645 .259 7.031
.109 17.381
1 105
200
sym
and [K]BC = [k]BC
.645
.259
7.031
.645
.259
.109
17.381
.259
.109
7.031
17.381
.5 105
7.031
17.381
1 105
(6.53-a)
(6.53-b)
Direct Stiffness: We can readily assemble the global stiffness matrix:
[K]
200
200
(.645 + .75)
(.259 + 0.)
(.109 + .00469)
sym
1.395
.259
.1137
sym
7.031
1.37
2 105
(7.031 + 0.)
(17.38 + 18.75)
(1 + 1) 105
(6.54-a)
(6.54-b)
Congruent Transformation, global axis, Boolean [A]

1. We start with the unconnected global stiffness matrix in global coordinate system:
e
{F}
1
PX
PY1
1
MZ
2
PX
PY2
2
MZ
200
{}
.645
.259
.109
sym
7.031
17.381
1 105
.75
0.
.00469
0
sym
0.
18.75
1 105
2. Next we determine the kinematics matrix A:

{} = [A] {}
1
u
1 0
0 1
v
0 0
=
2
1 0
u2
0 1
v2
0 0
0
0
1
0
0
1
1
UX
VY1
1Z
2
UX
VY2
2Z
(6.55-a)
(6.55-b)
(6.56-a)
(6.56-b)
3. Finally, if we take the product [A]T [Ke ] [A] we obtain the structure global stiffness matrix
|{z} |{z} |{z}
36 66 63
[K] in Eq. 6.54-b
Congruent Transformation (local axis):

1. Unconnected stiffness matrix in local coordinates:
e e
e
Victor Saouma
{p }
1
PX
PY1
1
MZ
2
PX
PY2
2
MZ
{ }
200
.75
sym
0.
.00469
0.
18.75
1 105
.75
sym
0.
.00469
0.
18.75
1 105
u1x
vy1
z1
u2x
vy2
z2
(6.57-a)
(6.57-b)
Draft
Chapter 7
FLEXIBILITY METHOD
7.1
1
Introduction
Recall the definition of the flexibility matrix

(7.1)
{} [d]{p}
where {}, [d], and {p} are the element relative displacements, element flexibility matrix, and forces
at the element degrees of freedom free to displace.
2
As with the congruent approach for the stiffness matrix, we define:

{F(e) }
{(e) }
=
=
b bF(1) c
b b(1) c
bF(2) c
b(2) c
...
...
bF(n) c cT
b(n) c cT
(7.2-a)
(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 define the
unassembled structure flexibility matrix as:
(e) (e)
(e)
F
d
(7.3)
=
[0]
R
R

where d(e) is the unassembled global flexibility 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.
7.2
Flexibility Matrix
5 We recall from Sect. 6.1.2 that we can automatically identify the redundant forces [F x ] and rewrite
Eq. 7.3 as:
#

" (e)

d00
0
[0]
F0
=
(7.4)
(e)
x
Fx
[0] dxx
Draft
7.2 Flexibility Matrix
7.2.2
12
135
Solution of Internal Forces and Reactions
The internal forces and reactions can in turn be obtained through Eq. 6.14:
{F0 } = [C1 ] {P} + [C2 ] {Fx }
(7.12)
which is combined with Eq. 7.11 to yield:

h
i
1
{F0 } = [C1 ] [C2 ] [Dxx ] [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:
h
i
1
{p } = [Dpp ] [Dpx ] [Dxx ] [Dxp ] {P}
|
{z
}
(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 f 5 be the redundant force,
and use the [C1 ] and [C2 ] matrices previously derived.
Solution:
C = L2L+H 2 = 0.8 and S = L2H+H 2 = 0.6
From Eq. 7.4 we obtain
"
#

(e)
0
d00
F0
0
=
(7.15-a)
(e)
x
Fx
0
dxx
u1
v1
Rx1
(e)
Ry1
2
2
f2
(e)
1.5
f1
L
(e)
1.25
f3
(7.15-b)
=
3
AE
(e)
1.5
f4
Rx4
0
u4
0
R
y4
v
(e)
1.25
f5
5
From Example 6.1.2 we have
1
0
S/C 1
1
0
0
1
[C1 ] =
1/C
0
S/C 0
0
0
S/C 0
Victor Saouma
1
0.75
1
0
1.25
0.75
0
0.75
0
1
0
1
0
0
0
0
S
[C2 ] =
=
1
0.8
0
0.8
0.6
1
0.6
0.8
0
(7.16)
Draft
Chapter 8
SPECIAL ANALYSIS
PROCEDURES
To be edited
8.1
Semi-Rigid Beams
Often times, a beam does not have either a flexible or rigid connection, but rather a semi-rigid one
(such as in type II stell connection), Fig. 8.1
The stiffness relationship for the beam element (with shear deformation) was previously derived in Eq.
2.42
2
M1
M2
(4 + y )EIz
1 +
(1 + y )L
(2 y )EIz
1 +
L(1 + y )
(2 y )EIz
2
L(1 + y )
(4 + y )EIz
2
L(1 + y )
(8.1-a)
(8.1-b)
If we now account for the springs at both ends, Fig. 8.2, with stiffnesses k1s and k2s , then
M1 = k1s (1 1 ) and M2 = k2s (2 2 )
(8.2)
The stiffness relationship between M1 and M2 and the joint rotations 1 and 2 is obtained by eliminating
1 and 2 from above, yielding
EI
EI
(S11 1 + S12 2 ) and M2 =
(S21 1 + S22 2 )
L
L
M1 =
aa
\[\[
`a`
\[\[
cc
b
bcb
Flexible
_^]_^] _]_]
_]^_]^_]_]
Rigid
Semi-Rigid
Figure 8.1: Flexible, Rigid, and Semi-Rigid Beams
(8.3)
Draft
Part II
Introduction to Finite Elements
Draft
Chapter 9
REVIEW OF ELASTICITY
9.1
1
Stress
A stress, Fig 9.1 is a second order cartesian tensor, ij where the 1st subscript (i) refers to the
X3
33
32
31
23
13
21
22
X3
X2
12
11
X1
X1
X2
Figure 9.1: Stress Components on an Infinitesimal Element

direction of outward facing normal, and the second one (j) to the direction of component force.
11 12 13
t1
t2
= ij = 21 22 23 =
t3
31 32 33
(9.1)
2 In fact the nine rectangular components ij of turn out to be the three sets of three vector components
(11 , 12 , 13 ), (21 , 22 , 23 ), (31 , 32 , 33 ) which correspond to the three tractions t1 , t2 and t3 which
are acting on the x1 , x2 and x3 faces (It should be noted that those tractions are not necesarily normal
to the faces, and they can be decomposed into a normal and shear traction if need be). In other words,
stresses are nothing else than the components of tractions (stress vector), Fig. 9.2.
The state of stress at a point cannot be specified entirely by a single vector with three components; it
requires the second-order tensor with all nine components.
Draft
9.2 Strain
151
X2
B
-t *
1 S
1
-t
h N
S3
t*n S
A
X1
-t 2 S2
X3
V
Figure 9.3: Stress Traction Relations

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:

e1 e2 e3

N = ABAC = 4 2 0
(9.6-a)
4 0 6
=
12e1 + 24e2 + 8e3
The unit normal of N is given by

n=
6
2
3
e1 + e2 + e3
7
7
7
Hence the stress vector (traction) will be
and thus t = 97 e1 + 57 e2 +
9.2
7
3
7
10
7 e3
6
7
2
7
(9.6-b)
7
c 5
0
5
3
1
0
1 = b 97
2
(9.7)
5
7
10
7
(9.8)
Strain
Given the displacement ui of a point, the strain ij is defined as
ij =
Victor Saouma
1
(ui,j + uj,i )
2
(9.9)
Draft
Chapter 10
VARIATIONAL AND ENERGY

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

W
def
F.ds
(10.1-a)
Fx dx + Fy dy
(10.1-b)
dW
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.
The first 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 simplifies to:
We = U
(10.3)
Draft
10.1 Work, Energy & Potentials; Definitions
159
Considering uniaxial stresses, in the absence of initial strains and stresses, and for linear elastic
systems, Eq. 10.9 reduces to
Z
1
(10.10)
U=
E d
2 |{z}
12
13
When this relation is applied to various one dimensional structural elements it leads to
Axial Members:
U
d
2
P
A
P
AE
=
=
d = Adx
Torsional Members:
xy
xy
d
J
xy
=
=
= rddrdx
Z r Z 2
=
r2 ddr
o
1
2
10.1.2.1
L
0
|{z}
E
Mz y
U=
1
2
x =
z
zy
= M
EIz
d = dAdx
y 2 dA
1
2
|{z}
E d
xy
G
Tr
J
U=
xy
xy
2
{z
}
|
1
2
Flexural Members:
Iz
Internal Work versus Strain Energy
U=
1
2
P2
dx
AE
(10.11)
T2
dx
GJ
(10.12)
M2
dx
EIz
(10.13)
L
0
L
0
During strain increment, the work done by internal forces in a differential element will be the negative
of that performed by the stresses acting upon it.
Z
Wi =
dd
(10.14)
14
15 If the strained elastic solid were permitted to slowly return to their unstrained state, then the solid
would return the work performed by the external forces. This is due to the release of strain energy
stored in the solid.
Victor Saouma
Draft
10.7 Variational Calculus; Preliminaries
197
136 It can be shown that in the principle of virtual displacements, the Euler equations are the equilibrium
equations, whereas in the principle of virtual forces, they are the compatibility equations.
Euler equations are differential 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.
137
138 Finite Element formulation are based on the weak form, whereas the formulation of Finite Differences
are based on the strong form.
139
We still have to define . The first variation of a functional expression is

Z b
F
0
F = F
u + u
F 0
F
0 u
u
Rb
u +
u dx
=
u
u0
= a F dx
a
(10.189)
As above, integration by parts of the second term yields
u
a
F
d F
u
dx u0
dx
(10.190)
We have just shown that finding the stationary

value of by setting = 0 is equivalent to finding

the extremal value of by setting d()
equal
to zero.

d
140
=0
Similarly, it can be shown that as with second derivatives in calculus, the second variation 2 can
be used to characterize the extremum as either a minimum or maximum.
141
10.7.2
142
Boundary Conditions
Revisiting the second part of Eq. 10.184, we had

b
F
(x)
=0
|{z} u0 a
|
{z
}
Ess.
Nat.
| {z }
Boundary Cond.
(10.191)
This can be achieved through the following combinations

(a) = 0
and
(a) = 0
and
F
(a) = 0
u0
F
(a) = 0
u0
(b) = 0
F
(b) = 0
u0
(10.192-a)
(10.192-b)
and
(b) = 0
(10.192-c)
and
F
(b) = 0
u0
(10.192-d)
143 Generalizing, for a problem with, one field variable, in which the highest derivative in the governing
differential 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 field variable itself) through m-1. Trial displacement functions are explicitely required
to satisfy this B.C. Mathematically, this corresponds to Dirichlet boundary-value problems.
Victor Saouma
Draft
10.7 Variational Calculus; Preliminaries

Euler Equation:
d
dx
199
du
EA
dx
=0
0<x<L
(10.196)
Natural Boundary Condition:

EA
du
P =0
dx
Solution II We have
EA
F (x, u, u ) =
2
0
at x = L
(10.197)
2
(10.198)
du
dx
(note that since P is an applied load at the end of the member, it does not appear as part of
F (x, u, u0 ) To evaluate the Euler Equation from Eq. 10.186, we evaluate
F
F
=0 &
= EAu0
u
u0
(10.199-a)
Thus, substituting, we obtain

d F
F
u
dx u0
du
EA
dx
d
(EAu0 ) = 0 Euler Equation
dx
0 B.C.
(10.200-a)
(10.200-b)
Example 10-13: Flexure of a Beam

The total potential energy of a beam supporting a uniform load p is given by

Z L
Z L
1
1
M pw dx =
(EIw00 )w00 pw dx
=
2
2
0
0
|
{z
}
(10.201)
Derive the first variational of .

Solution:
Extending Eq. 10.189, and integrating by part twice

Z L
Z L
F
F
00
F dx =
w
+
w
dx
=
w00
w
0
0
Z L
=
(EIw00 w00 pw)dx
0
(EIw00 w0 )|0
(10.202-b)
[(EIw00 )0 w0 pw] dx
Or
(10.202-a)
= (EIw00 w0 )|0 [(EIw 00 )0 w]|0 +

| {z } | {z } | {z } | {z }
Nat. Ess.
Nat.
Ess.
|
{z
}
BC
(EIw00 )00 = p
(10.202-c)
L
0
[(EIw00 )00 + p] wdx = 0

{z
}
|
Euler Eq.
(10.202-d)
for all x
which is the governing differential equation of beams and

Victor Saouma
Draft
10.8 Homework
10.8
201
Homework
ueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueueuu
vuevueveuveu veuveu veuveu xveuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu vuvu
vuevueveuveu veuveu veuveu xxveuveu veuveu veuveu veuveu veuveu wwveuveu veuveu veuveu }||} veuveu veuveu veuveu veuveu veuveu veuveu veuveu {{veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu veuveu vuvu
testes tsts veuvueveveuveuveveuveuveveuveuvexxxveuveuveveuveuveveuveuveveuveuveveuveuvewwwveuveuveveuveuveveuveuve}|}|}| veuveuveveuveuveveuveuveveuveuveveuveuvezyzyzy veuveuveveuveuve{veuveuveveuveuveveuveuveveuveuveveuveuveveuveuveveuveuveveuveuve~ veuveuveveuveuve veuveuveveuveuveveuveuvevuvuv
testes tsts vevevevexvevevevevevevevevevevevevezy veveveveveveveveveve~ vevevevevev
settsestts
tsetsetsts
For the cantilivered beam shown below:

q
L/2
L/2
subjected to a linearly varying vertical load of magnitude q = q0 (1 xl ):
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
(10.203)
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 first fit the coefficients 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
1 Application of the Principle of Virtual Displacement requires an assumed displacement field.

This
displacement field can be approximated by interpolation functions written in terms of:
1. Unknown polynomial coefficients, most appropriate for continuous systems, and the Rayleigh-Ritz
method
y = a 1 + a 2 x + a 3 x2 + a 4 x3
(11.1)
A major drawback of this approach, is that the coefficients have no physical meaning.
2. Unknown nodal deformations, most appropriate for discrete systems and Potential Energy based
formulations
(11.2)
y = = N 1 1 + N2 2 + . . . + N n n
For simple problems both Eqn. 11.1 and Eqn. 11.2 can readily provide the exact solutions of the
d4 y
q
governing differential equation (such as dx
4 = EI for flexure), but for more complex ones, one must use
an approximate one.
11.2
Shape Functions
3 For an element (finite or otherwise), we can write an expression for the generalized displacement
(translation/rotation), at any point in terms of all its known nodal ones, .
n
X
i=1
Ni (x)i = bN(x)c{}
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 6= j.
3. Summation of N at any point is equal to unity N = 1.

4. N can be derived on the bases of:
(a) Assumed deformation state defined in terms of polynomial series.
(b) Interpolation function (Lagrangian or Hermitian).
(11.3)
Draft
11.2 Shape Functions
12
205
Substituting and rearranging those expressions into Eq. 11.5 we obtain

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

u = a1 x + a2 = b x 1 c
| {z }
[p]

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

u1
0 1
a1
=
u2
L 1
a2
| {z } | {z } | {z }
[L]
{a}
{}
following inversion of [L], this leads to

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

u1
x
x
) L
u = b (1 L
c
{z
} u2
|
| {z }
[p][L]1
{}
| {z }
(11.11)
(11.12)
(11.13)
[N]
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
Victor Saouma
a 1 x3 + a 2 x2 + a 3 x + a 4
dv
= 3a1 x2 + 2a2 x + a3
dx
(11.15-a)
(11.15-b)
Draft
Chapter 12
FINITE ELEMENT
FORMULATION
1 Having introduced the virtual displacement method in chapter 10, the shape functions in chapter 11,
and finally having reviewed the basic equations of elasticity in chapter 9, we shall present a general
energy based formulation of the element stiffness matrix in this chapter.
2 Whereas chapter 2 derived the stiffness matrices of one dimensional rod elements, the approach used
could not be generalized to general finite element. Alternatively, the derivation of this chapter will be
applicable to both one dimensional rod elements or contnuum (2D or 3D) elements.
It is important to note that whereas the previously presented method to derive the stiffness 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 stiffness matrix
for flexural one dimensional elements. Despite its approximation, this so-called finite 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
bNc and the nodal displacements {}
(x) = bN(x)c{}
The strain is then defined as:
def
(x) = [B(x)]{}
(12.1)
(12.2)
where [B] is the matrix which relates nodal displacements to strain field and is clearly expressed in terms
of derivatives of N.
12.1.1
Axial Members
u(x)
x

u1
L
c
|{z}
u2
N1
N2
{z
} | {z }
{}
bNc
x
(1 )
b | {z L }
|
(12.3-a)
Draft
12.3 Element Stiffness Matrix Formulation
10
215
Let us now apply the principle of virtual displacement and restate some known relations:
U
W
Z
(12.9-a)
bc{}d
(12.9-b)
{} = [D]{} [D]{0 }
{} = [B]{}
(12.9-c)
(12.9-d)
{} = [B]{}
bc = bc[B]T
(12.9-e)
(12.9-f)
11 Combining Eqns. 12.9-a,

12.9-b, 12.9-c, 12.9-f, and 12.9-d, the internal virtual strain energy is
given by:
Z
Z
T
U =
bc[B] [D][B]{} d
bc[B]T [D]{0 } d
{z } | {z }
{z } | {z }
|
|
{
}
{
}
{
{ 0 }
(12.10)
Z
Z }
T
T
0
[B] [D][B] d{} bc
[B] [D]{ }d
= bc
12
The virtual external work in turn is given by:

W =
bc
| {z }
{F}
|{z}
Virt. Nodal Displ. Nodal Force
13
combining this equation with:
bcq(x)dx
(12.11)
{} = [N]{}
yields:
W = bc{F} + bc
14
(12.12)
[N]T q(x) dx
(12.13)
Equating the internal strain energy Eqn. 12.10 with the external work Eqn. 12.13, we obtain:
Z
Z
[B]T [D][B] d{} bc
[B]T [D]{0 }d =
bc
|
|
{z
}
{z
}
0
[k]
bc{F} + bc
{z
{F }
{z
[N] q(x) dx
{z
}
(12.14)
{F }
15 Cancelling out the bc term, this is the same equation of equilibrium as the one written earlier on.

It relates the (unknown) nodal displacement n , the
o structure stiffness matrix [k], the external nodal

e
force vector F , the distributed element force F , and the vector of initial displacement.
16
From this relation we define:
The element stiffness matrix:

[k] =
Victor Saouma
[B]T [D][B]d
(12.15)
Draft
Chapter 13
SOME FINITE ELEMENTS

13.1
Introduction
1 Having first 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 defined the [D] matrix, and finally having applied the virtual displacement
method to finite element in chapter 12, we now revisit some one dimensional element whose stiffness
matrix was earlier derived, and derive the stiffness matrices of additional two dimensional finite elements.
13.2
2
Truss Element
The shape functions of the truss element were derived in Eq. 11.9:
x
N1 = 1
L
x
N2 =
L
The corresponding strain displacement relation [B] is given by:
xx
du
dx
1
[ dN
dx
=
=
=
[
|
L1
{z
[B]
dN2
dx
1
L ]
]
(13.2-a)
For the truss Z

element, the constitutive matrix [D] reduces to the scalar E; Hence, substituting into
Eq. 12.15 [k] =
[B]T [D][B]d and with d = Adx for element with constant cross sectional area we
4
obtain:
[k] = A
L
0
[k] =
=
AE
L
L1
1
L
AE
2
L
1
1
L
0

E b L1
1
1

1
1
1
1
1
L
cdx
(13.3)
dx
(13.4)
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 effects and treat the problem as a quasi-static one.
2 If the structure is subjected to an impact load, than one must be primarily concerned with (stress)
wave propagation. In such a problem, we often have high frequencies and the duration of the dynamic
analysis is about the time it takes for the wave to travel across the structure.
If inertia forces are present, then we are confronted with a dynamic problem and can analyse it through
any one of the following solution procedures:
1. Natural frequencies and mode shapes (only linear elastic systems)

2. Time history analysis through modal analysis (again linear elastic), or direct integration.
4 Methods of structural dynamics are essentially independent of finite element analysis as these methods
presume that we already have the stiffness, mass, and damping matrices. Those matrices may be obtained
from a single degree of freedom system, from an idealization/simplification of a frame structure, or from
a very complex finite element mesh. The time history analysis procedure remains the same.
14.2
Variational Formulation
5 In a general three-dimensional continuum, the equations of motion of an elementary volume without

damping is
LT + b = m
u
(14.1)
where m is the mass density matrix equal to I, and b

operator L is

0
x
y
0
0
L=
y x

z
0
0 z
6
is the vector of body forces. The Differential
0
0
(14.2)
For linear elastic material

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

fet
N b d +
and the vector of internal forces

fit
NT tt d
(14.14)
BT t d
(14.15)
BT DBdu
(14.16)
But since = D = DB, then

fit =
It should be noted that Eq. 14.13 defines the consistent mass matrix, which is a fully populated
matrix. Alternatively, a lumped mass matrix can be defined, resulting only in a diagonal matrix. For an
axial element, the lumped mass matrix is obtained by placing half of the total element mass as a particle
at each node.
10
11
For linear elastic systems, and if damping was considered

t + Cu t + Ku = fet
Mu
(14.17)
the damping matrix matrix is often expressed in terms of the mass and stiffness matrix, called Rayleigh
damping, as
(14.18)
C = M + K
14.2.2
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
Z
Z
Z
Z
uT bt+t d
(14.19)
uT tt+t d +
T t+t d =
uT md
ut+t +
13
For linear problems

t+t = De But+t
Victor Saouma
(14.20)
Draft
14.3 Time Discretization
227
In practical finite element analysis, we are therefore mainly interested in a few effective 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 effectiveness. In the present
implementation the direct integration method is employed.
20
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 different form is carried out.
14.3.1
Explicit Time Integration
For the explicit time integration, we adopt at starting point a slight variation of the differential
equation previously derived, one which includes the effects of damping.
22
+ Cu + Ku = fe
Mu
|{z}
|{z} |{z} |{z}
FI (t)
FD (t)
Fi (t)
(14.29)
fe (t)
, u
where M, C and K are the mass, damping and stiffness matrices, fe is the external load vector and u
and u are the acceleration, velocity and displacement vectors of the finite element assemblage.
23 An alternative way to consider this equation, is to examine the equation of statics at time t, where
FI (t), FD (t), Fi (t), Fe (t) are the inertia, damping, internal and external elastic forces respectively.
14.3.1.1
Linear Systems
For the semi-discretized equation of motion, we adopt the central difference method. This method
is based on a finite difference approximation of the time derivatives of displacement (velocity and acceleration). Assuming a linear change in displacement over each time step, for the velocity and the
acceleration at time t:
24
t
u
t
u
ut+t utt
2t
utt 2ut + ut+t
t2
(14.30-a)
(14.30-b)
where t is the time step.

25
t + Cu t + Ke u = fet ) we obtain
Substituting into Eq. 14.17 (Mu
t+t
1
t
u
K 2 M ut 1 M 1 C utt
M
+
C
=
f
e
t2
t2
2t
t2
2t
|{z}
|{z}
|{z}
|{z}
|{z}
|
a0
{z
b
M
a1
from where we can solve for ut+t .

26
a2
{z
bfet
a0
a1
(14.31)
It should be noted that:

1. Solution is based on the equilibrium condition at time t
2. No factorization of the stiffness 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:
Victor Saouma
Draft
14.3 Time Discretization
229
31 Since no structural stiffness 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
T
(14.42)
If this condition is violated, numerical instability occurs.

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

Elastic Stability Limit
Bifurcation
Bifurcation
Softening
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 effects of finite 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 stiffness (when a member is
subjected to a tensile axial load).
First-order inelastic equilibrium equations written in terms of the geometry of the undeformed structure, accounts for material non-linearity.
Second-order inelastic equations of equilibrium written in terms of the geometry of the deformed
shape, can account for both geometric and material nonlinearities. Most suitable to determine
failure or ultimate loads.
Draft
15.1 Introduction
Thus buckling will occur if
237
P
EI

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

cr =
2 E

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

y, u
dx
w
P
M
x
P
v
x
V+ V dx
x
dx
P
M+ M dx
x
i
j
Figure 15.3: Simply Supported Beam Column; Differential Segment; Effect of Axial Force P
the horizontal axis is small, leads to

dM
(dx)2
dV
dv
M M+
dx + w
+ V +
dx P
dx = 0
dx
2
dx
dx
(15.9)
Neglecting the terms in dx2 which are small, and then differentiating each term with respect to x, we
obtain
d2 M
dV
d2 v
P
=0
(15.10)
dx2
dx
dx2
16
17
However, considering equilibrium in the y direction gives

dV
= w
dx
Victor Saouma
(15.11)
Draft
15.3 Elastic Instability; Bifurcation Analysis

obtain
Ke =
and
AL2
I
AL2
EI
L3
0
0
0
1
0
0
P
0
Kg =
L
0
0
245
4
2
AL
I
2
2 AL
I
0
0
0
4
0
0
0
0
0
3
0
0
4L2
6L
2L2
3
0
0
2 2
15 L
L
10
L2
30
5
0
0
6L
24
0
5
0
0
L
10
12
5
0
0
2L2
0
8L2
6
0
0
L2
30
0
4 2
15 L
(15.54)
(15.55)
4. Noting that in this case Kg = Kg for P = 1, the determinant |Ke + Kg | = 0 leads to

1
2

1 AL
I
AL
2
4 I

3 0

5 0
6 0
5. Introducing =
AL2
I
4
2
AL
I
2
2 AL
I
0
0
0
and =
3
0
0
2 L4
4L2 15
EI
1 L3
6L + 10
EI
1 L4
2L2 + 30
EI
L2
EI ,
1
4

1
4 2
3 0
0

5 0
0
6 0
0
5
0
0
1 L3
6L + 10
EI
L2
24 12
5 EI
0
6
0
0
1 L4
2L2 + 30
EI
0
4 L4
8L2 15
EI
the determinant becomes

3
0
0
2 2 15
6L + 10
2 + 30
5
6

0
0

0
0

=0
2
+
6L + 10
30

0
12 2 5

0
4 2 15

=0

(15.56)
(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
,
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
15.3 Elastic Instability; Bifurcation Analysis
247
The element stiffness matrices are given by
k1e
k1g
k2e
k3e
k3g
u1
20
1, 208
u1
0.01
0.10
P

2
1, 208
96, 667
2
0.10
16.00
128, 890
64, 440
u1
47
1, 678
3
1, 678
80, 556
u1
0.01667
0.1
P

The global equilibrium relation can now be written as
0
0
0
3
64, 440

128, 890
0
0
3
0
0
0.1
9.6
(Ke P Kg ) = 0
u1

(66.75) P (0.026666)

(1, 208.33) P (0.1)

(1, 678.24) P (0.1)
2
(1, 208.33) P (0.1)
(225, 556.) P (16.)
(64, 444.) P (0)
(15.60-a)
(15.60-b)
(15.60-c)
(15.60-d)
(15.60-e)
(15.61)
3

(1, 678.24) P (0.1)
(64, 444.4) P (0) = 0
(209, 444.) P (9.6)
(15.62)
The smallest buckling load amplification factor is thus equal to 2, 017 kips.
(* Initialize constants *)
a1=0
a2=0
a3=0
i1=100
i2=200
i3=50
l1=10 12
l2=15 12
l3=6 12
e1=29000
e2=e1
e3=e1
(* Define elastic stiffness matrices *)
ke[e_,a_,l_,i_]:={
{e a/l , 0
, 0
, -e a/l
{0
, 12 e i/l^3 , 6 e i/l^2 , 0
{0
, 6 e i/l^2
, 4 e i/l
, 0
{-e a/l , 0
, 0
, e a/l
Victor Saouma
,
,
,
,
0
-12 e i/l^3
-6 e i/l^2
0
,
,
,
,
0
6 e i/l^2
2 e i/l
0
},
},
},
},
Draft
15.5 Summary
255
STRONG FORM
?
2nd Order D.E.
2 B.C.
WEAK FORM
?
4th Order D.E.
4 B.C.
?
2
d
v
d
u
y
x =
+
dx
dx2
?
d2 y P = 0
EI
dx2
v = A sin kx B cos kx
U=
?
2
4
d
v
d
v
EI 4 P 2 = w
dx
dx
v = C1 sin kx + C2 cos kx + C3 x + C4
1
2
1
2
dv
dx
2
E2 d
?
K=
Kg
Ke
?
P = (Ke + Kg )u
?
|Ke + Kg | = 0
Figure 15.5: Summary of Stability Solutions
Victor Saouma
Draft
Chapter 16
REFERENCES
Basic Structural Analysis :
1. Arbabi, F., Structural Analysis and Behavior, McGraw-Hill, Inc., 1991
2. Beaufait, F.W., Basic Concepts of Structural Analysis, Prentice-Hall Inc., Englewood Cliffs,
N.J., 1977
3. Chajes, A., Structural Analysis, Prentice-Hall, Inc., Englewood Cliffs, 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 Cliffs,
N.J., 1982
8. Laursen, H.I., Structural Analysis, 2nd Ed., McGraw-Hill, N.Y., 1978
9. Morris, J.C., Wilbur, S., and Utku, S., Elementary Structural Analysis, McGraw-Hill, N.Y.,
1976
10. Wang, C.K., Intermediate Structural Analysis, McGraw-Hill, N.Y., 1983
Matrix Analysis :
1. Argyris, J.H., Recent Advances in Matrix Methods of Structural Analysis, Pergamon Press,
Oxford, 1964
2. Beaufait, F.W., Rowan Jr., W.H., Hoadley, P.G., and Hackett, R.M., Computer Methods of
Structural Analysis, 4th Edition, 1982
3. Bhatt, P., Programming the Matrix Analysis of Skeletal Structures, Halsted Press, 1986
4. Elias, Z.M., Theory and Methods of Structural Analysis, John Wiley & Sons, 1986
5. Holzer, S.M., Computer Analysis of Structures. Elsevier, 1985
6. Livesley, R., Matrix Methods of Structural Analysis, Pergamon Press, Oxford, 964
7. Martin, H.C., Introduction to Matrix Methods of Structural Analysis, McGraw-Hill, N.Y., 1966
8. McGuire, W., and Gallagher, R.H., Matrix Structural Analysis, John Wiley and Sons Inc.,
N.Y., 1979
9. Meek, J.L., Matrix Structural Analysis, McGraw-Hill, N.Y., 1971
10. Meyers, V.J., Matrix Analysis of Structures, Harper and Row, Publ., N.Y., 1983
11. Przemieniecki, J.S., Theory of Matrix Structural Analaysis, McGraw-Hill, N.Y., 1968
12. Weaver Jr, W., and Gere, J.M., Matrix Analysis of Framed Structures, 2nd Ed., Van Nostrand
Co., N.Y., 1980
Introduction to Finite Element and Programming :
1. Bathe, K.J., Finite Element Procedures in Engineering Analysis, Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1982
Draft
Appendix A
REVIEW of MATRIX ALGEBRA

Because of the discretization of the structure into a finite 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
Definitions
Matrix:
A11
A21
..
.
[A] =
Ai1
.
..
Am1
A12
A22
..
.
...
...
..
.
A1j
A2j
..
.
...
...
..
.
A1n
A2n
..
.
Ai2
..
.
...
..
.
...
Aij
..
.
...
..
.
...
Ain
..
.
Am2
Amj
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:
A row vector would be
B1
B2
..
.
{X} =
B
..
Bm
bCc = b B1
B2
...
Bi
(A.2)
...
Bm c
(A.3)
Note that scalars, vectors, and matrices are tensors of order 0, 1, and 2 respectively.
Square matrix: are matrices with equal number of rows and columns. [A]mm
Symmetry: Aij = Aji
Identity matrix: is a square matrix with all its entries equal to zero except the diagonal terms which
are equal to one. It is often denoted as [I], and

0, if i 6= j
(A.4)
Iij =
1, if i = j
Draft
A.3 Determinants
261
Scalar Multiplication:
[B]
Bij
= k [A]
= kAij
(1.11-a)
Matrix Multiplication: of two matrices is possible if the number of columns of the first one is equal
to the number of rows of the second.
[A]mn
Aij
[B]mp [C]pn
(1.12-a)
bBi c1p {Cj }p1

|
{z
}
11
p
X
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] 6= [B][A]
A.3
Determinants
The Determinant of a matrix [A]nn , denoted as det A or |A|, is recursively defined as

det A =
n
X
(1)1+j a1j det A1j
(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

(A.14)
a21 a22 = a11 a22 a12 a21
For a 3 3 matrix

a11 a12

a21 a22

a31 a32
a13
a23
a33

a
= a11 22
a32

a
a23
a12 21

a31
a33

a
a23
+ a13 21

a31
a33
= 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

a22
a32
(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 floating operations should be performed in order to evaluate the
determinant.
This is why it is impractical to use Cramers rule to solve a system of linear equations.
Some important properties of deteminants:
Victor Saouma
Draft
A.6 Eigenvalues and Eigenvectors
A.6
263
Eigenvalues and Eigenvectors
A special form of the system of linear equation
A11 A12 . . .
A21 A22 . . .
[A] = .
..
..
..
.
.
Ai1 Ai2 . . .
A1n
A2n
..
.
Ann
x1
x2
..
.
xn
is one in which the right hand side is a multiple of the solution:
B1
B2
=
..
Bn
(1.21)
[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
or

A11

A21

[A] =
..

.

Ai1
A12
...
A22 . . .
..
..
.
.
Ai2
...
(1.24)

=0

Ann
A1n
A2n
..
.
(1.25)
When the determinant is expanded, we obtain an nth order polynomial in terms of which is known as
the characteristic equation of [A]. The n solutions (which can be real or complex) are the eigenvalues
of [A], and each one of them i satisfies
[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 finite element formulations.
Victor Saouma
Draft
Appendix B
SOLUTIONS OF LINEAR
EQUATIONS
Note this chapter is incomplete
B.1
Introduction
1 Given a system of linear equations [A] nn {x} = {b} (which may result from the direct stiffness
method), we seek to solve for {x}. Symbolically this operation is represented by: {x} = [A] 1 {b}
2
There are two approaches for this operation:
Direct inversion using Cramers rule where [A]1 = [adjA]

[A] . However, this approach is computationally
very inefficient 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, finite 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 defined 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
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.
Draft
B.2 Direct Methods
267
2. Elimination of the first column:

(a) row 1=0.1(row 1)
(b) row 2=(row2)+20(new row 1)
(c) row 3=(row 3) -5(new row 1)
1 0.1
0 5
0 2.5
0 0 0.1
1 0
4
0 1
5.5
0.1
2
0.5
0.5
10
7.5
3. Elimination of second column

(a) row 2=0.2(row 2)
(b) row 1=(row 1)-0.1(new row 2)
(c) row 3=(row 3) -2.5(new row 2)
1 0
0 1
0 0
0.7
2
2.5
0.06
0.4
1.5
0.02
0.2
0.5
4. Elimination of the third column

(a) row 3=0.4(row 3)
(b) row 1=(row 1)+0.7(new row 3)
(c) row 2=(row 2)-2(new row 3)
1 0
0 1
0 0
0 0.36
0 1.6
1 0.6
0.16
0.6
0.2
B.2.1.1
0 0.02
0.8
0
1
3.5
0.28
1
2
0.8
1.4
0.4
| {z
{x}
This last equation is [I|A1 ]
(2.6-a)
(2.7-a)
(2.8-a)
Algorithm
Based on the preceding numerical examples, we define a two step algorithm for the Gaussian ellimination.
Defining akij to be the coefficient of the ith row & j th column at the k th reduction step with i k &
j k:
5
Reduction:
ak+1
ik
ak+1
ij
bk+1
ij
=0
=
akij
bkij
k
ak
ik akj
k
akk
k
ak
ik bkj
ak
kk
Backsubstitution:
xij =
biij
k<in
k < i n; k < j n
(2.9)
k < i n; 1 < j m
Xn
k=i+1
aiii
aiik xkj
(2.10)
Note that Gauss-Jordan produces both the solution of the equations as well as the inverse of the original
matrix. However, if the inverse is not desired it requires three times (N 3 ) more operations than Gauss
3
or LU decomposition ( N3 ).
Victor Saouma
Draft
B.2 Direct Methods
269
4. Take row by row or column by column
lij =
aij
Xj1
uij = aij
lii = 1
lik ukj
k=1
ujj
Xi1
lik ukj
k=1
i>j
(2.16)
ij
Note:
1. Computed elements lij or uij may always overwrite corresponding element aij
2. If [A] is symmetric [L]T 6= [U], symmetry is destroyed in [A]F
For symmetric matrices, LU decomposition reduces to:

uij = aij
lii = 1
uji
lij = ujj
Xi1
k=1
lik ukj
ij
(2.17)
Example B-3: Example

Given:
7
4
A=
1
3
9
5
6
2
1
2
3
1
2
7
4
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
= ua21
11
= a22 l21 u12
= a23 l21 u13
= a24 l21 u14
= 74
= 5 4 79
= 2 + 4 71
= 7 4 72
= 10.1429
= 2.5714
= 10.1429
Row 3:
l31
l32
u33
u34
= ua31
11
31 u12
= a32 l
u22
= a33 l31 u13 l32 u23
= a34 l31 u14 l32 u24
= 17
= 6(0.1429)(9)
10.1429
= 3 (0.1429)(1) (0.4647)(2.5714)
= 4 (0.1429)(2) (0.4647)(8.1429)
= 0.4647
= 1.6622
= 8.0698
Row 4:
l41
l42
l43
u44
= ua41
11
41 u12
= a42 l
u22
a43 l41 u13 l42 u23
=
u33
= a44 l41 u14 l42 u24 l43 u34
Victor Saouma
= 73
= 2(0.4286)(9)
10.1429
= 1(0.4286)(1)(0.5775)(2.5714)
1.6622
= 5 (0.4286)(2) (0.5775)(8.1429) (1.2371)(8.0698)
= 0.5775
= 1.2371
= 8.8285
Draft
B.3 Indirect Methods
271
Given:
4
6
A=
10
4
6
13
13
6
10
13
27
2
a11
=
=
=
=
4
6
2
72
(2.28-a)
Solution:
Column 1:
l11
l21
l31
l41
=
=
=
=
a21
l11
a31
l11
a41
l11
=2
=3
=5
=2
6
2
10
2
4
2
Column 2:
l22
l32
l42
=
=
=
2
a22 l21
=
=
=
a32 l31 l21

l22
a42 l41 l21
l22
13 32
13(5)(3)
2
6(2)(3)
2
=2
= 1
=0
Column 3:
l33
l43
=
=
2 l2
a33 l31
32
=
=
a43 l41 l31 l42 l32

l33
27 52 (1)2
2(2)(5)(0)(1)
1
=1
= 8
Column 4:
l44
2 l2 l2
a44 l41
42
43
p
72 (2)2 (0)2 (8)2
=2
or
2
3
5
2
|
0
2
1
0
B.2.4
B.3
16
{z
[L]
0
0
1
8
0
2
0
0
0 0
2
0
}|
{z
[A]
3
2
0
0
5
1
1
0
{z
[U]
2
0
8
2
}
(2.33-a)
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.
Victor Saouma
Draft
B.4 Ill Conditioning
B.4.2
273
Pre Conditioning
If a matrix [K] has an unacceptably high condition number, it can be preconditionedthrough a congruent operation:
[K0 ] = [D1 ][K][D2 ]
(2.41)
22
However there are no general rules for selecting [D1 ] and [D2 ].
B.4.3
Residual and Iterative Improvements
Victor Saouma
Draft
Appendix C
TENSOR NOTATION
NEEDS SOME EDITING
1
Equations of elasticity are expressed in terms of tensors, where

A tensor is a physical quantity, independent of any particular coordinate system yet specified most
conveniently by referring to an appropriate system of coordinates.
A tensor is classified by the rank or order
A Tensor of order zero is specified 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.
3 To express tensors, there are three distinct notations which can be used: 1) Engineering; 2) indicial;
or 3) Dyadic.
4 Whereas the Engineering notation may be the simplest and most intuitive one, it often leads to long
and repetitive equations. Alternatively, the tensor and the dyadic form will lead to shorter and more
compact forms.
C.1
Engineering Notation
In the engineering notation, we carry on the various subscript(s) associated with each coordinate axis,
for example xx , xy .
C.2
Dyadic/Vector Notation
Uses bold face characters for tensors of order one and higher, , . This notation is independent of
coordinate systems.
Since scalar operations are in general not applicable to vectors, we define

A+B
AB
A
AB
B+A
(3.1-a)
= BA
= Ax i + A y j + Az k
(3.1-b)
(3.1-c)
= |A||B| cos(A, B)
= A x Bx + A y By + A z Bz
(3.1-d)
Draft
C.3 Indicial/Tensorial Notation
277
this simple compacted equation (expressed as x = cz in dyadic notation), when expanded would yield:
x1
x2
=
=
c11 z1 + c12 z2 + c13 z3

c21 z1 + c22 z2 + c23 z3
x3
c31 z1 + c32 z2 + c33 z3
(3.8-a)
Similarly:
Aij = Bip Cjq Dpq
Victor Saouma
(3.9)
A11
A12
= B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
= B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
A21
A22
= B21 C11 D11 + B21 C12 D12 + B22 C11 D21 + B22 C12 D22
= B21 C21 D11 + B21 C22 D12 + B22 C21 D21 + B22 C22 D22
(3.10-a)
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.
D.1
Integration by Parts
The integration by part formula is

Z
b
0
u(x)v (x)dx =
a
or
D.2
udv =
a
b
uv|a
v(x)u0 (x)dx
(4.1)
(4.2)
vdu
a
Green-Gradient Theorem
Greens theorem is
D.3
b
u(x)v(x)|a
(Rdx + Sdy) =
Z
R
S
x
y
dxdy
(4.3)
Gauss-Divergence Theorem
The general form of the Gauss integral theorem is

Z
or
v.nd =
vi ni d =
divvd
(4.4)
vi,i d
(4.5)

Draft: Finite Element I

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Draft: Finite Element I

Enviado por

Direitos autorais:

Formatos disponíveis

Draft

Lecture Notes in:

Matrix Structural Analysis of Framed Structures

2 ELEMENT STIFFNESS MATRIX

6 EQUATIONS OF STATICS and KINEMATICS

8 SPECIAL ANALYSIS PROCEDURES

Finite Element I; Framed Structures

E 10-8 Tapered Beam; Fourrier Series . . . . . . . . . . . . . . .

12 FINITE ELEMENT FORMULATION

13 SOME FINITE ELEMENTS

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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

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

Example of Nodal Definition .

Examples of Influence Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3D Transformations of Linear Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

Internal Element Force Definition for the Statics Matrix . . . . . . . . . . . . . . . . . . . 118

Possible Combinations of Real and Hypothetical Formulations . .

11.1 Characteristics of Beam Element Shape Functions . . . . . . . . . . . . . . . . . . . . . . 208

Why Matrix Structural Analysis?

1.3 Structural Idealization

Structural design must satisfy:

Structural analysis must satisfy

Finite Element I; Framed Structures

1.3 Structural Idealization

Table 1.3: Example of Group Number

Figure 1.1: Global Coordinate System

We should differentiate between 2 coordinate systems:

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

Figure 1.2: Local Coordinate Systems

Finite Element I; Framed Structures

1.4 Degrees of Freedom

Figure 1.5: Independent Displacements

Table 1.4: Degrees of Freedom of Different Structure Types Systems

Finite Element I; Framed Structures

1.5 Course Organization

Finite Element I; Framed Structures

Matrix Structural Analysis of

ELEMENT STIFFNESS MATRIX

Table 2.1: Examples of Influence Coefficients

2.4 Stiffness Coefficients

Similarly, we would obtain:

Those results can be summarized in a matrix form as:

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

In the stiffness method, we

Finite Element I; Framed Structures

2.5 Force-Displacement Relations

Figure 2.2: Definition of Element Stiffness Coefficients

Finite Element I; Framed Structures

2.7 Remarks on Element Stiffness Matrices

Upon substitution, the grid element stiffness matrix is given by

Remarks on Element Stiffness Matrices

Finite Element I; Framed Structures

Finite Element I; Framed Structures

STIFFNESS METHOD; Part I:

The Stiffness Method

Let us define a transformation matrix [(e) ] such that:

Note that we use the same matrix

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

[K(e) ] = [(e) ]T [k(e) ][(e) ]