Chair of Structural Mechanics

TUM School of Engineering and Design

Technical University of Munich

Stability of Structures
Prof. Dr.-Ing. Gerhard Müller
Corinna Treimer, M.Sc.

Winter Term 2022/2023

Copyright © Prof. Dr. Ing. Gerhard Müller

Published by the Chair of Structural Mechanics, Technical University of Munich

For academic use only. Reproduction and distribution are not permitted without the permis-
sion of the author.

Munich, October 10, 2022.

Contents
1 Introduction 1
1.1 Local and Global Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Non–Linearities in the Basic Relations of Structural Analysis . . . . . . . . 1
1.3 Geometrically and physically non–linear analysis of rigid body – spring systems 7
1.3.1 Retrospection: Principle of virtual displacements (example) . . . . . 7
1.3.2 Hinged Rigid Cantilever with Rotational Spring: Levels of Approxima-
tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Snap-through problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4.1 Introductory Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4.2 Kinematics, Constitutive Law . . . . . . . . . . . . . . . . . . . . . 16
1.4.3 Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2 Second Order Analysis and Stability of Plane Frames 20

2.1 Preliminaries and Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2 Basic Relations of Beam Columns with variable Cross-Sections on Elastic Spring
Beds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.1 The Governing Equation of Bending . . . . . . . . . . . . . . . . . . 21
2.2.2 Element Matrices on the Basis of exact solutions . . . . . . . . . . . 24
2.2.3 Relevant Differential Equations and their solutions . . . . . . . . . . 27
2.2.4 Derivation of the P.V.D. starting with the Governing Differential Equa-
tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2.5 Element Matrices for Bending on the Basis of Cubic Test Functions
(Finite Element Method) . . . . . . . . . . . . . . . . . . . . . . . . 40
2.3 Analysis and Structural Behavior of Plane Frames (Second Order Analysis) 43
2.3.1 Methods of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.3.2 Advantages and Disadvantages of Exact and Approximate Stiffness Ma-
trices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.3.3 The Buckling Coefficient ε as a Measure of Stability of a Member . 46
2.3.4 The Principle of Superposition in the Second Order Theory . . . . . 47
2.3.5 Redistribution of Internal Forces and Moments in Beam Structures . 48
2.3.6 Overcritical Loading of Subsystems . . . . . . . . . . . . . . . . . . 49
2.3.7 Considering Elastic Supports at the Boundaries . . . . . . . . . . . 51
2.4 Stability of Beams and Frames Subjected to Plane Loading . . . . . . . . . 53
2.4.1 Determination of the Critical Load of Frames with Exact Stiffness Ma-
trices (Second Order Theory) . . . . . . . . . . . . . . . . . . . . . . 53

Stability of Structures
Winter Term 2022/2023
III
 2.4.2 Linearized Buckling Analysis with Geometrical Stiffness Matrices . . 55
2.4.3 Geometrical Nonlinear Analysis and Eigenvalue Problem: Example . 57
2.4.4 Effective Length of Beam Columns of Plane Frames . . . . . . . . . 64

3 Second Order Analysis and Stability of Space Frames 72

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.1.1 Failure Modes of Bars . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.2 General and Kinematic Assumptions, Constitutive Law . . . . . . . . . . . 78
3.2.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.2 Kinematic Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.2.3 Consitutive Laws for Stresses . . . . . . . . . . . . . . . . . . . . . . 83
3.3 Bending and Torsion in a 1 Order Analysis
st
. . . . . . . . . . . . . . . . . 84
3.3.1 Principle of Virtual Displacements: Initial Formulation . . . . . . . 84
3.3.2 Virtual Work of Internal Forces, Transformation in Principal Axes . 84
3.4 Second Oder Analysis of Combined Torsion and Bending . . . . . . . . . . 89
3.4.1 Preliminary Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.4.2 The Principle of Virtual Displacements: Initial Formulation . . . . . 90
3.4.3 Supplement: Torsion Caused By Inclination of the Longitudinal Fibers 91
3.4.4 Loads and Stress Resultants of the Deformed Configuration and their
Transformation to the Reference Frame of the Undeformed Structure 96
3.4.5 Principle of Virtual Displacements: Resulting Formulation . . . . . 102
3.4.6 Differential Equations, boundary Conditions, Internal Forces . . . . 105
3.4.7 Internal Forces in the Reference Frames of the Initial and of the Actual
States – Matrix Formulation . . . . . . . . . . . . . . . . . . . . . . 114
3.4.8 Special Case with an Exact Solution:
Girder in Yoke Supports having a single symmetric Cross–section and
M y = const. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.5 Torsional Buckling and Lateral–Torsional Buckling of Beam Columns and Gird-
ers of Uniform Cross–Section;
Governing Equations, Analytical Solutions and Structural Behaviour . . . . 122
3.5.1 Buckling and torsional Buckling of Thin–Walled Bars
of Double Symmetric Cross–Sections
with a Centric Concentrated Compressive Force . . . . . . . . . . . 123
3.5.2 Buckling and Lateral Torsional Buckling of Bars
of Arbitrary Thin–Walled Cross–Section
with a Centric Concentrated Compressive Force . . . . . . . . . . . 126
3.5.3 Buckling and Lateral Torsional Buckling of Bars
of Thin–Walled Monosymmetric Cross–Section,
with a Concentrated Compressive Force, Acting Eccentrically on the
z–Axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
3.5.4 Lateral Buckling of Deep Girders
of Thin–Walled Monosymmetric Cross–Section
(f. i. steel work cross–sections) . . . . . . . . . . . . . . . . . . . . . 130

Stability of Structures Contents

Winter Term 2022/2023 IV
 3.5.5 Lateral Buckling of Deep Girders
of Thin–Walled monosymmetric Cross–Section
being Stressed by M̊y = const. . . . . . . . . . . . . . . . . . . . . . 131
3.5.6 Lateral–Torsional Buckling of Glued Laminated Beams of Rectangular
Cross–Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

Bibliography 138

Stability of Structures Contents

Winter Term 2022/2023 V
1 Introduction
In this lecture, stability problems of beam-like structures are discussed with a focus on the
description of physical phenomena such as buckling, torsional buckling, or lateral-torsional
buckling, and a precise step by step derivation of the differential equations and equation
systems.
Second order analysis and stability investigations are considered for plane frames and beam
structures (spatial systems) for varying boundary conditions such as bedding for instance. The
kinematic relations are discussed and the differential equations are derived using energy- and
work- principles. Analytical solutions for simplified systems are handled along with numerical
approaches (Finite Element Method). The analytical implementation in computer algebra
systems as well as numerical codes are discussed and practiced with the help of exercise
sheets.
The lecture is conceived on the basis of the lecture notes of Apl. Prof. Dr.-Ing. Gunter
Kiener [Kiener 2004; Wunderlich and Kiener 2004], who was a lecturer at TU München from
1972 to 2004. Prof. Kiener has been giving lectures in Numerics, Stability, and Torsion of
Structures as well as Tensor Analysis. The distinguishing feature of his lectures, which I have
appreciated very much in the year 2000 as one of his students, which is also the backbone
of these lecture notes, was the focus on very precise and clear mechanical derivations and
formulations. I thank Prof. Kiener very much for providing this material.

Martin Buchschmid

1.1 Local and Global Failure

1.2 Non–Linearities in the Basic Relations of Structural

Analysis

Structural analysis aims to prove that a structure can safely support the anticipated loadings
and provide sufficient serviceability. This includes the investigation of relations between ap-
plied forces and moments (due to self weight, wind load, traffic, etc.), imposed displacements

Stability of Structures
Winter Term 2022/2023
1
(support settlements etc.), or strains (due to thermal expansions, shrinkage, creep, etc.), and
the internal forces and support reactions, stresses, strains, and displacements caused by these
actions.

Four classes of quantities are related:

a) Forces and moments acting on a structure P

b) Stresses or internal forces and moments σ
c) Strains (axial strain, curvature, rate of twist) ε
d) Displacements and rotations u

Quantities of class b) only occur due to applied actions, whereas quantities of classes a), c),
and d) can be both imposed and caused.
The relations between the different classes can be illustrated as follows:

external internal
forces (1)
p σ

(4) (2)

displacements u ε
or strains (3)
Figure 1.1

Relation (1) involves equations of equilibrium usually relating externally applied forces and
moments. These equations can be derived considering either the undeformed or the
deformed system, according to the assumptions of the mechanical model employed. Both
the differential considerations and suitable energy methods (e.g., principle of virtual
work) yield the same results because they are essentially equivalent.
Relation (2) is governed by one or more different constitutive laws depending on the used
material. These laws relate strains to corresponding stresses or internal forces and mo-
ments, respectively. It is, as per definition, required that measures of strain correspond
to suitable measures of stress to give valid expressions of work.
Relation (3) between displacements and strains is obtained from the kinematic assumptions
of a mechanical model. Whether these assumptions are justified can be examined using
a more accurate model or via experiment.

Stability of Structures 1 Introduction

Winter Term 2022/2023 2
Relations (1), (2), and (3) relate applied forces to the displacements caused by these. By
elimination of the strains, we can find a set of equations involving only the unknown displace-
ments.

Any of the relations from (1) to (3) can be linear or nonlinear.

If all of them are linear, then relation (4) will also be linear. In this case, we can always use
linear superposition, e.g., to split up loads in symmetric and anti–symmetric components. The
solutions of analyses of linear systems are always clearly defined.

If one or more of the relations are nonlinear, then relation (4) will be nonlinear. This implies
that the principle of superposition is not valid in general. Furthermore, there might exist
several equilibrium paths to certain problems.

Depending on the degree of linearization, we have the choice between various theories and
methods of computation (see table). On which theory we can base the structural analysis,
and which one is necessary or most advantageous, is determined by the structure itself and its
loadings (i.e., the validity of linearizations in any of the relations (1) to (3)), and by economical
and practical considerations. A higher degree of linearization simplifies the computation, but
at the same time we get a less realistic model of the behaviour of a structure.

Stability of Structures 1 Introduction

Winter Term 2022/2023 3
 Plast. – Plast. / TH.II.O. / plastic or non – linear methods b3
a
Plast. – Plast. / TH.I.O. / plastic or non – linear methods Spl ≤Spl,max

Elast. – Plast. / TH.II.O. / lin. – elast. ev. redistr b2

p Elast. – Plast. / TH.I.O. / lin. – elast. ev. redistr
S≤Smax

b1

non–linear in plastic zones

non–linear in plastic hinges

Elast. – Elast. / TH.II.O.
σ≤σmax
Elast. – Elast. / TH.I.O.

linear in general
(1): Equilibrium
non-linear
linear

linear

linear
(4): Applied Actions –
Displacements

(2): Constitutive Law

d linear
c1 κy , κz , ψ, κω in general
u
v (3): Kinematics
c2
w φ(xi )
non-linear in plastic
ϑ
hinges
Figure 1.2: Applied and internal forces respectively stresses, strains and displacements in the analy-
sis of beam-like structures

Stability of Structures 1 Introduction

Winter Term 2022/2023 4
 Table 1.1: Methods of Analysis in Structural Analysis
Kinematics Const. Law Equilibrium Method of Analysis / Level of theory
(3) (2) (1)
1) l l l First Order Analysis
The equilibrium equations are derived con-
sidering the undeformed configuration.
Mainly used for continuous beams, grid
structures, plane, and space frames sub-
jected to bending; trusses; undercritically
loaded plates, walls, and shell structures.
2) l (χ)1 l nl Second Order Analysis
nl (ϵS )1 The equilibrium equations are derived with
the assumption of small deformations. A
stability analysis reduces to a linear eigen-
value problem, where the eigenvectors rep-
resent the buckling modes, which are de-
termined in shape but not in magni-
tude. The behaviour of systems subjected
to loads above the critical load (post–
buckling behaviour) cannot be examined
Continuous beams, grid structures and
frames with large axial compressive forces;
flat arches; undercritically loaded shear
walls, plates and shell structures that are
amenable to buckling.
3) nl (ϵS , χ)1 l nl Geometrically nonlinear theory with
large deformations but small strains
The equilibrium equations are derived con-
sidering the deformed configuration. It is
possible to examine overcritically loaded
systems (post–buckling behaviour)
Cable structures; overcritically loaded
shear walls, plates, and shell structures
that are amenable to buckling.

Stability of Structures 1 Introduction

Winter Term 2022/2023 5
 4) l nl l First order analysis of physically non-
linear problems; plastic hinges and
yield line theory
Suitable for systems with small deforma-
tions and small strains and a nonlinear
constitutive law. The equilibrium equa-
tions are derived considering the unde-
formed configuration
Mainly used for continuous beams sub-
jected to bending; grid structures and
frames in steel and concrete design, if re-
distributions of internal forces and mo-
ments to the entire structure due to plas-
ticity of the material are to be taken into
account.
5) l (χ) nl nl Second order analysis of physically
nonlinear problems; plastic hinges
and yield line theory
Suitable for systems with small deforma-
tions and small strains and a nonlinear
constitutive law. The equilibrium equa-
tions are derived with the assumption of
small deformations.
Continuous beams, grid structures and (in
particular displaced) frames in steel and
concrete design subjected to large axial
loads, if redistributions of internal forces
and moments to the entire structure due
to the plasticity of the material are to be
taken into account; shear walls, plates, and
shell structures that are amenable to buck-
ling, where a nonlinear material behaviour
is of advantage.

Stability of Structures 1 Introduction

Winter Term 2022/2023 6
 6) nl (χ, ϵS )1 nl nl Geometrically and physically nonlin-
ear theory
Suitable for systems with large deforma-
tions and large strains and a nonlinear con-
stitutive law. The equilibrium equations
are derived considering the deformed con-
figuration. Post–buckling behaviour may
be examined.

1.3 Geometrically and physically non–linear analysis of rigid

body – spring systems

1.3.1 Retrospection: Principle of virtual displacements (example)

1 1
2 2 AV

Figure 1.3: Reference configuration and deformed state of a simple beam

Virtual displacements are imposed to the loaded system in order formulate the equilibrium
with the help of the virtual work theorem.
After cutting the left support and introducing the vertical reaction force AV , the system shows
one single degree of freedom of motion; the deflected beam and the applied loads can rotate
about the right hand pin support as the centre of rotation to a neighboring state.

1
e.g. in frame structures

Stability of Structures 1 Introduction

Winter Term 2022/2023 7
 P δφ b

wA (δφ)
wP (δφ)

AV (δφ)
Figure 1.4: Finite displacement

If we apply a finite angle of rotation, the support reaction and the internal forces will change
while moving from the deformed position to the neighboring state. Hence, we will have a
completely non–linear problem.

Thus, we conclude:

1. To keep the real state of stress constant while imposing virtual displacements, a purposive
virtual displacement needs to be infinitesimally small.

2. If virtual displacements are chosen to be infinitesimally small, kinematic relations concerning

such displacement may be linearised. In the given example, movements of any particular point
of the bar may occur tangential to the real functions of displacements (f. i. circular arcs).

P
δφ

δwA (δφ)
δwP (δφ)

AV

Figure 1.5: Virtual displacement

3. Because of the resulting linear kinematic relations, virtual displacements may be written
in any size of order.

4. Further, the displacements of the structure in the deformed state may be disregarded.

δWi + δWe = 0

Stability of Structures 1 Introduction

Winter Term 2022/2023 8
 − AV δ wA + P̄ δwP = 0

δwP δwA
=
1
2
1

1
δwP = δwA
2
1
− AV δwA + P̄ δwA = 0
2
1
(−AV + P̄ ) δwA = 0
2
1
AV = P̄
2

1.3.2 Hinged Rigid Cantilever with Rotational Spring: Levels of

Approximation

In this chapter, different levels of approximation of a geometrically as well as materially non-

linear analysis are shown on the basis of a simple one–degree–of–freedom structural system.
The presented models of analysis 1–6 correspond to the respective categories of Table 1.1 on
page 5 in Section 1.2. In general, the governing equations for structural systems with more
than one degree of freedom are specified in matrix notation.

Reference configuration Present configuration

(deformed state, instan-
taneous configuration)
fh P
H fv
fv
φ0 φ0 φ
l EI → ∞
EA → ∞ l cos(φ0 + φ)
φ0 + φ
kφ

fh
Figure 1.6: Imperfect structural system

Stability of Structures 1 Introduction

Winter Term 2022/2023 9
 Kinematics: fv = l[1 − cos(φ0 + φ)] (1.1)

fh = l sin(φ0 + φ) (1.2)

Constitutive law: M = kφ φ (1.3)

Equilibrium (P.v.D.): −δW = −δ(Wi + We ) = 0 (1.4)

During an imposed virtual displacement, internal work is performed in the elastic spring and
external work is done by the applied forces P and H.

Because we investigate a one-degree-of-freedom system, we choose a rotation as virtual dis-

placement δφ, which moves the bar from the deformed state to a neighboring state. Because
this virtual movement has to begin from the instantaneous configuration and should be in-
finitesimally small, the state of forces may be kept constant.

−δ(Wi + We ) = kφ φ δφ − P δfv − H δfh = 0 (1.5)

The virtual quantities δφ, δfv and δfh depend upon each other by the prescribed kinematic
relations of the system. Because they are infinitesimal small, they may be calculated as linear
terms of a Taylor series expansion:

df 1 d2 f
f (φ + dφ) = f (φ) + dφ + dφ2 + . . .
dφ 2 dφ2
dfv
δfv = δφ = l sin(φ0 + φ)δφ (1.6)
dφ
dfh
δfh = δφ = l cos(φ0 + φ)δφ (1.7)
dφ

Stability of Structures 1 Introduction

Winter Term 2022/2023 10
 l δφ cos(φ0 + φ)
φ0
φ l δφ sin(φ0 + φ)

δφ

Figure 1.7: Virtual displacement (neighbouring state)

In contrast to real movement, the virtual displacement does not occur on a circular arc but
on a straight line being at right angels to the radius ray (see Section 1.3.1).

Model of Analysis 3: Geometrically nonlinear analysis considering large displacements

kφ φ δφ − P l sin(φ0 + φ) δφ − H l cos(φ0 + φ) δφ = 0 (1.8)

[kφ φ − P l sin(φ0 + φ) − H l cos(φ0 + φ)] δφ = 0

Because of the assumed arbitrary and non-vanishing virtual displacements, the sum of
the terms in brackets vanishes.

kφ φ − P l sin(φ0 + φ) − H l cos(φ0 + φ) = 0

kφ φ − P l sin(φ0 + φ)
=Hl (1.9)
cos(φ0 + φ)

For multi-degree-of-freedom systems, we may write in matrix notation:

a(P, φ, φ0 ) = P

If we allow only vertical loads, the above relation (1.9) converts into:

kφ φ − P l sin(φ0 + φ) = 0 (1.10)

kφ φ
P = (1.11)
l sin(φ0 + φ)

Discussion of equation (1.11) :

Stability of Structures 1 Introduction

Winter Term 2022/2023 11
 dP kφ [sin(φ0 + φ) − φ cos(φ0 + φ)]
= (1.12)
dφ l sin2 (φ0 + φ)

1. Imperfect system with initial tilt angle φ0 ̸= 0

P (φ = 0) = 0

dP kφ
(φ = 0) =
dφ l sin φ0

Transition to a perfect reference state (φ0 → 0):

dP
→∞
dφ

2. Perfect system without initial tilt angle: φ0 = 0

kφ φ
P = (1.13)
l sin φ
kφ φ kφ kφ
lim P (φ) = lim = lim =
φ→0 φ→0 l sin φ φ→0 l cos φ l

dP kφ sin φ − φ cos φ
=
dφ l sin2 φ

dP kφ sin φ − φ cos φ
lim = lim
φ→0 dφ φ→0 l sin2 φ
kφ cos φ − (cos φ − φ sin φ)
= lim
φ→0 l 2 sin φ cos φ
kφ φ
= lim =0
φ→0 2 l cos φ

Model of Analysis 2: Linearization for the purpose of a second order theory of small displace-
ments

Assumption: φ, φ0 small, → sin(φ0 + φ) → φ0 + φ

→ cos(φ0 + φ) → 1

Stability of Structures 1 Introduction

Winter Term 2022/2023 12
 Hence, eq.(1.9) reduces to:

kφ φ − P l (φ0 + φ) − H l = 0

(kφ − P l)φ − (H l + P l φ0 ) = 0 (1.14)

For multi-degree-of-freedom systems, we may write in matrix notation:

A(P)V = P

The solution of second order theory problems, in general, is well-defined. However,

because of the initially introduced assumptions, the results are only valid in a limited
range.

It has to be seen that A depends upon the applied loads.

In particular: H = 0

(kφ − P l)φ = P l φ0 A(P)V = P (1.15)

kφ φ
P = (1.16)
l (φ0 + φ)

dP kφ (φ + φ0 ) − φ kφ φ0
= =
dφ l (φ + φ0 ) 2 l (φ0 + φ)2

φ=0: P =0

dP kφ kφ
= <
dφ l φ0 l sin φ0

Perfect structure: φ0 = 0

(kφ − P l) φ = 0 A(P)V = 0 (1.17)

We get an Eigenvalue problem. Displacements remain undetermined:

kφ
P crit = (1.18)
l

Model of Analysis 1: Analysis without taking displacements into account (first order analy-
sis).

Stability of Structures 1 Introduction

Winter Term 2022/2023 13
 Assumption: φ0 = 0;
The applied load P does not affect the equilibrium of moments.

kφ φ − H l = 0 KV = P (1.19)

Influence of non-linear material behaviour:

Model of Analysis 6: Geometrically and materially nonlinear analysis with large displace-
ments.

kφ (φ) φ − P l sin(φ0 + φ) − H l cos(φ0 + φ) = 0 (1.20)

A(P, φ, φ0 ) = P

Model of Analysis 5: Linearization for a second order theory of small displacements, taking
into account material non-linearities

(kφ (φ) − P l) φ − (H l + P l φ0 ) = 0 (1.21)

K(P, φ) V = P(φ0 )

Model of Analysis 4: Materially nonlinear first order analysis

kφ (φ) φ − H l = 0 (1.22)

K(φ) V = P

Stability of Structures 1 Introduction

Winter Term 2022/2023 14
 P̄ klφ

P = kφ φ
l sin(φ)

P = kφ φ
l sin(φ0 +φ)

P = kφ
l

P = kφ φ
l φ0 +φ

(a) φ = 0 − 2 π

P̄ klφ

(b) φ = 0 − 58 π
Figure 1.8: Results for P̄ sketched for different models of analysis

Stability of Structures 1 Introduction

Winter Term 2022/2023 15
1.4 Snap-through problems

1.4.1 Introductory Remarks

EA → ∞

EA

EA

Figure 1.9: Examples: Shallow arches, shallow shells

1.4.2 Kinematics, Constitutive Law

P
2

w
l h
dα l + ∆l
α0 α

Figure 1.10: Kinematics

Stability of Structures 1 Introduction

Winter Term 2022/2023 16
 (h − w)2 + a2 = (l + ∆l)2
q
a2 + (h − w)2 = l (1 + ∆l
l
) = l (1 + ε)
q
ε(w) = 1
l
a2 + (h − w)2 − 1

Constitutive Law

1q 2
 
σ = εE = a + (h − w)2 − 1 E
l

1.4.3 Equilibrium

P.V.D.:
Z
σ δε dV − P δw = 0

P , σ: applied load and stresses in the deformed state; they are assumed to be constant on
virtual displacements and virtual strains
δε, δw: infinitesimally small, starting from the deformed state, which is equilibrated
Virtual displacement:

dε
δε = δw
dw
1 1h 2 i− 1
= a + (h − w)2 2 2 (h − w) (−1) δw
l 2
h−w
=− q δw
l a2 + (h − w)2

Stability of Structures 1 Introduction

Winter Term 2022/2023 17
 P
2
sin α = h−w
l+∆l
δw · sin α

l l + ∆l δw
h α

h−w
∆α δα
α

Figure 1.11: Illustrative representation of the virtual displacements

From Fig. 1.11, we read (bearing in mind that the elongation of the bars is related to the
unstrained length):

δw sin α
−δε =
l
δw (h − w)
=
l (l + ∆l)
h−w
= q δw compare above
l a2 + (h − w)2

Equilibrium in the deformed state (P.V.D.):

Z Z 
1q 2 −(h − w)

2 a + (h − w)2 − 1 E q δw dA dx − P δw = 0
l l a2 + (h − w)2

For any arbitrary δw we conclude:

 
h−w  l
2 EA −1 + q =P
l a2 + (h − w)2
" #
h−w l
2 EA −1 + √ 2 =P
l l − 2 h w + w2

Stability of Structures 1 Introduction

Winter Term 2022/2023 18
 P
+P +P
compressive force

2 snap through
P =0

w
tensile force

snap back
P =0
−P
−P P =0

Figure 1.12: Interdependence of applied load and vertex displacement

Stability of Structures 1 Introduction

Winter Term 2022/2023 19
2 Second Order Analysis and Stability of
Plane Frames

2.1 Preliminaries and Assumptions

We shall assume that

1. The central axis (x-axis) of a beam coincides with its centroidal axis.
2. The axes of a cross section (y- and z-axes) are principal axes. Moreover, the x- and
z-axes lie within the plane of the structure.
3. The structure consisting of individual beams, or the structural system itself, may have
imperfections according to the pertinent standards (indicated by the subscript 0).
4. Local failures or out-of-plane failures are either prevented by constructive measures or
separate analysis is performed.
5. The effects of strains caused by shearing stresses or shear forces are negligible (Bernoulli
hypothesis).
6. Linear relations according to the engineering beam theory are assumed between strains
(axial strains and curvatures) of the central axis and its displacement.
Hence, we will develop a theory of small displacements.
7. A homogeneous and isotropic material is assumed. Furthermore, according to the en-
gineering beam theory, the normal stresses are governed by a one-dimensional linear
constitutive law.
Thus, only small strains are admissible.
We will discuss, however, the effects of a nonlinear constitutive law in Elastic-Plastic Analysis of Fr
.
8. The structural system under investigation is generally free of stresses in the undeformed
configuration.
9. The loads act within the plane of the structure.
10. The loads do not change their directions while causing structural deformations.

Stability of Structures
Winter Term 2022/2023
20
 11. Any distributed axial load may be applied on a beam, which may cause a varying normal
force diagram in a first order analysis. However, for the iteration process of a second
order analysis, we will assume a constant (possibly averaged) normal force in a beam
defined by two nodes.

2.2 Basic Relations of Beam Columns with variable

Cross-Sections on Elastic Spring Beds

2.2.1 The Governing Equation of Bending

Here, the relevant relations of bending are compiled. We confine our interest to a second order
theory of bending because we assume that the basic relations for the elongation of bars under
normal forces hold unchanged.
Kinematic relations:

φ = −w′ (2.1)

κ = φ′ (2.2)

Constitutive law:

M = EI(κ − κ) (2.3)

∆T
κ = αT + κS + κK + κpl (2.4)
h

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 21
 my (x)

pz (x)
x
w0 (x) w0 (x + dx)
z, w0 , w w(x)
w(x + dx)
L
M (x) T (x) L
T (x + dx) M (x + dx)

kw (x) · w(x)

kφ w′ (x)

Equilibrium:

L′ = 0 (2.5)

T ′ − kw w + pz = 0 (2.6)

M ′ − T + L(w′ + w′0 ) + kφ w′ + my = 0 (2.7)

Taking the derivative of (2.7) with respect to x and substituting (2.6) yields

M ′′ + L(w′′ + w′′0 ) + kφ w′′ − kw w + pz + m′y = 0 (2.8)

Normal and shearing forces and longitudinal, respectively transverse forces (within the range
of validity of the 2nd order analysis for small quantities w′ + w′0 and with |(w′ + w′0 ) · T | ≪ |L|
and |(w′ + w′0 ) · Q| ≪ |N |(Q ≪ N and T ≪ L)):

N ≈L (2.9)

T ≈ Q + (w′ + w′0 ) · N (2.10)

Q ≈ T − (w′ + w′0 ) · L (2.11)

Relevant quantities:

φ = −w′ (2.12)

M = −EI(w′′ + κ) (2.13)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 22
 T = −EIw′′′ − EI ′ w′′ + (L + kφ )w′ + my + Lw′0 − EIκ′ − EI ′ κ (2.14)

Q = −EIw′′′ − EI ′ w′′ + kφ w′ + my − EIκ′ − EI ′ κ (2.15)

Governing equation:

(EIw′′ )′′ − (L + kφ ) w′′ + kw w − pz − m′y − L w′′0 + (EIκ)′′ = 0 (2.16)

Remark 1: Extreme values of the bending moment:

!
M ′ = −EIw′′′ − EI ′ w′′ − EIκ′ − EI ′ κ = Q(x) − kφ w′ − my = 0

For beams without a spring bed and without a distributed bending moment, the extreme value
of the bending moment M (x) coincides with zero values of the shearing force Q(x).
Remark 2: All terms of eq. (2.16) have units of a distributed force pz : [force/length]
Thus, this relation describes the statically indeterminate split of pz (as well as of the quantities
m′y , L w′′0 and (EIκ)′′ ) to the different components of the resistance of the structural member:

(EIw′′ )′′ (Bending), − Lw′′ (Th.II.O.), kφ w′′ (spring bed), kw w (spring bed).

Remark 3: Concerning the mechanical meaning of destabilizing forces of a second order theory:

N w′′

N w′ N x
′
w

w′ + w′′ dx + . . . N (w′ + w′′ dx + . . .)

N
dx

Remark 4: Normal, respectively shearing forces and longitudinal, respectively transverse

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 23
forces:

cos arctan(w′ + w′0 ) − sin arctan(w′ + w′0 )

    
Q T
  =  
N sin arctan(w′ + w′0 ) cos arctan(w′ + w′0 ) L

cos arctan(w′ + w′0 ) sin arctan(w′ + w′0 )

    
T Q
  =  
L − sin arctan(w + w0 ) cos arctan(w + w0 ) N
′ ′ ′ ′

For small arguments, the relations

sin(w′ + w′0 ) ≈ w′ + w′0 and cos(w′ + w′0 ) ≈ 1

may hold:

1 −(w′ + w′0 )
    
Q T
 ≈ ′  
N w + w′0 1 L

1 w′ + w′0
    
T Q
  ≈  
L −(w′ + w′0 ) 1 N

Thus, with the additional assumptions |(w′ + w′0 ) · T | ≪ |L| and |(w′ + w′0 ) · Q| ≪ |N |
(Q ≪ N und T ≪ L), we have:

N ≈L (2.17)

T ≈ Q + (w′ + w′0 ) · N (2.18)

Q ≈ T − (w′ + w′0 ) · L (2.19)

2.2.2 Element Matrices on the Basis of exact solutions

Stiffness matrices, vectors of fixed end forces and moments, transfer matrices, and load vectors
are denoted as element matrices in this context.
To determine the stiffness matrix of a bar we only need the homogeneous part of eq. (2.16):

(EIw′′ )′′ − (L + kφ )w′′ + kw w = 0 (2.20)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 24
The solution of this linear, ordinary fourth order differential equation (with coefficients, in
general depending upon x) may be written in the following manner:

w(x) = C1 w1 (x) + C2 w2 (x) + C3 w3 (x) + C4 w4 (x) (2.21)

With the denotations

NF (x) = [w1 (x) w2 (x) w3 (x) w4 (x)] (2.22)

cT = [C1 C2 C3 C4 ] (2.23)

we may write in matricial notation:

w(x) = NF c (2.24)

The functions of the row matrix NF represent a fundamental set of solutions of eq. (2.20).
Hence, we can present the displacements and stress resultants of equations (2.12), (2.13), and
(2.14) from page 22 as functions of the constant column matrix c (the entries of which can
not be interpreted in a mechanical sense).
    
w(x) NF C1
    
−N′F
 
 φ(x)    C 
  2 Nu
= =
  
     c
 T (x)  −EI NF − EI NF + (L + kφ )NF 
′′′ ′ ′′ ′  C  Ns
    
     3
M (x) −EI N′′F C4

z(x) = A(x) · c (2.25)

These four constants in (2.25) can be expressed in dependence of the four displacements and
rotations of the member ends.

w(xa ) NF (xa )
      
wa C1
−w ′ (x ) −N′ (x ) C  Nu (xa )
        
φ 
 a a  F a   2
= = = v = [G]−1 c (2.26)
 
        c
 wb   w(xb )   NF (xb )  C3  Nu (xb )
      
      
φb −w′ (xb ) −N′F (xb ) C4

From this follows:

c = Gv (2.27)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 25
 Hence: w(x) = NF G v (2.28)

With: N (x) = NF G (2.29)

w(x) = N v (2.30)

z(x) = A(x) G v (2.31)

The functions of the row matrix NF are the exact shape functions of the member.

We obtain the exact stiffness matrix in formulating the transverse forces and bending moments
at the ends of the element (in sign convention 2):

−T (xa )
   
Ta
−M (x ) −Ns (xa ) −Ns (xa )
       
M 
 a a 
= = G v=k v =⇒  G=k (2.32)

     
 Tb   T (xb )  Ns (xb ) Ns (xb )
   
   
Mb M (xb )

To remember: A column of the stiffness matrix contains the end forces and end moments if
you set the related end displacement equal to +1 and the rest of them are 0.

My

Pz sign convention 2

my1 my2

pz1 pz2
φa φb

z
wa wb Ma Ta Tb Mb

ls

a l

Positive sense of direction of end displacements, loads and end forces and moments.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 26
2.2.3 Relevant Differential Equations and their solutions

2.2.3.1 Beam Column, EI = const., N = const. < 0, kw = 0

Governing equation:

EIw′′′′ + |N |w′′ = pz + m′y + N w′′0 − EIκ′′ (2.33)

Solution of the homogeneous PDE:

wh (x) = C̃1 1 + C̃2 x + C̃3 sin ε̃x + C̃4 cos ε̃x (2.34)
s
|N |
ε̃ = (2.35)
EI

Remark 1: If we introduce dimensionless coordinates ξ = x

l
and use d(...)
dξ
= (. . .)′ we can
rewrite eq. (2.34):

EI w′′′′ + |N | l2 w′′ = l4 pz + l3 m′y + N l2 w′′0 − EI l2 κ′′ (2.36)

wh (ξ) = C1 1 + C2 ξ + C3 sin εξ + C4 cos εξ (2.37)

s
|N |
Characteristic parameter: ε=l (2.38)
EI

Remark 2: Expression (2.38) denotes the only mechanical descriptive quantity for the charac-
teristic parameter. (The same result is obtained by inserting ε̃ = εl in (2.34).)
Remark 3: Stiffness matrix in a local coordinate system:

EA EA
 
0 0 − 0 0
 l l 
EI EI EI EI
 
0 [2(A′ +B ′ )−D′ ] 3 −(A′ +B ′ ) 2 0 −[2(A′ +B ′ )−D′ ] 3 −(A′ +B ′ ) 2
 
 
 l l l l 
′ EI ′ EI ′ EI ′ EI
 
′ ′
 0 −(A +B ) 2 A 0 (A +B ) 2 B 
l l l l
 
k =
 

 EA EA 
 − 0 0 0 0 
 l l 
EI EI EI EI
 
−[2(A′ +B ′ )−D′ ] 3 (A′ +B ′ ) 2 [2(A′ +B ′ ) − D′ ] 3 (A′ +B ′ ) 2
 
 0 0 
 l l l l 
′ EI ′ EI ′ EI ′ EI
 
′ ′
0 −(A +B ) 2 B 0 (A +B ) 2 A
l l l l
(2.39)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 27
 ε (sin ε − ε cos ε) A′ 2 − B ′ 2 ε2 sin ε
A =
′
C =
′
=
2 (1 − cos ε) − ε sin ε A′ sin ε − ε cos ε

ε (ε − sin ε) |N |
B′ = D ′ = ε2 = l 2
2 (1 − cos ε) − ε sin ε EI

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 28
Remark 3: Global and local stability failure in thin walled beam columns
Px
Critical load (Euler - case 2)

π 2 EI
Pcrit =
l2

Pz

Px
stress problem

Moment (2nd order theory)

− Normal Force (≈ L = −P x )

Cross section and distribution of

normal stresses caused by Normal
Forces and Bending Moment

Top view on a part of the flange

(pressure):
The linearly distributed stresses can
be collected in line-forces and
line-moments.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 29
 In case of sufficient compressive
forces the destabilizing forces
Pz
concerning 2nd order analysis can
result in reasonable displacements
perpendicular to the center-plane of
the flange (buckling). The flanges
act as elastically bedded (across the
web) plates.

2.2.3.2 Beam on a Continuous Elastic Support;

EI = const., N = 0, kw > 0 (Winkler E., 1867)

2.2.3.3 Beam Column on a Continuous Elastic Support;

EI = const.; N := const. < 0, kw = : const. > 0

Governing equation:

EIw′′′′ + |N | w′′ + kw w = pz + m′y − EIκ′′ (2.40)

Solution of the homogeneous d.e., characterizing parameters:

v q 
|N |
u
kw
u
EI
± 2 EI
1. N 2 < 4 EI kw ε1,2 = (2.41)
t
2

wh (x) = C1 e−ε2 x cos ε1 x + C2 e−ε2 x sin ε1 x + C3 eε2 x cos ε1 x + C4 eε2 x sin ε1 x (2.42)

Boundary disturbances of a beam column on a continuous elastic support are fading if the bar
is sufficiently long:

P
e

If N vanishes, the characteristic parameters are taking the form of the beam on a spring bed.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 30
2. In the case of
v q 
|N |
u
kw
u
EI
− 2 EI
ε2 = =0 (2.43)
t
2

we obtain the critical normal force of the beam column:

s
kw
N 2 = 4 EI kw ε1 = ε = (2.44)
4

EI

The exponential terms e±ε2 x take a value of 1 and we receive the homogenous solution:

wh (x) = C1 cos εx + C2 sin εx + C3 x cos εx + C4 x sin εx (2.45)

s √
|N | ± N 2 − 4 EI kw
3. N > 4 EI kw
2
ε1,2 = (2.46)
2 EI

wh (x) = C1 cos ε1 x + C2 sin ε1 x + C3 cos ε2 x + C4 sin ε2 x (2.47)

Remark 1: We can derive the critical compressive force of a simple beam column on hinge
supports in a different approach:

w(0) = 0 w(l) = 0

My (0) = −EIy w′′ (0) = 0 My (l) = −EIy w′′ (l) = 0

nπx
Trial function: w = C sin (2.48)
l
π nπx
w′ = C n cos (2.49)
l l
2
π nπx

w = −C n
′′
sin (2.50)
l l

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 31
 3
π nπx

w = −C n
′′′
cos (2.51)
l l
4
π nπx

w ′′′′
=C n sin (2.52)
l l

All boundary conditions are satisfied. If we substitute the above trial function in the governing
differential equation, we have:
4 2
nπ nπx nπ nπx nπx
 
EI C sin − |N | C sin + kw C sin =0
l l l l l
4 2
nπ nπ
 
EI − |N | + kw = 0 (2.53)
l l

Hence,
" #
l2 n4 π 4 n2 π 2 l2
Pcrit = EI +kw = EI + kw
n2 π 2 l4 l2 n2 π 2
s !2 #
1
" # "
l4 kw EI l4 kw EI
= n2 + 2 4 π2 2 = n2 + π2
n π EI l nπ 2 EI l2
EI
= φ28 π 2 (2.54)
l2

 s 2
1 l4 kw 
φ28 (n) = n2 +  2 (2.55)
nπ EI

(Denotation according to A. Pflüger, Stabilitätsprobleme der Elastostatik S.364)

Eq. (2.54) yields Pcrit as a function of n. Extreme values of Pcrit depend upon n:
s
2nπ 2 2l2 1
 2
∂Pcrit n kw
= 2 EI − 3 2 kw = 0 Hence: = 2 (2.56)
∂n l nπ l π EI
q
Pcrit = 2 kw EI (2.57)

Critical loads of a hinge supported beam column on a spring bed.

Remark 3: Examples: - piles under compression - top and bottom layers of sandwich beams
- Analogous: axisymmetrically loaded cylindrical shells with constant thickness (1st and 2nd
order theory)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 32
 φ28

n=4
n=5

n=3

 q 2
n = 2 φ28 (n) = n2 + 1
n·π 2
l4 kw
EI

n=1 q
l 4 kw
EI

q
l4 kw
Figure 2.1: The system parameters l, kw , EI are included in the term EI on the x-axis. On the
y -axis there is the critical load Pcrit =φ28 π 2 EI
l2in dependency of the parameter φ28 .
For a given system, corresponding to one point on the x-axis, one can determine the
critical load of the system by reading the lowest value of φ28 , given by the red envelope
curve. Additionally q on can deduce the failure mode n that occurs for this combination of
4
l, kw , EI . E.g. for l EI
kw
= 40 the system will collapse with n = 2 sinus half waves, as
Pcrit for n = 2 is smaller as Pcrit for n = 1.

2.2.3.4 Beam in Tension, EI = const.; N = const. > 0; kw = 0

2.2.3.5 Beam in Tension on a Continuous Elastic Support,

EI = const.; N = const. > 0; kw > 0

2.2.3.6 Beam with Constant Width and Linearly Varying Height, EI = var.; N =0; kw = 0

b a a l b b

x
x h = ha (2.58)
ha x ξ hb a

x3
EI(x) = EIa (2.59)
a3
Ma Mb
Qa Qb

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 33
In addition to the usual assumptions of a technical theory of bending and torsion of prismatic
beams, we adopt that the rate of change of characteristic parameters in dependence upon x
are so small that we may apply a one-dimensional beam model for the structural analysis.
In this context, we assume for the relevant dimension h(x) that the following relation holds:

d 1 d
g= h(x) = h(x) ≤ 0.2
dx l dξ

Governing equation:

EIa h 3 ′′′′ i
3
x w +6x 2 ′′′
w +6x w ′′
= pz + m′y −[EI(x) κ]′′ (2.60)
a

Solution of the homogeneous ODE:

1
wh (x) = C1 1+ C2 x+ C3 + C4 ln x (2.61)
x

Remark 1: Stiffness matrix (bending) for a beam of constant width and linearly varying depth
(linear theory of 1st order):

l
α= β =1+α η = (2 + α) ln β − 2α
a

α3 (α+2) α3 α3 (α+2) β α3
 
− 2 − −
l3 l l3 l2
  

Ta
   wa
2β ln β − 3α − 2α
2 2
α 3
β (2β ln β − α2 − 2α)
 
  
 
 Ma  EIa
 −  ϕa 
 
=

l l2 l  
α (α+2)
  
3
β α3
 
 Tb 
  η 


 wb 

 
l3 l2

Mb  
ϕb
β 2 (2 ln β − α (2−α))
 
 
symm.
l
p = k Iexakt v

Further examples of beams with varying cross-sections:

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 34
 a b

h ba x x bb h

ba a l bb

a b

ba ha x x bb hb

ba a l bb

a b

ha x x bb hb

d ah
ba ah l bb

a b tF
tF
ts ts ha x x ξ bb hb ts ts
tF
tF
d ah
ba ah l bb

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 35
 a b
AF
AF
ha x x hb
AF
AF
s ah l s

2.2.3.7 Beam Column with Constant Width and Linearly Varying Height,
EI = var.; N = const.>0; kw = 0

x
h = ha (2.62)
a
x3
EI(x) = EIa (2.63)
a3

Governing equation:

EIa h 3 ′′′′ i
x w + 6x 2 ′′′
w + 6x w ′′
− N w′′ = pz + m′y − [EI(x) κ]′′ (2.64)
a3

Solution of the homogeneous ODE:

√ 2ε √ 2ε
! !
wh (x) = C1 1 + C2 x + C3 x J1 √ + C4 x N1 √ (2.65)
x x

s
|N |
Characteristic parameter: ε= a3 (2.66)
EIa

Bessel and Neumann functions of the first order compared with trigonometric functions:

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 36
 J1 (x) (Bessel function of 1st order)

N1 (x) (Neumann function of 1st order)

sin(x)

cos(x)

Figure 2.2: Fundamental System

Appendix 2.2.3.4 Beam in Bending and Tension,

EI = const.; N = const. > 0; kw = 0

Governing equation:

EI w′′′′ − N w′′ = pz + m′y + N w′′0 − EI κ′′ (2.67)

Solution of the homogeneous PDE:

εx εx
wh (x) = C1 1 + C2 x + C3 sinh + C4 cosh (2.68)
l l

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 37
 s
|N |
Characteristic parameter: ε=l (2.69)
EI

Appendix 2.2.3.5 Beam on a continuous elastic support in Bending and Tension,

EI = konst; N = const. > 0; kw > 0

Governing equation:

EI w′′′′ − N w′′ + kw w = pz + m′y − EI κ′′ (2.70)

Solution of the homogeneous ODE:

v q 
u
kw N
u
EI
± 2EI
N 2 < 4EI kw ε1,2 = (2.71)
t
2

wh (x) = C1 e−ε1 x cos ε2 x + C2 e−ε1 x sin ε2 x + C3 eε1 x cos ε2 x + C4 eε1 x sin ε2 x (2.72)

s s
kw N
N = 4EI kw ε= = (2.73)
2 4

EI 2 EI

wh (x) = C1 cosh εx + C2 sinh εx + C3 x cosh εx + C4 x sinh εx (2.74)

v
√ 
− 4EI kw
u
u |N | ± N 2
N 2 > 4EI kw ε1,2 = (2.75)
t
2EI

wh (x) = C̃1 cosh ε1 x + C̃2 sinh ε1 x + C̃3 cosh ε2 x + C̃4 sinh ε2 x (2.76)

= C1 eε1 x + C2 eε2 x + C3 e−ε1 x + C4 e−ε2 x (2.77)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 38
2.2.4 Derivation of the P.V.D. starting with the Governing Differential
Equation

We start with equation (2.16) (Govering Differential Equation of bending of a beam acc. Th
II. Order with elastic bedding):
Z
(EIw ) − (L + kφ ) w + kw w − pz −
′′ ′′ ′′
m′y − L w′′0 + (EIκ) = 0 | · δw(x)|
′′
. . . dx
Z bh i
(EIw′′ )′′ − (L + kφ ) w′′ + kw w − pz − m′y − L w′′0 + (EIκ)′′ δw(x)dx = 0
a

Partial Integration:
Z b Z b
g δw dx = −
′
g δw′ dx + [g δw]ba
a a
Z b Z b Z b
g IV δw dx = − g III δwI dx + [g III δw]ba = g II δwII dx − [g II δwI ]ba + [g III δw]ba
a a a

Z
EI w′′ δw′′ dx − [EI w′′ δw′ ]ba + [(EI w′′ )′ δw]ba
Z
+ L w′ δw′ dx − [L w′ δw]ba
Z
+ kφ w′ δw′ dx − [kφ w′ δw]ba

Z Z Z
+ kw w δw dx − pz δw dx + my δw′ dx − [my δw]ba
Z
+ EI κ δw′′ dx − [EI κ δw′ ]ba + [(EI κ)′ δw]ba
Z
+ L w′0 δw′ dx − [L w′0 δw]ba = 0

Z
[EI w′′ δw′′ + (L + kφ ) w′ δw′ + kw w δw] dx
Z
+ [−pz δw + (my + L w′0 ) δw′ + EI κ δw′′ ] dx

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 39
 b
+ [{(EI w′′ )′ − (L + kφ ) w′ − my − L w′0 + (EI κ)′ } δw]a + [{−EI w′′ − EI κ} δw′ ]ba = 0

Z
− [EI w′′ δw′′ + (L + kφ ) w′ δw′ + kw w δw] dx
Z
b
+ [pz δw − (my + L w′0 ) δw′ − EI κ δw′′ ] dx + [T (x) δw(x)]ba − [M (x) δw′ (x)]a = 0

. . . . . . − T (a) δw(a) + T (b) δw(b) + M (a) δw′ (a) − M (b) δw′ (b) = 0

Z
− [EI w′′ δw′′ + (L + kφ ) w′ δw′ + kw w δw] dx
Z φa φb
+ [pz δw − (my + L w′0 ) δw′ − EI κ δw ] dx
′′

wa wb Ma Ta Tb Mb
+Ta δwa + Tb δwb + Ma δφa + Mb δφb = 0 (2.78)

2.2.5 Element Matrices for Bending on the Basis of Cubic Test Functions
(Finite Element Method)

Starting from eq. (2.78) (weak formulation) a finite element should be derived for a beam
column with a varying height on an elastic support. As test functions we choose the shape
functions of a prismatic (uniform) beam without elastic support.

w(x) = N v (2.79) w′ (x) = N′ v (2.80) κ(x) = −w′′ = −N′′ v (2.81)

δw(x) = N δv (2.82) δw′ = N′ δv (2.83) δw′′ (x) = −N′′ δv (2.84)

vT = [wa φa wb φb ] (2.85) pT = [Ta Ma Tb Mb ] (2.86)

Z
− [EI w′′ δw′′ + (L + kφ ) w′ δw′ + kw w δw] dx
Z
+ [pz δw − (my + L w′0 ) δw′ − EI κ δw′′ ] dx

+ Ta δwa + Tb δwb + Ma δφa + Mb δφb = 0

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 40
 Z h i
− δvT N′′T EI(x) N′′ v + δvT N′T (L + kφ ) N′ v + δvT NT kw (x) N v dx
Z h i
+ δvT NT pz (x) − δvT N′T (my (x) + L · w0 (x)′ ) + δvT N′′T EI κ(x) dx

+ δvT p = 0 (2.87)

Z
kp + kgk = N′′T EI(x) N′′ dx (2.88)
Z
kgL =L N′T N′ dx (2.89)
Z
kgφ = kφ N′T N′ dx (2.90)
Z
kgw = NT kw (x) N dx (2.91)

Z
pp = pz (x) NT dx (2.92)
Z
pm =− my (x) N′T dx (2.93)
Z
pL = −L w0 (x)′ N′T dx (2.94)
Z
pκ = EI κ(x) N′′T dx (2.95)

n o
δvT − [kp + kgk + kgL + kgφ + kgw ] v + pp + pm + pL + pκ + p = 0

− [kp + kgk + kgL + kgφ + kgw ] v + pp + pm + pL + pκ + p = 0 (2.96)

k = kp + kgk + kgL + kgφ + kgw (2.97)

p0 = −(pp + pm + pL + pκ ) (2.98)

p = k v + p0 (2.99)

Evalutation of eqs. (2.89) to (2.91) yields the desired matrices based on cubic form functions,
for instance:

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 41
 36 −3 · l −36 −3 · l
 
L
kgL = kl
4 · l2 3·l
 
30 · l  −3 · l −l2 
kl =
 
 
−36 3·l 36 3·l
 
kφ
kgφ =
 
kl  
30 · l −3 · l −l 2
3·l 4 · l2

Derivation of ”destabilizing forces” linked with kgL :

From: p = (kp + kgL ) v

φa φb
neglecting the influence of kp :
wa wb
L
Ta = (36 wa − 3 l φa − 36 wb − 3 l φb )
30l

L
wb − wa Ta = (36 wa − 3 l φa − 36 wb − 3 l φb )
φa = − ; φb = φa ; 30l
l
L wb − wa wb − wa
 
= 36 wa + 3 l − 36 wb + 3 l
30l l l
+L
+L L
Ta = (36 wa + 3 wb − 3 wa − 36 wb + 3 wb − 3 wa )
30l
Tb L wb − wa
= (30 wa − 30 wb ) = −L · ;
30l l

Stiffness matrix kgw (produced by a continuous elastic support):

156 −22 · l 54 13 · l
 

4 · l2 −13 · l −3 · l2
 
 −22 · l 
kw · l  
kgw =
 
420  54 −13 · l 156 22 · l
 

 
13 · l −3 · l2 22 · l 4 · l2
 

For a beam of constant width and linearly varying height, we get the following for the matrices
kp + kgk :

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 42
(with µ = h(a)
h(a+l)
, α= 1−µ
µ
and b = const. , compare Section 2.2.3.6) :

kp+gk = kp + kgk

 12
− l62 − 12 − l62

l3 l3
 
 4 6 2 
l l2 l
kp =
 
EIa 
 12 6


l3 l2
 
 
4
l

kgk = EIa ·
         
12 15
 l3 10 α + 65 α2 + 7 3
20 α − l62 α+ 7 2
10 α + 51 α3 − 12
l3
15
10 α + 65 α2 + 7 3
20 α − l62 2α + 17 2
10 α + 12 α3
       
4 3 2 2 1 3 6 7 2 1 3 2 3 13 2 2 3
4α + 5α + 10 α α+ 10 α + 5α 2α
+ 10 α
+ 5α
 

 l l2 l 
    
12 15
+ 65 α2 + 7 3 6
2α + 17 2 1 3
 
l3 10 α 20 α l2 10 α + 2 α
 
 
   
4 9 19 2 11 3
l 4α + 10 α + 20 α

2.3 Analysis and Structural Behavior of Plane Frames

(Second Order Analysis)

2.3.1 Methods of Analysis

The following methods are available for a first/second order analysis of plane frames:
1. Differential Equation Methods
These methods directly solve the governing differential equations by substituting a suitable
fundamental solution. The constants have generally no mechanical interpretation and are
determined by boundary conditions. This approach is only limited to very simple structural
systems and therefore not recommended in engineering practice.
2. Iterative Hand Methods
These methods are only applicable for simple systems (e.g., cantilever beams with a vary-
ing cross section). (See e.g., Chr. Petersen, Statik und Stabilität der Baukonstruktionen,
Chapters 3.1. (Verfahren iterativer Annäherung) and 3.2. (Verfahren approximativer Verfor-
mungsaffinität)).

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 43
3. The Force Method for Plane Frames
The force method was extended systematically in the papers of J. Andelfinger (Das erweiterte
Kraftgrößenverfahren zur Berechnung von Stabilitätsproblemen ebener Tragsysteme, 1964)
and G. Hees (1975) for stability problems and second order analyses. It is not suitable for
computers and requires additional considerations for systems undergoing side sway. Hence,
the force method is not recommended in engineering practice.
4. The Transfer Matrix Method
As for a first order analysis, this method is restricted to continuous beam systems (e.g.,
continuous columns over several stories of a building), and therefore is of minor importance
as an independent method. However, the matrix method (transfer matrix and load vectors)
is of great importance in combination with analysis programs for beam structures (that use
exact stiffness matrices). This algorithm is advantageous to compute the stiffness matrices
and fixed end forces and moments of combined structural members in a first pass, and the
internal forces and moments over the length of a member in a second pass.
5. The Rotation Method
This is a modified version of the general stiffness method, with exact stiffness matrices and
the assumption of infinitesimally high axial stiffness (EA ⇒ ∞). For hand use only.
6. General Displacement (Stiffness) Method with Exact Stiffness Matrices
7. Finite Element Method with Geometrical Stiffness Matrices for Destabilizing Forces of a
Second Order Analysis, for Elastic Foundations and for Varying Cross Sections
Plane beam structures, with loads and supports within the plane of the structure (referred
to as plane frames), have an exceptional position in structural mechanics. Provided that the
assumptions of the mechanical model employed hold (for individual members e.g. Bernoulli
beam, Timoshenko beam, beam on elastic foundation, second order theory of beams under
tension or compression), the mathematically exact solutions to the differential or integral for-
mulations describing the mechanical problem are already known (the integral virtual work
formulation and the corresponding system of differential equations are mathematically equiv-
alent) and discussed in detail. Therefore, the methods mentioned in 6. and 7. are commonly
implemented in analysis programs for beam structures used in engineering practice.
Now, we shall investigate the nature of the solutions obtained from a second order analysis,
both for the use of approximate (according to the finite element method) and exact relations.

2.3.2 Advantages and Disadvantages of Exact and Approximate Stiffness

Matrices

The destabilizing forces of a geometrically nonlinear beam theory according to a second order
analysis may be included in the local relations of a beam element with exact or approximate

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 44
stiffness matrices. The latter matrices may be obtained, e.g., as the second term of a series
expansion of the exact stiffness matrices as a function of the normal force or the buckling
coefficient in an element. The first term of the series represents the stiffness matrix of a
geometrically linear first order theory. Both the approaches have theoretical and practical
advantages and disadvantages, which we shall briefly discuss in the following:
The advantages of an exact solution are

1. The theoretical error of the element formulations within the scope of the underlying
mechanical model is zero for all quantities.
2. It follows that in the discretization process, every member defined by the geometry of
the structure can be assembled into the calculation model as one single element. The
global results are theoretically exact and independent of how the structure is divided
into its elements. However, this requires that not only the stiffness matrices but also the
fixed-end forces and moments are obtained in the same manner.
3. The relations between the various quantities at a point of the element remain theoreti-
cally exact, which allows checks for conditions that must also be satisfied exactly.
4. Any type of distributed loads can be modeled theoretically exact, independent of the
discretization.
5. In contrast to a merely numerical approach, all the elements of the characteristic matrices
remain valid, even if they are isolated from their context. Therefore, they may be used
any time for checks.
6. Exact solutions provide the best imaginable basis for estimating the accuracy and con-
vergence of approximate solutions.
7. Analytical solutions usually allow a direct examination of functional dependencies and
therefore the behavior of structural members and structures.

Disadvantages of an exact solution may be

1. Members subjected to tension and compression require their own respective set of stiff-
ness matrices, which must be provided by the computer program.
2. Exact solutions may contain complex analytical functions that are not available as stan-
dard mathematical functions in common high-level programming languages.
3. Rigorous solutions sometimes do not allow the numerical transition e.g. from a beam
subjected to compressive forces to a beam without axial forces. In this case, analytically
undefined expressions may occur with unforeseeable numerical consequences (computer
programs that assemble exact matrices must avoid these cases by performing suitable
checks).
4. Analytical solutions are eventually also evaluated numerically, and are therefore sub-
jected to errors inherent in the numerical algorithms.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 45
 5. For practical applications, it is usually necessary to determine the quantities at selected
points of a structure.

p1

p2

2.3.3 The Buckling Coefficient ε as a Measure of Stability of a Member

q
The buckling coeffieient ε = l |N EI
|
(eq. (2.38) on page 27) characterizes the stability of a
structural member (e.g. a column), if it is modeled with a single element according to the
stiffness method. If geometrical stiffness matrices of the finite element method are employed,
it is usually necessary for accuracy and convergence to choose a finer discretization scheme.
For this, a beam is modeled with several finite elements. Thus, the buckling coefficient cor-
responding to an individual finite element does not characterize the degree of stability of the
pertaining structural member. However, it gives us an estimate of the accuracy of the cor-
responding geometrical stiffness matrix of the individual finite element: For ε < 1, it can be
assumed that the destabilizing forces of a second order theory are sufficiently accounted for.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 46
2.3.4 The Principle of Superposition in the Second Order Theory

The principle of superposition holds for the second order theory only if the resulting normal
force is already known for the computation of the individual load cases.

EI w′′′′ + |N | wII = p1 + p2 + . . .

However, this approach is not applicable in practice, because the resulting normal force is
usually not known.
p
P
The principle of superposition is

V a) valid
P

p
P1

b) not valid
V
P2

p
W

c) not valid

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 47
 P P P P
P 2 2 2 2

d) load rearrangement
̸= + not possible

2.3.5 Redistribution of Internal Forces and Moments in Beam Structures

In a first order analysis, all internal forces and moments increase linearly if a given combination
of loads is increased proportionally. In a second order analysis, the stiffness of a beam element
is nonlinearly dependent on the normal force in the element. The stiffness of elements or
structural members subjected to compression decreases whereas the stiffness of elements or
structural members subjected to tension increases. This causes, in general, a redistribution
of internal forces and moments in the entire structure. If selected internal forces are plotted
over the load parameter λ, we can observe either a proportional or an inversely proportional
increase. In some cases, even a sign change in the service load domain is possible.
Example: Plane frame, bearing a horizontal load, changing direction of action

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 48
 W
2EI,A,l E = 2,1 · 108 kN
m2
l = 5m
I = 18260 cm4 Wkrit = 8808,26 kN
√
EI,A,l EI,A, 2 l A = 120 cm2 −Wkrit = 8732,99 kN

19011,1489 −8031,3371
−5460,6574 6131,5397
W = ±6600 kN ≈ 75 % Wkrit

M II

W
moment-curves for are scaled with −1

displacements 2nd Order Theory

2.3.6 Overcritical Loading of Subsystems

For poorly designed structures, it may occur that individual parts of the structure reach or
exceed their critical loads upon increasing the load combination, although the entire struc-
ture is still far from reaching its lowest critical load. These parts of the structure carry the
corresponding normal forces but they also apply their destabilizing forces on the remaining
structure. With respect to the numerical computation, this may require a higher number of

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 49
iteration steps, which in turn can cause a corresponding redistribution of internal forces and
moments (including a possible sign change of the fixed-end forces).

λ·P
1
1 2

2 2m

3 E = 2,1 · 108 kNm2

3 2m Beam:

I = 0,0001627 m4 ; A = 0,00727 m2
4
Column:
4 2m
I = 0,0000264 m4 ; A = 0,00478 m2

5 Load:

P = 100 kN
5 2m

4m

An analysis with exact stiffness matrices yields:

λ NCrossbar TCrossbar MCrossbar,left MCrossbar,right NPoint 6 TPoint 6

10.000 0.2219 −11.9477 49.5601 1.7750 −988.0523 0.2219

14.000 2.6172 −9.8983 60.5114 −20.9295 −1390.1017 2.6174

17.000 322.0957 932.1431 −1153.2460 −2575.3521 −2632.1431 322.0957

20.000 −6.6425 −44.8766 126.3782 53.1057 −1955.1234 −6.6425

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 50
 λ = 20

λ = 14
Displacements (oversized)
IInd Order Theory Buckling Shape λkrit = 17,3

λ = 10

λ = 10 λ = 20 λ = 20

λ = 14 λ = 14

Moment Curves
IInd Order Theory

The computation (Beam 5 and Column 10 finite elements, iteration of normal forces) results
in:

λ NCrossbar TCrossbar MCrossbar,left MCrossbar,right NPoint 6 TPoint 6 uColumn

10.000 0.1944 −11.5366 45.6236 0.5241 −988.4634 0.1944 −7.0224 · 10−3

14.000 2.1142 −11.6273 57.9987 −11.4689 −1388.3727 2.1142 −2.0768 · 10−2

17.000 126.7237 309.1161 −349.5890 −886.7527 −2009.1161 126.7237 −8.4308 · 10−1

20.000 −6.2948 −41.4839 115.1485 45.6598 −1958.5161 −6.2948 3.4548 · 10−2

2.3.7 Considering Elastic Supports at the Boundaries

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 51
System and Load

EA → ∞
P EI = 900 kNm2
kφ = 5000 kNm
P = 120 kN
2 M = 80 kNm
4,0

M
kφ kφ
1 3
4,0

4
=
ˆ
kφ

4,0 4,0

Moments Ist Order Theory Moments IInd Order Theory

z

z
x

−17,48
−19,11

−20,88 −22,51
−9,22 −9,94
x x x x
19,11

17,48

9,19 9,91
z

z z z z
x

20,90 22,54

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 52
Moments IInd Order Theory

z
kφ = 1500 kNm

−20,80
−19,19
−6,63
x x
20,80

6,61
z

z z
19,21
x

2.4 Stability of Beams and Frames Subjected to Plane

Loading

2.4.1 Determination of the Critical Load of Frames with Exact Stiffness

Matrices (Second Order Theory)

λ · PV λ · pV

λ · PH

The entire load on a plane frame consisting of distributed loads and point loads is applied
gradually with a load paramater λ starting from λ = 0 and λ = 1.
The fixed-end forces of a beam elementq are nonlinearly dependent on the load parameter λ
|λ N (ε)|
and the buckling coefficient ε(λ) = l EI
, which must be determined iteratively:

p = k(ε(λ)) v+p0 (ε(λ)) (2.100)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 53
Hence, the global system of equations for the geometrical unknowns is given as:
0
K(εi (λ)) V = λ P∗ −P (εi (λ)) = P (2.101)

A solution of this nonlinear system of equations can only be obtained iteratively. If K and
0
P are expressed as functions of the buckling coefficient, which is obtained from the normal
forces of the previous iteration step, then we can find a unique solution if

det K ̸= 0 (2.102)

If the determinant of K vanishes, then there is no unique solution to this equation system.
This leads to a method for the determination of the critical load parameter λi . The (generally
infinite number of) roots of the function det(K) of λi yield corresponding load magnitudes, at
which equilibrium is satisfied for a corresponding displacement shape.

det(K)

λkrit,i
λkrit,i+1 λ

A displacement vector Vi (λi ) is assigned to each eigenvalue λi , whose components are deter-
mined from

K(λi ) · Vi = 0, (2.103)

if one component of Vi is given. With λi , only the shape of the displacements, but not the
actual magnitudes of the displacements are determined.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 54
 Remark: The ite- λ(0) = 1
ration step is indi-
cated by a super- p
i(0)
λ(0) = 1
script in parenthe- (0)
P pi(0)
ses.
(0)
ki(0) P
i(0)
Eigenvalue problem
K(0) kb of the iteration step (n)
(0)
V(0) Kb
vi(0) V(0)
pi(0) vi(0)
Without taking
N i(0) redistributions pi(0)
.. of internal forces
. into account N i(0)
(n)
.. ⇒ Vλi , λi
λ(n)
.
q
|λ(n) N i |
εi(n) = li EI
λ(n)
ki(n) ki(n)
g
N-Iteration
K(n) K(n)
g

det K(n) K(n) = Kb + K(n)

g
(0)
With taking pi0(n) λ(n) P
redistributions (n)
of internal P V(n)
forces into
account
V(n) vi(n)
vi(n) pi(n)
pi(n) N i(n)
N i(n)

2.4.2 Linearized Buckling Analysis with Geometrical Stiffness Matrices

If we express the stiffness matrix of a beam, which contains transcendental functions, as a

function of the buckling coefficient ε by a series expansion and truncate after the linear term,
then we obtain the basic relation of the stiffness method for an element as:

p = (kb + kg (λ)) v+p0 (2.104)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 55
Therefore, the geometrical stiffness matrix kg is linearly dependent on the normal force or the
buckling coefficient, which are assumed to be constant for each element. The fixed-end forces
and moments are equal to those of the first order theory.
Similarly, we can express the homogeneous global system of equations in the following form:

 
Kb + K̃g (λ) · V = 0 (2.105)

Assuming that the relative distribution of the normal forces in the entire structure remains
constant if the load combination is increased proportionally over the load parameter λ, we can
extract the parameter λ from K̃g :

(λ) (λ=1)
K̃g = λ · Kg (2.106)
 
Kb +λ · Kg · V = 0 (2.107)

Setting

Kg = −Kg (2.108)

we get:

(Kb −λ · Kg ) · V = 0 (2.109)

The eigenvalue problem given with this equation possesses the following properties:
1. both Kb and Kg are symmetrical.
2. If Kb and Kg are banded, then they possess the same band structure.
3. Kb is always positive definite, i.e., VTi Kb Vi > 0 for Vi ̸= 0
4. Kg is generally positive semi-definite, i.e., VTi Kg Vi ≥ 0 for Vi ̸= 0

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 56
2.4.3 Geometrical Nonlinear Analysis and Eigenvalue Problem: Example

2.4.3.1 Procedure for a Problem with two Degrees Of Freedom

Reference Configuration Deformed Configuration

λ · P2 λ · P3

3
w3 a(1 − cos φ1 )
a[1 − cos(φ1 + φ2 )]
a φ1
φ2
kφ2 2
w2 a(1 − cos φ1 )
φ1
a

kφ1 1

Kinematics: w2 = a(1 − cos φ1 ) (2.110)

w3 = a(1 − cos φ1 ) + a[1 − cos(φ1 + φ2 )] (2.111)

Constitutive Equation: M1 = kφ1 φ1 (2.112)

M2 = kφ2 φ2 (2.113)

EI ⇒ ∞ EA ⇒ ∞

Equillibrium (Virt. Work): − δ(Ai +Aa ) =

M1 δφ1 + M2 δφ2 − λ P 2 δw2 − λ P 3 δw3 = 0 (2.114)

∂w2 ∂w2
δw2 = δφ1 + δφ2
∂φ1 ∂φ2

= a sin φ1 δφ1 (2.115)

δw3 = a [(sin φ1 + sin(φ1 + φ2 ) · 1)δφ1 + sin(φ1 + φ2 ) · 1 · δφ2 ]

(2.116)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 57
 Thus: kφ1 φ1 δφ1 + kφ2 φ2 δφ2 − λ P 2 a sin φ1 δφ1

− λ P 3 a [(sin φ1 + sin(φ1 + φ2 )) δφ1 + sin(φ1 + φ2 ) δφ2 ] = 0 (2.117)

h i
kφ1 φ1 − λ P 2 a sin φ1 − λ P 3 a (sin φ1 + sin(φ1 + φ2 )) δφ1 +
h i
kφ2 φ2 − λ P 3 a sin(φ1 + φ2 ) δφ2 = 0 (2.118)

h h ii
kφ1 φ1 − λ a (P 2 + P 3 ) sin φ1 + P 3 sin(φ1 + φ2 ) δφ1 +
h i
kφ2 φ2 − λ P 3 a sin(φ1 + φ2 ) δφ2 = 0 (2.119)

h i
g1 (φ1 , φ2 , λ) = kφ1 φ1 − λ a (P 2 + P 3 ) sin φ1 + P 3 sin(φ1 + φ2 ) = 0 (2.120)

g2 (φ1 , φ2 , λ) = kφ2 φ2 − λ P 3 a sin(φ1 + φ2 ) = 0 (2.121)

Eq. (2.119) can be written in matrix-notation:

A(φ1 , φ2 , λ) = [0] (2.122)

A linearisation regarding 2nd order theory where sin φ ≈ φ and cos φ ≈ 1 yields:
h i
kφ1 φ1 − λ a (P 2 + P 3 ) φ1 + P 3 (φ1 + φ2 ) = 0 (2.123)

kφ2 φ2 − λ P 3 a (φ1 + φ2 ) = 0 (2.124)

resp.

K(λ) V = [0] (2.125)

Remark.:
Both functions h i
g1 (φ1 , φ2 , λ) = kφ1 φ1 − λ a (P 2 + P 3 ) sin φ1 + P 3 sin(φ1 + φ2 ) = 0
and
g2 (φ1 , φ2 , λ) = kφ2 φ2 − λ P 3 a sin(φ1 + φ2 ) = 0
represents for arbitrarily defined λ a surface-curve in the φ1 − φ2 − plane spanned by the
variables. The intersections with the horizontal zero-plane mark the states of equilibrium for

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 58
one degree of freedom:

Circular cut part 2-3: M2 = kφ2 φ2 = λ P 3 a sin(φ1 + φ2 )

part 1-2: M1 = kφ1 φ1 = M2 + λ (P 2 + P 3 ) a sin φ1

The points of intersection of both the contour lines mark the solution for the problem. The
solution of the linearized equation system is obtained analogously to eq. (2.125).

2.4.3.2 System Matrices

a = 1,5 m
P 2 = 4 kN kφ1 = 20 kNm
P 3 = 2 kN kφ2 = 10 kNm

λ · P2 λ · P3
K = Kb − λ · Kg 3

a φ1
20 0 12 3
   
φ2
Kb =  ; Kg =  ;
0 10 3 3 kφ2 2

φ1
a
20 − λ · 12 −λ · 3
 

K = 
kφ1 1
−λ · 3 10 − λ · 3

 
φ1
Vλ =   K(λ) V = 0
φ2

Roots of the determinant of the coefficient-matrix:

(20 − 12λ) · (10 − 3λ) − 9λ2 = 0

200 − 60λ − 120λ + 36λ2 − 9λ2 = 0

27λ2 − 180λ + 200 = 0

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 59
 1  √ 
λ1,2 = +180 ± 1802 − 4 · 27 · 200
2 · 27
1
= (180 ± 103,923) = 3, 3 ± 1,9245
54

2.4.3.3 Eigenvalues, Eigenvectors

Eigenvalues: λ1 = 1,4088 λ2 = 5,2578

Eigenvectors:

1. λ1 = 1,4088

φ1 = 1

−(20 − 12λ1 )
φ2 =
−3λ1

1 1.Eigenform
 

= +0,7322 V1 =  
0,7322

2. λ2 = 5,2578

φ1 = 1

−(20 − 12λ2 )
φ2 =
−3λ2

1
 
2.Eigenform
= −2,732 V2 =  
−2,732

2.4.3.4 Figurative Presentation of the Solutions of Systems (2.119) and (2.125)

a = 1,5 m, P2 = 4 kN, P3 = 2 kN, kφ1 = 20kNm, kφ2 = 10kNm

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 60
Subcritical load λ = 1,0

(a) Geometric non-linear computation - Points of inter-

section of the zero contour lines of equations (2.120)
and (2.121)

(b) 2nd order analysis (2.123) and (2.124)

Figure 2.3: Solution for a subcritical load (λ = 1,0)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 61
Critical load λ = 1,4088 – 1st eigenvalue

(a) Geometric non-linear computation - Points of inter-

section of the zero contour lines of equations (2.120)
and (2.121)

(b) 2nd order analysis (2.123) and (2.124)

Figure 2.4: Solution for the critical load (λ = 1,4088)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 62
Supercritical load λ = 1,7 (λ1 < λ < λ2 )

(a) Geometric non-linear computation - Points of inter-

section of the zero contour lines of equations (2.120)
and (2.121)

(b) 2nd order analysis (2.123) and (2.124)

Figure 2.5: Solution for a supercritical load (λ = 1,7)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 63
2.4.4 Effective Length of Beam Columns of Plane Frames

2.4.4.1 To Commemorate: Critical Loads (Euler), Effective Length

π 2 EI P
I
Pcrit = simply supported
l2
π 2 EI P
II
Pcrit =  2 clamped
l
2

π 2 EI left: clamped P
III
Pcrit =  2 right: simply supported
√l
2

π 2 EI left: clamped P
IV
Pcrit =
(2 · l)2 right: free end

Using the concept of effective length

sk = β · l

we can consistently express the critical load independent of the boundary conditions:

π 2 EI
Pcrit =
s2k

π 2 EI
Pcrit =
β 2 l2

In the case of a linear constitutive law, we find the stresses under the critical load to be:

π 2 EI
σcrit =
A β 2 l2

Introducing the radius of gyration

s
I
i=
A

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 64
we get:

i2
σcrit = π 2 E
β 2 l2
βl
λ= (slenderness ratio)
i
π2 E
σcrit =
λ2

compare DIN EN 1993-1-1:

s
A fy λ
λ̄ = = qE
Ncr π fy

2.4.4.2 Effective Length of Beam Columns and their Eigenmodes

Governing equation of every bar in compression under critical load:

EI w(x)′′′′ + |λ1 · L| w(x)′′ = 0 (2.126)

The deflection line in the first eigenmode must be a solution of this homogenous equation:
s
|λ1 · L|
w(x) = C1 + Ĉ2 x + C3 cos ε̂1 x + C4 sin ε̂1 x ε̂1 = (2.127)
EI
x x x
resp: w(x) = C1 + Ĉ2 l + C3 cos ε̂1 l + C4 sin ε̂1 l
l l l
s
x x x |λ1 · L|
= C1 + C2 + C3 cos ε1 + C4 sin ε1 ε1 = l · (2.128)
l l l EI

ε21
|λ1 · L| = EI C2 = Ĉ2 · l (2.129)
l2

1 ε1 x ε1 x
w(x)′ = +C2 − C3 sin ε1 + C4 cos ε1
l l l l l

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 65
 ε21 x ε21 x
w(x)′′ = −C3 2
cos ε 1 − C 4 2 sin ε1
l l l l
ε31 x ε31 x
w(x)′′′ = +C3 3
sin ε 1 − C 4 3
cos ε1
l l l l

C2 can be mechanically interpreted:

T (x) = −EI w′′′ + λ1 · L w′ (2.130)

= −EI w′′′ − |λ1 · L| w′ (2.131)

" #
ε3 x ε3 x
= EI −C3 31 sin ε1 + C4 31 cos ε1
l l l l

1 ε21 ε1 x ε1 x
 
+ λ1 L C2 − 2 EI −C3 sin ε1 + C4 cos ε1
l l l l l l
λ1 · L
T (x) = + C2 (2.132)
l

Using:

T
C2 = + l (2.133)
λ1 · L

we can rewrite the deflection line:

T x x
w(x) = C1 + x + C3 cos ε1 + C4 sin ε1 (2.134)
λ1 · L l l

The coordinates of the inflection points may be calculated in the local coordinate system of
the bar from the following condition:

ε2 x ε2 x !
w(x)′′ = −C3 2
cos ε 1 − C 4 2
sin ε1 = 0
l l l l

Hence:
ε1 xwi
C3 = −C4 tan (2.135)
l

Here, the quantities xwi are the desired local coordinates of the inflection points.
Introducing a new coordinate system, whose origin is in the neighboring inflection point (left

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 66
hand side), leads to the transformation (respectively translation) relations:

x = x − xwi (2.136)

x = x + xwi (2.137)

xwi ≤ 0

xw2 x
βR · lR βR · lR
lR

xw1 x

T
wc′
λ1 · L λ1 · L
T

Rem.: From eq. (2.132) we get λ1T·L = Cl2 = wc′ , being the inclination of the chord of inflection
points of the deformed bar (see chapter 2.2.5, destabilizing forces).
Using eqs. (2.137) and (2.135) we get from eq. (2.134):

T x + xwi x + xwi
w(x) = C1 + (x + xwi ) + C3 cos ε1 + C4 sin ε1
λ1 · L l l
T xw x + xwi x + xwi
= C1 + (x + xwi ) − C4 tan ε1 i cos ε1 + C4 sin ε1
λ1 · L l l l
"
T xw x xw x xw
 
= C1 + (x + xwi ) + C4 − tan ε1 i cos ε1 cos ε1 i − sin ε1 sin ε1 1
λ1 · L l l l l l
#
x wwi x xw
+ sin ε1 cos ε1 + cos ε1 sin ε1 i
l l l l

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 67
 "
T x xw x xw xw
= C1 + (x + xwi ) + C4 − cos ε1 sin ε1 i + sin ε1 tan ε1 i sin ε1 i +
λ1 · L l l l l l
#
x xw x wwi
+ sin ε1 cos ε1 i + cos ε1 sin ε1
l l l l

1
" #
T x xwi xwi x xwi
= C1 + (x + xwi ) + C4 
xw
 sin ε1 sin ε 1 sin ε 1 + sin ε 1 cos2
ε 1
λ1 · L cos ε1 i l l l l l
l

T 1 x
w(x) = C1 + (x + xwi ) + C4 
x
 sin ε1
λ1 · L cos ε1 i
w l
l
T ε1 x
w(x)′ = + + C4 
x
 cos ε1
λ1 · L l cos ε1 i
w l
l
ε2 x
w(x)′′ = −C4 1
xw
 sin ε1
l2 cos ε1 i l
l

1. point of inflection (point of origin of the new coordinate system):

xw1 = 0

2. point of inflection:

ε1 xw2 π·l π
=π xw2 = =β·l Thus: β =
l ε1 ε1

Rem.: The effective length of a (structural) member is identical with the distance of neigh-
boring inflection points of the function of the deflection line of the bar in the first eigenmode.
This definition assumes straight bars of constant cross-section and constant axial compressive
forces.

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 68
2.4.4.3 Developing the Effective Length in Comparing Critical Loads

Having calculated the least λ1 of the overall system, we can compute the respective compressive
forces λ1 · L of every member. The related buckling coefficient is:
s
|λ1 · L|
ε1 = l ·
EI

If we equate this compressive force with the critical force of an Euler 2 column (having the
same member length and stiffness), we develop:
v
ε2 2
! π EI π π 2 EI
u
|λ1 · L| = 21 EI = 2 2 β= β=
u
t
l β l ε1 |λ1 · L| l2

In these relations, the longitudinal force L is related to λ = 1 (from which we started the
eigenvalue analysis).

2.4.4.4 Example

Pendulum Bar with spring and rotational spring

kw kw · w b
b
Tb

Ma
kφ
a

kφ · φa

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 69
Fixed end forces:

EI C ′ EI C ′ − D′ wb
Tb = φa +
l l l l l
EI ′ EI ′ wb
Ma = C φa + C
l l l

Equilibrium (deformed state):

C ′ − D′ C′
 
+ kw 0
   
 EI 3
EI 2  wb
l l =
C′ C′
  
φ 0
 
EI 2 EI + kφ a
l l

buckling condition (determinant = 0):

C ′ − D′ C′ C ′2
" #" #
EI + kw EI + kφ − EI 2 4 = 0
l3 l l

C ′ (C ′ − D′ ) (C ′ − D′ ) C′ 2 C
′2
EI 2 + EI kφ + EI kw + kw kφ − EI =0
l4 l3 l l4
(C ′ − D′ ) C′
" # " #
l3 l
C (C − D ) + kφ l
′ ′ ′
+ kw l 3 + kw kφ − C ′2 = 0
EI EI EI EI

kφ l kw l3
Parameter: γ= δ=
EI EI

C ′2 − C ′ D′ + γ (C ′ − D′ ) + δ C ′ + δ γ − C ′2 = 0

− C ′ D′ + γ (C ′ − D′ ) + δ C ′ + δ γ = 0

With (compare Section 2.2.3.1)

ε2 sin ε
C = ′
sin ε − ε cos ε
D ′ = ε2

ε3 cos ε
C ′ − D′ =
sin ε − ε cos ε

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 70
one obtains:

ε4 sin ε ε3 cos ε ε2 sin ε

− +γ +δ +δγ = 0
sin ε − ε cos ε sin ε − ε cos ε sin ε − ε cos ε

Buckling equation (compare. Petersen, Statik und Stabilität der Baukonstruktionen, Table
5.4, p. 341)

Stability of Structures 2 Second Order Analysis and Stability of Plane Frames

Winter Term 2022/2023 71
3 Second Order Analysis and Stability of
Space Frames

3.1 Introduction

3.1.1 Failure Modes of Bars

1.Prandtl’s problem

SA
Cantilever in the undeformed state (side view
and cross–section)

λ·F
Applied load (acting in the symmetry plane of
the beam) and bending moment
M

λ·F
Deflection line and applied load (1st order ana-
lysis)
w

F crit = λcrit · F

Spatial displacement mode under the first cri-

tical load

Stability of Structures
Winter Term 2022/2023
72
 λ · px
2.Torsional buckling of a thin–walled beam in compres-
sion
Cross–section and Rotation of
yoke support in cross-sections in
the ends of the midspan
column

Because of the shape of the cross–section and the supports, the

webs of the bar may be regarded as separate plate strips, which
are simply supported in three edges and having one edge free to
move. Because of the assumption of no cross–sectional distorsion,
every one of the plate strips undergoes axissymmetric displace-
ments (with respect to the axis of the bar)

3.Torsional instability of a beam

F F

2h

l b

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 73
 If the end section of the bar rotates by an-
F gle ϑl , we can calculate an external mo-
h sin ϑl ment:

M X = F · h · sin ϑl

h If we assume h ≪ l, for small displace-

ϑl ϑl ments the following relation holds:

h ϑl
M X = GIT
l

In the range of GIl T < F · h, the geometri-

cally nonlinear problem leads to the solu-
tion:

Mx Mx ( GIl T > F · h) Mx ( GIl T = F · h) F · h · sin ϑl − GIT

ϑl
=0
l

GIT
In the special case of = F · h for the
Mx ( GIl T < F · h) l
linearised problem of (sin ϑl ≈ ϑl ), we can
Mx find a neighbouring state of deformation
in equilibrium (in the sense of a stability
M x = F h sin ϑl
problem, which essentially is an eigenvalue
problem):

ϑn.l. ϑ 
GIT

F · h− ϑl = 0
l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 74
4. Possible modes of failure of a beam column in fork supports, having different
cross–sections and bearing an eccentrically acting longitudinal force

P P P
P yP P zP
Yoke support

P yP Yoke support
P zP
P P P

S=M S=M S=M

y y y

z z z
Stress problem Buckling, Lateral–torsional
Torsional buckling buckling

S=M S=M S=M

y y y

z z z
Lateral–torsional Buckling, Lateral–torsional
buckling Torsional buckling buckling

S S S
y y y
M M M

z z z
Stress problem Lateral–torsional Lateral–torsional
buckling buckling

S: Centre of gravity M: Centre of shear Point of application of loads

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 75
5. Possible modes of failure of a girder in fork supports, having different cross–
sections and with a constant applied moment

My My
x

y y

z
z
Stress problem Lateral buckling

Lateral buckling (about an appro-

ximatevly fixed axis of rotation)

Stress problem Lateral buckling

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 76
 3.1 Introduction
3.2.1 Kinematics for biaxial bending and torsion 3.2.2 Material law

3.4.3 Supplements to torsion

3.4.4 Transformation of stress resultants of the deformed configura-

tion to the coordinate system of the reference state

3.4.5 P. V. D (preliminary form)

3.4.5 Simplifying assumptions concerning p. v. v.

3.4.5 P. V. D (final formulation)

3.4.6 Eulerian equations: governing equa-

tions of a bar, being three-dimensionally
loaded; boundary conditions, stress resultants

3.4.8 Special case of an exact solution (2nd order analysis)

3.5 Exact solutions of the eigenvalue problems of bars

3.6 Weak formulation

of the principle of virtuel displacements
Finite element for bars
Critical loads of members in codes

3.6.6 Analysis of space frames,

2nd order analysis

3.6.6 Numerical solution of the eigenvalue problem of space frames

2nd order analysis of space frames in codes

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 77
3.2 General and Kinematic Assumptions, Constitutive Law

3.2.1 Assumptions

1. The beam elements under investigation here are assumed to have a straight and prismatic
shape, i.e., the cross-section is constant along the central axis.
2. The individual members (beam elements) are made of a homogeneous, isotropic material.
3. In the reference configuration, we shall assure that the beams are unstressed and possess
their initial geometry. A local Cartesian coordinate frame is assigned to each individual
member, where the x-axis and the central axis coincide.
Applying external loads causes deformations and displacements of the structure, and the
external loads are in equilibrium with the internal forces and moments.
4. The equilibrium equations are derived using the principle of virtual displacements (the
Euler equations of the integral principle) according to a first order analysis, i.e., without
taking the deformations of the beam into account.
5. The shape of the cross sections in the y-z-plane does not change besides a rotation.
Axial strains, shearing strains, and stresses in the cross-sectional plane are negligible,
particularly for warping torsion of thin-walled bars.
6. We will adopt the Bernoulli hypothesis for the shearing deformations in bending, where
the effect of shearing strains on the state of stress is neglected, which implies that the
sections remain plane and orthogonal to the central axis.
7. Depending on the geometry of a section, the boundary conditions and the applied load,
we can employ the following mechanical models for the computation of the effects of
torsion:
a) Bars with a solid cross section or thick-walled sections with or without holes are
amenable to the general uniform torsion theory, provided that warping is not re-
strained.
b) Sufficiently accurate results for thin-walled tubular or multicellular sections can be
obtained with Bredt’s theory.
c) Thin-walled open sections with negligible warping restraints can be computed using
the specialized uniform torsion theory for thin-walled sections.
d) Thin -walled open-closed cross sections (with mainly open segments) are in a state
of stress that is a superposition of those described above.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 78
 e) If the effects of restrained warping on thin-walled open sections are not negligible,
it is necessary to apply the classical theory of non-uniform torsion. In complete
analogy to bending, we neglect in this context the influence of shearing strains on the
distribution of axial displacements respectively warping normal stresses (hypothesis
of Wagner).
f) If the effects of restrained warping on thin-walled closed sections are not negligible,
we also need to take warping shearing strains into account (not dealt with in this
context).

3.2.2 Kinematic Relations

The assumptions of Chapter 1: Introduction (particularly the plane-seetions hypothesis (5),

the assumption on the undistorted profile (6), and a product expression for the warping (7.5))
enable us to express the displacements u(x, y, z), v(x, y, z) and w(x, y, z) of any point P (x, y, z)
in the interior of a beam in terms of the displacements u, v, w and ϑ of an arbitrary axis (which
is axis of rotation at the same time) (kinematics):

u(x, y, z) = u · 1 − v ′ · y − w′ · z − ϑ′ · ω0 (y, z) (3.1)

v(x, y, z) = v − ϑ · z (3.2)

w(x, y, z) = w + ϑ · y (3.3)

y
P (y,z)
ϑ
ϑ·y
ϑ·z

Figure 3.1: Displacement of a material point P because of a cross-sectional rotation

Please note that in the above expressions, the functions u(x, y, z), v(x, y, z) and w(x, y, z)
refer to any point in the interior or on the surface of the bar under investigation, whereas the
displacements u, v, w and ϑ (without explicitly mentioning their dependence on x) are related
to the axis of the bar.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 79
The warping function ω0 (y, z) in this context includes both the warping of thick-walled or
solid sections, as well as the (strain - free ) warping of the cross -sectional segment center lines
according to the theory of thin-walled open sections.
With the definitions

φy = −w′ (3.4)
φz = −v ′ (3.5)
ψ = −ϑ′ (3.6)

we get:

u(x, y, z) = u + φz y + φy z + ψω0 (3.7)

The strains in the interior of the beam may in general be expressed as a function of the
displacement fields as:

ε(x, y, z) = u(x, y, z)x (3.8)

γxy (x, y, z) = v(x, y, z)x + u(x, y, z)y (3.9)

γxz (x, y, z) = w(x, y, z)x + u(x, y, z)z (3.10)

Substituting expressions (3.7), (3.2), and (3.3) yields:

ε(x, y, z) = u′ + φ′z y + φ′y z + ψ ′ ω0 (3.11)

γxy (x, y, z) = v ′ − ϑ′ z + φz + ψ ω0,y (3.12)

γxz (x, y, z) = w′ + ϑ′ y + φy + ψ ω0,z (3.13)

The first and the third term on the right side of (3.12) correspond to combined bending and
shearing in the x-y-plane. They vanish according to assumption 6.
Because of the hypothesis of Wagner, the remaining terms are related to uniform torsion
(which are not produced by changing axial warping stresses with respect to x). To emphasize
this fact, we introduce an index V (de St. Venant) with ω and get:

γxy,V (x, y, z) ≈ ψ · (ωV,y + z) (3.14)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 80
 y φz resp. uy x w′
v u

v′ φy resp. uz

u x w z

Figure 3.2: Shearing in a x-y -, respectively x-z -plane

Similarly, the first and the third term on the right side of (3.13) correspond to combined
bending and shearing in the x-z-plane. They vanish according to assumption 6.
Again because of the hypothesis of Wagner, we may associate the remaining terms to uniform
torsion:

γxz,V (x, y, z) ≈ ψ(ωV,z − y) (3.15)

The strains of the central axis become:

κz ≈ φ′z (3.16)

κy ≈ φ′y (3.17)

κω ≈ ψ ′ (3.18)

Here, we can rewrite the strains of the central axis in terms of the derivatives of the displace-
ments as:

ε = u′ (3.19)

κz = −v ′′ (3.20)

κy = −w′′ (3.21)

κω = −ϑ′′ (3.22)

Substituting (3.19) and (3.16) in (3.11) yields:

ε(x, y, z) = ε + κz y + κy z + κω ω0 , (3.23)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 81
and substituting (3.19), (3.20), and (3.21) yields:

ε(x, y, z) = u′ − v ′′ y − w′′ z − ϑ′′ ω0 (3.24)

Therefore, we get for the variation of the strains

δε = δu′ (3.25)

δκz = −δv ′′ (3.26)

δκy = −δw′′ (3.27)

δκω = δϑ′′ (3.28)

respectively

δε(x, y, z) = δε + δκz y + δκy z + δκω ω0 (3.29)

δγxy,V (x, y, z) = −δϑ′ (ωV,y + z) (3.30)

δγxz,V (x, y, z) = −δϑ′ (ωV,z − y) (3.31)

With a given warping function ωV we only need to consider variations of functions of x.

In order to simplify further developments, we will rewrite eqs. (3.1) and (3.24) formally as:

u(x, y, z) = (u)1 + (−v ′ )y + (−w′ )z + (−ϑ′ )ω0 (y, z) (3.32)

ε(x, y, z) = (u′ )1 + (−v ′′ )y + (−w′′ )z + (−ϑ′′ )ω0 (y, z) (3.33)

The terms in brackets on the right sides are strains of the central axis, and therefore a function
of x and dependent on the loading. The strains of the second equation may be combined to a
vector κ(x) as:

κT (x) = [u′ − v ′′ − w′′ − ϑ′′ ] (3.34)

resp. κT (x) = [ε κz κy κω ]

The factors without brackets in the product expression on the right side are merely a function
of the coordinates of the section and are independent of the loading. They may be interpreted
as the unit warping corresponding to stretching, rotation about the y- and z-axis, and torsion.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 82
 ω T (y, z) = [1 y z ω0 (y, z)] (3.35)

Using the above conventions, we obtain the strain at an arbitrary point of the beam:

ε(x, y, z) = ω T (y, z) κ(x) (3.36)

3.2.3 Consitutive Laws for Stresses

Because of the adopted hypothesis (5.) (shape of the cross section), (6.) (Bernoulli) and (7.a)
(Wagner), we can only afford to introduce a constitutive law for normal stresses (produced by
axial stretching, biaxial bending and constrained warping effects) and for shearing stresses,
which does not depend on the rate of change of normal stresses.
In any case we adopt linear elastic constitutive laws.
We assume a one-dimensional Hooke’s law for the longitudinal stresses resulting from normal
strains, bending, and warping torsion without taking the lateral strains into account (fiber
model):

σ(x, y, z) = E ε(x, y, z) (3.37)

With (3.36) in matrix notation:

σ(x, y, z) = E ω T κ (3.38)

The shearing stresses from the uniform torsion theory are computed according to the general-
ized Hooke’s law of three-dimensional elasticity theory as:

τxy,V (x, y, z) = Gγxy,V (x, y, z) (3.39)

τxz,V (x, y, z) = Gγxz,V (x, y, z) (3.40)

Here, again we attach an index V to both shearing stresses and shearing strains.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 83
3.3 Bending and Torsion in a 1st Order Analysis

3.3.1 Principle of Virtual Displacements: Initial Formulation

In General:

− δW = −δ (Wi + Wa ) = 0 (3.41)
Z Z
− δWi = [σx δεx + σy δεy + σz δεz + τxy δγxy + τyz δγyz + τzx δγzx ] dA dx (3.42)
l A

3.3.2 Virtual Work of Internal Forces, Transformation in Principal Axes

Based on the assumption that the shape of the cross sections in the y-z-plane does not change
as well on the hypotheses of Bernoulli and Wagner (see lecture notes) equation (3.42) can be
written as:
Z Z
− δWi = [σx δεx + τxy,V δγxy,V + τzx,V δγzx,V ] dA dx (3.43)
l A

Using the constitutive laws for stresses

Z Z
− δWi = [Eε δε + G γxy,V δγxy,V + G γzx,V δγzx,V ] dA dx (3.44)
l A

and the kinematic relations:

Z Z 
− δWi = E (ε + κz y + κy z + κω ωw ) δ (ε + κz y + κy z + κω ωw ) +
l A

+ Gϑ′ (ωV,y + z) δϑ′ (ωV,y + z) + Gϑ′ (ωV,z − y) δϑ′ (ωV,z − y) dA dx (3.45)

The cross-sectional integrations can be applied in advance:

Z   Z Z Z Z 
− δWi = δε E ε dA + κz y dA + κy z dA + κω ωw dA +
l
 Z Z Z Z 
+ δκz E ε y dA + κz y dA + κy
2
yz dA + κω yωw dA +
 Z Z Z Z 
+ δκy E ε z dA + κz yz dA + κy z dA + κω
2
zωw dA +

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 84
  Z Z Z Z 
+ δκω E ε ωw dA + κz yωw dA + κy zωw dA + κω ωw2 dA +
Z h i 
+ δϑ′ G ϑ′ (ωV,y + z)2 + (ωV,z − y)2 dA dx (3.46)

We obtain for the cross-sectional values related to an arbitrary axis:

Z
dA =A (3.47)
Z Z Z
y dA = Sz z dA = Sy ωw dA = Sω (3.48)
Z Z Z
y z dA = Iyz y ωw dA = Iyω z ωw dA = Iyω (3.49)
Z Z Z
y 2 dA = Iz z 2 dA = Iy ωw2 dA = Iω (3.50)
Z h i
(ωV,y + z)2 + (ωV,z − y)2 dA = IT (3.51)

Using (3.47) and (3.51) we obtain for (3.46):

Z
− δWi = {δε E [ε A + κz Sz + κy Sy + κω Sω ] +
l

+ δκz E [ε Sz + κz Iz + κy Iyz + κω Iyω ] +

+ δκy E [ε Sy + κz Iyz + κy Iy + κω Izω ] +
+ δκω E [ε Sω + κz Iyω + κy Izω + κω Iω ] +
+ δϑ′ G ϑ′ IT dx} (3.52)

The cross-sectional values (3.48) to (3.49) vanish, if in the system of principal axes the axis of
the center of shear is considered separately. Loads and internal forces have to be defined as
depicted below:
Reference Frame for thin-walled cross sections:

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 85
 N (y N ,z N )

x
y

S (y S ,z S )
y x
α

y M (y M ,z M )
α
D (y D ,z D )

z
z z

• Arbitrary reference frame: x, y, z

Axis of rotation: D
Start for integration for ω: N (y N , z N )
• System of center of gravity: x, y, z (Sy , Sz )
Axis of rotation: D
Start for integration for ω: N (y N , z N ) (Sω = 0)
• Principal axes: x, y, z (Iyz = 0)
Axis of rotation: Center of Shear M
Start for integration for ω: N (yN , zN ) (Iyω = Izω = 0)

In the system of principal axes where the center of shear is the axis of rotation (3.52) is
simplified:
Z
−δWi = {δε EAε + δκz EIz κz + δκy EIy κy + δκω EIω κω + δϑ′ GIT ϑ′ } dx (3.53)
l

With

N = EA ε MxV = GIT ϑ′ My = EIy κy Mz = EIz κz Mω = EIω κω (3.54)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 86
 N = EA u′ MxV = GIT ϑ′ My = −EIy w′′ Mz = −EIz v ′′ Mω = −EIω ϑ′′ (3.55)

one obtains:
Z
−δWi = {N δu′ − My δw′′ − Mz δv ′′ − Mω δϑ′′ + MxV δϑ′ } dx (3.56)
l

3.3.2.1 Virtual Work of the Loads

As a start the virtual work of the external loads reads:

Z h i
δWe = px δup + py δvp + pz δwp + mx δϑ + my δφy + mz δφz dx+
l
h ib
+ N δu + Qy δv + Qz δw + M x δϑ + M y δφy + M z δφz + M ω δψ (3.57)
a

yp
ypM yM
y S
zM
M
Qy φy v ϑ

zp
zpM
ϑ
P
my py px

mx pz

mz z
w
φz

Qz

up = u − yp v ′ − zp w′ − wp ϑ′ , vp = v − ϑ zpM , wp = w + ϑ ypM (3.58)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 87
 Z h
δWe = px (δu − yp δv ′ − zp δw′ − wp δϑ′ ) + py (δv − δϑ zpM ) + pz (δw + δϑ ypM ) +
l

+mx δϑ + my δφy + mz δφz ] dx+

h ib
+ N δu + Qy δv + Qz δw + M x δϑ + M y δφy + M z δφz + M ω δψ (3.59)
a

3.3.2.2 Principle of Virtual Displacements: Resulting Formulation

Z
− δW = [N δu′ + MxV δϑ′ − My δw′′ − Mz δv ′′ − Mω δϑ′′ ] dx
Z h
− px (δu − yp δv ′ − zp δw′ − wp δϑ′ ) + py (δv − δϑ zpM ) + pz (δw + δϑ ypM ) +

+ mx δϑ + my δφy + mz δφz ] dx
h ib
− N δu + Qy δv + Qz δw + M x δϑ + M y δφy + M z δφz + M ω δψ =0 (3.60)
a

Z  h i
− δW = [−N ′ − px ] δu + −MxV
′
+ py zpM − pz ypM − px ωp − mx − Mω′′ δϑ
h i h i 
+ −My′′ − pz − px zp − m′y δw + −Mz′′ − py − p x y p − m′z δv dx
h  ib
+ N − N δu
a
h  ib h  ib
+ MxV + Mω′ + px ωp − M x δϑ + −Mω + M ω δϑ′
a a
h  ib h  ib
+ My′ + my + px zp − Qz δw + −My + M y δw′
a a
h  ib h  ib
+ Mz′ + mz + px yp − Qy δv + −Mz + M z δv ′ =0 (3.61)
a a

Decisive differential equations, boundary conditions (for loads) and internal forces:
1. ′
MxV + Mω′′ + mx − py zpM + pz ypM + p′x ωp = 0

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 88
 mx − py zpM + pz ypM + p′x ωp

Mω′ (x) MxV MxV + MxV

′
dx + . . . Mω′ + Mω′′ dx + . . .
= MxW = MxW + MxW
′
dx + . . .

BC1 : M x = +MxV + Mω′ + px ωp BC2 : M ω = +Mω (3.62)

2. My′′ + pz + m′y + p′x zp = 0

my + px zp

pz
x My + My′ dx + . . .

My (x) Qz (x) Qz + Q′z dx + . . .

z,w

BC1 : Qz = +My′ + my + px zp BC2 : M y = +My (3.63)

V : Q′z + pz = 0 (3.64)
X

M : − Qz + My′ + my + px zp = 0 (3.65)
X

⇔ My′′ + pz + m′y + p′x zp = 0 (3.66)

3.4 Second Oder Analysis of Combined Torsion and Bending

3.4.1 Preliminary Remarks

Based on the assumptions given in Chapter 3.2, in this chapter, we shall develop the theory of a
second order analysis of beams and beam structures subjected to three–dimensional loads. The
derivation of the governing equations is performed for a straight beam by using the principle
of virtual displacements.
Starting with this virtual work principle, we will eventually obtain the governing differential
equations in form of Euler’s equations. The mechanical interpretation of the relevant terms
in these equations is discussed subsequently. The internal forces and moments of the second

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 89
order theory are given in the reference frame of the initial configuration by the boundary terms
of the virtual work principle. Further, we will derive and discuss in Chapter 3.5, the relations
that govern the most important cases of three–dimensional buckling of single–span beams and
girders encountered in engineering practice (buckling or eigenvalue problems, respectively:
torsional buckling, lateral buckling, lateral–torsional buckling).

yM ypM b
yBM x
L(b)
l u(b)
a S Internal forces, displace-
M y (b) ments and loads at a
M z (b) beam element of length l.
px (x) u(a)
ϑ(b)
M y (a) M ω (b)
S φz (b) M x (b)
L(a)
M z (a) φy (b) M T y (b)
y
w(b)
v(b)
φz (a) T z (b) B
zM
φy (a) T y (a)
M
w(a) mz (x)
M x (a) y(a)
M ω (a)
ϑ(a) pz (x)
zBM B
T z (a)
p

zpM

p
mx (x) py (x) my (x)

x, y, z : Principal axes system (initial configuration)

S–S : Centroidal axis
M–M : Shear center axis, coordinates: yM , zM
z p–p : Axis of the applied loads, coordiantes (related to M): ypM , zpM
B–B : Axis of the elastic foundation, coordinates (related to M): yBM , zBM

3.4.2 The Principle of Virtual Displacements: Initial Formulation

The principle of virtual displacements also holds for a second order analysis. However, we
need to impose the virtual displacement on the deformed configuration.

According to assumption in Chapter 3.2 (the loads are defined in the initial configuration and
do not change their direction) as well as requirement 9 (the virtual work integral is formulated

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 90
with respect to the initial configuration), equations (3.60) and (3.61) hold if we transform the
internal forces and moments of the deformed configuration (in the following designated by a
tilde, if necessary) to the reference frame of the initial configuration.

Unlike in Chapter 3.3, here, we will assume that the distributed longitudinal load px acts in
the centroidal axis of the beam and that the contribution of the virtual work px ω (0, 0) δϑ′ to
the integral expression of the external loads may be neglected.

Please note that although the loads are generally defined in the undeformed configuration, they
follow the displacements of the structure from the undeformed configuration to the deformed
configuration without changing their direction (only in this configuration we can have an
equilibrium condition). Because we formulate the following relations based on the reference
frame of the initial configuration, it is also necessary to consider the resulting loads in this
reference frame.

The effect of the displacement of the beam on the bimoment is negligible.

The forces and moments applied at the ends of the beam are positive according to the internal
forces and moments along the beam (i.e., definition 1).

−δ(Wi + Wa ) = 0 (3.67)

Z
−δWi = [ L δu′ + Mx δϑ′ − My δw′′ − Mz δv ′′ − Mω δϑ′′

+ vBM kv δvBM + wBM kw δwBM + ϑ kϑ δϑ ] dx (3.68)

Z 
δWa = px δu + py δ(v − ϑ zpM ) + pz δ(w + ϑ ypM )

+ mx δϑ + my δφy + mz δφz ] dx

+ [ L(x) δu + T y (x) δv + T z (x) δw

b
+ M x (x) δϑ + M y (x) δφy + M z (x) δφz + M ω (x) δϑ ′
(3.69)
a

3.4.3 Supplement: Torsion Caused By Inclination of the Longitudinal

Fibers

Remark: The following chapter considers the deformed configuration.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 91
 ϑ
M
x
rM
dx dϑ
rM ϑ′ σ dA
σ dA
β

A torsional moment causes twisting of cross sections adjacent at a distance dx about an angle
dϑ about the shear center axis. With rM equal to the distance from the shear center to an
individual fiber, we get for the inclination β of the fiber with respect to the longitudinal axis
of the beam:

dϑ
β = rM
dx

Therefore, the differential force σ dA acting in the direction of the inclined fiber possesses a
component parallel to the cross–sectional plane of magnitude

dF = rM ϑ′ σ dA

which causes a differential torque:

dMxσ = rM
2
ϑ′ σ dA

Integration over the cross-section yields:

Z
Mxσ = ϑ′ σ rM
2
dA

It follows:
!
Z
N My Mz MωM
Mxσ = ϑ ′
+ z+ y+ ωM 2
rM dA (3.70)
A Iy Iz Iω
" #
N Z 2 My Z Mz Z MωM Z
=ϑ rM dA +
′
z rM dA +
2
y rM dA +
2 2
ωM rM dA
A Iy Iz Iω

If we compute the radial distance rM from the shear center in the principal axes system, we

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 92
obtain:

2
rM = (y − yM )2 + (z − zM )2 (3.71)

= y 2 − 2 y yM + yM
2
+ z 2 − 2 z zM + zM
2

= y 2 + z 2 − 2 (y yM + z zM ) + yM
2
+ zM
2

Before we continue our discussion, let us make the following definitions:

Z
IP S = (y 2 + z 2 ) dA (3.72)

1 Z
rSy = z (y 2 + z 2 ) dA (3.73)
Iy
1 Z
rSz = y (y 2 + z 2 ) dA (3.74)
Iz
1 Z
rM ω = ωM (y 2 + z 2 ) dA (3.75)
Iω

We refer to IP S as the polar moment of inertia of the cross section. The cross–sectional distances
rSy and rSz possess the unit of length.

rSy vanishes for cross-sections that are symmetrical about the y–axis, whereas rSz vanishes for
cross-sections that are symmetrical about the z–axis. The cross-sectional property rM ω does
not have a unit. For cross-sections that are symmetrical to both the y– and the z–axis, rSy ,
rSz and rM ω are equal to zero.

Using the definitions (3.72)–(3.75), we can write the integral expressions in (3.70) as
Z Z  
2
rM dA = y 2 − 2 y yM + yM
2
+ z 2 − 2 z zM + zM
2
dA = IP S + yM
2
A + zM
2
A

Z Z  
z 2
rM dA = z y 2 − 2 y yM + yM
2
+ z 2 − 2 z zM + zM
2
dA
Z Z
= z (y 2 + z 2 ) dA − 2 zM z 2 dA = rSy Iy − 2 zM Iy

Z Z  
y 2
rM dA = y y 2 − 2 y yM + yM
2
+ z 2 − 2 z zM + zM
2
dA
Z Z
= y (y + z ) dA − 2 yM
2 2
y 2 dA = rSz Iz − 2 yM Iz

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 93
 Z Z  
2
ωM rM dA = ωM y 2 − 2 y yM + yM
2
+ z 2 − 2 z zM + zM
2
dA = rM ω Iω

Hence, (3.70) can be rewritten as:

"
N 
Mxσ = ϑ ′
IP S + yM
2
A + zM
2
A +
A
#
My Mz MωM
+ (rSy Iy − 2 zM Iy ) + (rSz Iz − 2 yM Iz ) + rM ω Iω
Iy Iz Iω

IP S
   
=ϑ N ′
+ yM
2
+ zM
2
+ My (rSy − 2 zM ) + Mz (rSz − 2 yM ) + MωM rM ω
A

With the following abbreviations

IP S
i2P S = (3.76)
A
i2P M = i2P S + yM
2
+ zM
2
(3.77)

by = rSy − 2 zM (3.78)

bz = rSz − 2 yM (3.79)

we obtain
h   i
Mxσ = ϑ′ N i2P S + yM
2
+ zM
2
+ My (rSy − 2 zM ) + Mz (rSz − 2 yM ) + MωM rM ω

or
h i
Mxσ = ϑ′ N i2P M + My by + Mz bz + MωM rM ω (3.80)

respectively.

If we substitute the terms in brackets by the following expressions

KT σ = K̊T σ + KTTσ (3.81)

K̊T σ = N i2P S + My (rSy − zM ) + Mz (rSz − yM ) + MωM rM ω

 
KTTσ = N yM
2
+ zM
2
− My zM − MZ yM

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 94
We finally get:

Mxσ = KT σ ϑ′ (3.82)

Remark: Expressions for the cross-sectional properties rSy and rSz , as well as for the distances
from the shear center yM and zM can be found in Der Stahlbau, Chr. Petersen, page 375.

Cross Section in System

of Principal Axes yM zM rSy rSz by bz

in general ̸= 0 ̸= 0 ̸= 0 ̸= 0 = rSy − 2 zM = rSz − 2 yM

symmetrical y =0 ̸= 0 ̸= 0 =0 = rSy − 2 zM =0
to z – axis
z

symmetrical ̸= 0 =0 =0 ̸= 0 =0 = rSz − 2 yM
y
to y – axis
z

symmetrical y =0 =0 =0 =0 =0 =0
to both axes
z

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 95
3.4.4 Loads and Stress Resultants of the Deformed Configuration and
their Transformation to the Reference Frame of the Undeformed
Structure

Pz

Ty
y

My
x
Qỹ
Mx

ỹ Mω
Mỹ

Mx̃ L
z x̃
Tz z̃
Mω̃
Qz̃
Nx̃

Mz Mz̃

p
pz
v
M
w
py
S
M y

B
S
ỹ

z
z̃

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 96
The axes of the coordinate frame of the deformed state are rotated by angles ϑ, v ′ and w′
(related to the coordinate frame of the reference configuration). The same facts are valid for
the vectors of the internal forces and moments. Thus, these vectors have to be transformed to
the reference frame.
ϑ v′

y v′ w′
w′ ϑ
w′ ϑ x
ỹ

Sequence of rotations in the plot:

1. Rotation about the x–Axis by ϑ x̃
2. Rotation about the z–Axis by v ′ v′
z
3. Rotation about the y–Axis by w′

z̃

Inclination of the gravitational axis compared to the shear axis because of ϑ ̸= const.:
w′
y S
Ñ
zM · ϑ′
zM · dϑ
zM
dϑ zM M yM

Ñ
x dϑ yM · dϑ yM · ϑ′
yM z
ϑ

v′

Remark: Ñ is acting in the centre of gravity.

The transformation of the vectors of stress resultants of the deformed configuration

(x̃, ỹ, z̃) to the reference system is shown in the following plot.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 97
 S x
w′ − ϑ′ yM

x̃ Mz̃
M Nx̃
w′

Mx̃
z̃ Qz̃
z
y − yM
ϑ
Qỹ
Mỹ
ϑ

S x Qz̃
v ′ + ϑ′ zM
Mỹ
M x̃ Nx̃ z − zM Mz̃
v′

Mx̃
ỹ Qỹ
y

Matrical notation (the ∗ indicating - up to now - undefined entries):

1 −v ′ −w′ 0 0 0
    
L Nx̃
    
 v + zM ϑ′ 1 0 0 0
 ′
 Ty 
 
−ϑ   Qỹ 
 
    
1 0 0 0
 ′
 Tz  w − yM ϑ′ ϑ   Qz̃ 
   
=
    
1 −(v ′ + zM ϑ′ )
    
′ ′ 
M 
 x

 ∗ ∗ ∗ w − yM ϑ  Mx̃ 
    
My  ∗ 0 0 v′ 1 ϑ  Mỹ 
    

    
Mz ∗ 0 0 −w′ −ϑ 1 Mz̃

The above given entries to the transformation matrix result from a split of the vectors of stress
resultants (actual state) with respect to the reference frame (undeformed state).

In the actual configuration, the normal force (N ≈ L) acts in the relocated centre of gravity
S̃.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 98
Related to the cross-sectional axes of the reference configuration a positive normal force pro-
duces positive moments N · (w − ϑ yM ) and N · (v + ϑ zM ).
Related to the shear of the reference configuration the shearing forces (of the deformed config-
uration) are acting with lever arms v and w (which are the displacements of the shear centre
itself). Regarding their sign (and sense of action), we get torsional moments −Qy · w and
Qz · v.

v S
y
w zM
S̃ ϑ zM
M
ϑ yM
(S) w
ϑ
zM
∼ Qỹ

ϑ M̃ v

ỹ ∼ Qỹ
z
ϑ

z̃

yM yM

Matrical notation (the ∗ still indicating undifined entries):

1 −v ′ −w′ 0 0 0
    
L Nx̃
    
 v + zM ϑ′ 1 0 0 0
 ′
 Ty 
 
−ϑ   Qỹ 
 
    
1 0 0 0
 ′
 Tz  w − yM ϑ′ ϑ   Qz̃ 
   
=
    
1 −(v ′ + zM ϑ′ )
    
′ ′ 
M 
 x

 ∗ −w v w − yM ϑ  Mx̃ 
    
My   w − ϑ yM 0 0 v′ 1 ϑ  Mỹ 
    
    
Mz v + ϑ zM 0 0 −w′ −ϑ 1 Mz̃

In the coordinate system of the reference configuration, the normal force Ñ (of the deformed
state) can be decomposed into its component Ñ · (v ′ + zM ϑ′ ) which is in parallel to the y–axis
and in another component Ñ · (w′ − yM ϑ′ ) which is in parallel to the z–axis.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 99
If we confine our considerations to linear terms, we again get torsional moments as fol-
lows:

Ñ · (w′ − yM ϑ′ ) · (v − yM + ϑ zM ) ≈ −Ñ · (w′ − yM ϑ′ ) · yM and

− Ñ · (v ′ + zM ϑ′ ) · (w − zM − ϑ yM ) ≈ Ñ · (v ′ + zM ϑ′ ) · zM

S
y
zM
Ñ (v ′ + zM ϑ′ )
S̃ ϑ zM
M
ϑ yM
(S) w
ϑ
zM
Ñ (w′ − zM ϑ′ )
M̃ v
z

yM yM

Matrical notation (now with all entries defined):

1 −v ′ −w′ 0 0 0
    
L Ñx
    
 Ty 
  
 v ′ + zM ϑ′ 1 −ϑ 0 0 0   Q̃y 
 
    
 Tz  w′ − yM ϑ′ ϑ 1 0 0 0   Q̃z 
    

=
    
z (v ′ + zM ϑ′ ) − yM (w′ − yM ϑ′ ) −w 1 −(v ′ + zM ϑ′ )
    
′ ′ 
M 
 x  M v w − yM ϑ  M̃x 
    
My  w − ϑ yM 0 0 v′ 1 ϑ  M̃y 
    

    
Mz v + ϑ zM 0 0 −w′ −ϑ 1 M̃z

sA = TGA
⃗ sG (3.83)

Remark: The portion K̊T σ ϑ′ of the complete torsional moment, specified in Chapter 3.4.3, is
included in Mx̃ .

Effect of applied distributed forces of the actual configuration related to the co-
ordinate system of the reference state; influence of displacements; displacements
of the axis of elastic supports

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 100
 yp
ypM yM
v S
y w
ypM zM
v
w M zp
M̃
zpM

v P zpM
ϑ
w
(P)
ypM ϑ
P̃
py zpM ϑ

pz z

Displacements of the axis of loads: Displacements of the axis of elastic supports:

vpM = v − zpM ϑ vBM = v − zBM ϑ

wpM = w + ypM ϑ wBM = w + yBM ϑ

Matrical notation (points indicating zero entries):

1
    
px . . . . . px̃
    
 py 
  
 . 1 . . . .
  pỹ 
 
    
 pz  . . 1 . . .   pz̃ 
    
  
= (3.84)
   
−(w + ypM ϑ) v − zpM ϑ 1 .
    
m   . .  mx̃ 
 
 x  
    
my  w − ϑ yM . . . 1 .  mỹ 
    
    
mz v + ϑ zM . . . . 1 mz̃

Remark: px is acting in the axis of gravity.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 101
3.4.5 Principle of Virtual Displacements: Resulting Formulation

Using eqs. (3.83) and (3.84) we can rewrite eq.(3.67) (to recall: displacements are related to
the coordinate system of the reference state):
Z h i
Ñ − Q̃y v ′ − Q̃z w′ δu′

+ Ñ (zM (v ′ + zM ϑ′ ) − yM (w′ − yM ϑ′ )) − Q̃y w + Q̃z v

+ M̃ xV + M̃ xσ − M̃y (v ′ + ϑ′ zM ) + M̃z (w′ − ϑ′ yM ) dϑ′
h i
+ −Ñ (w − yM ϑ) − M̃x v ′ − M̃ y − M̃z ϑ δw′′
h i
+ −Ñ (v + zM ϑ) − M̃x w′ − M̃y ϑ − M̃ z δv ′′
h i
+ M̃ ω δϑ′′

+ (v − zBM ϑ) kv δ(v − zBM ϑ) + (w + yBM ϑ) kw δ(w + yBM ϑ) + ϑ kϑ dϑ

− px δu − py (δv − zpM δϑ) − pz (δw + ypM δϑ)

h i
− mx − py (w + ypM ϑ) + pz (v − zpM ϑ) δϑ
!
− [−my − px (w − yM ϑ)] δw′ − [−mz − px (v + zM ϑ)] δv ′ dx

h ib
− L δu + T y δv + T z δw + M x δϑ − M y δw′ − M ω δϑ′ =0 (3.85)
a

The above present problem is obviously non–linear cause the stress resultants Ñ , Q̃y , Q̃z , M̃x
a. s. o. are depending upon the displacements u, v, w and ϑ.

Thus, we have to choose a solution procedure in steps: in terms, which include products of
displacements and stress resultants (both existing in the deformed state), we accept internal
forces and moments resulting from a first order analysis to be sufficiently exact. We denote
these stress resultants by a small circle, for instance M̊y . (In German literature, the complete
set of circled internal forces and moments is denoted as deformed state).

Those internal forces, not being multiplied by displacements, are substituted by the respective

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 102
products of strain and stiffness of the bar.

Ñ = EA u′ M̃y = −EIy w′′ M̃z = −EIz v ′′ M̃ω = −EIω ϑ′′ M̃xV = GIT ϑ′

Remark: The last term Mω rM ω in K̊T σ is neglected henceforth.

Thus, we get:
Z h i
EA u′ − Q̊y v ′ − Q̊ w′ δu′


+ N̊ (zM (v ′ + zM ϑ′ ) − yM (w′ − yM ϑ′ )) − Q̊y w + Q̊z v

+ GIT ϑ + K̊T σ ϑ − M̊y (v + ϑ zM ) + M̊z (w − ϑ yM ) δϑ′
′ ′ ′ ′ ′

h i
+ −N̊ (w − yM ϑ) − M̊x v ′ + EIy w′′ − M̊z ϑ δw′′
h i
+ −N̊ (v + zM ϑ) + M̊x v ′ + M̊y ϑ + EIz v ′′ δv ′′

+ [EIω ϑ′′ ] δϑ′′

+ kv (v − zBM ϑ) δv + kw (w + yBM ϑ) δw
h i
+ kv (−zBM v + zBM
2
ϑ) + kw (yBM w + yBM
2
ϑ) + kϑ ϑ δϑ

− px δu − py (δv − zpM δϑ) − pz (δw + ypM δϑ)

h i
− mx − py (w + ypM ϑ) + pz (v − zpM ϑ) δϑ
!
− [−my − px (w − yM ϑ)] δw − [−mz − px (v + zM ϑ)] δv dx
′ ′

h ib
− L δu + T y δv + T z δw + M x δϑ − M y δw′ − M z δv ′ − M ω δϑ =0 (3.86)
a

Eq.(3.86) is further developed taking account of the following items:

1. Setting (see chapter 3.4.2):

K̊T σ ≈ N̊ i2P S + M̊y (rSy − zM ) + M̊z (rSz − yM ) + M̊ω rM ω

2. We apply integration by parts to eliminate derivatives of variations of displacements and

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 103
 derivatives of virtual displacements in the integrand:
Z  
−δA = {. . .}δu(x) + {. . .}δϑ(x) + {. . .}δw(x) + {. . .}δv(x) dx

b b
+ [{. . .}δu]ba + [{. . .}δϑ(x)]ba + [{. . .}δϑ(x)′ ]a + [{. . .}δw(x)′ ]a
b
− [{. . .}δw(x)]ba + [{. . .}δv(x)′ ]a − [{. . .}δv(x)]ba

3. We collect terms with respect to virtual displacement functions.

4. We accept first order analysis equilibrium equations to hold for the complete set of circled
stress resultants (Schnittgrößen des Grundzustandes).

Having achieved some auxiliary calculations (including rearrangement of terms) we get the
final form of the principle of virtual displacements:
" #
Z
−δW =  ′ ′ ′ ′
− {EA u − Q̊y v − Q̊z w } − px δu(x)

"
h i′
+ {EIω ϑ′′ }′′ − {GIT ϑ′ }′ − N̊ (i2P M ϑ′ + zM v ′ − yM w′ )

h i′ h i′ h i′
− M̊y by ϑ′ + M̊y v ′′ − M̊z bz ϑ′ − M̊z w′′ − M̊ω rM ω ϑ′

+ kv (−zBM · v + zBM
2
ϑ) + kw (yBM · w + yBM
2
ϑ) + kϑ ϑ
#
+ mz w′ − my v ′ + py ypM ϑ + pz zpM ϑ − mx + py zpM − pz ypM δϑ(x)

"
+ {−N̊ (w − yM ϑ)′ }′ + {−M̊x v ′ + EIy w′′ − M̊z ϑ}′′

#
+ kw (w + yBM ϑ) − m′y − pz δw(x)

"
+ {−N̊ (v + zM ϑ)′ }′ + {−M̊x w′ + M̊y ϑ + EIz v ′′ }′′

# 

+ kv (v − zBM ϑ) − m′z − py δv(x) dx

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 104
 " #b
+ {EA u − Q̊y v − Q̊z w − L} δu(x)
′ ′ ′

a
"
h i
+ N̊ i2P S ϑ′ + zM v ′ − yM w′ + M̊y by ϑ′ − M̊y v ′ + Q̊z v

#b
+ M̊z bz ϑ′ + M̊z w′ − Q̊ w + GIT ϑ′ − (EIω ϑ′′ )′ − M x ) δϑ
a
" #b
+ {EIω ϑ + M ω } δϑ(x)
′′ ′

a
" #b
+ {−N̊ (w − yM ϑ) − M̊x v + EIy w − M̊z ϑ + M y } δw(x)
′ ′′ ′

a
" #b
− ({−N̊ (w − yM ϑ) − M̊x v ′ + EIy w′′ − M̊z ϑ}′ − my − px (w − yM ϑ) + T z ) δw(x)
a
" #b
+ {−N̊ (v + zM ϑ) + M̊x w + EIz v + M̊y ϑ + M z } δv(x)
′ ′′ ′ ′

a
" #b
− ({−N̊ (v + zM ϑ) + M̊x w + EIz v + M̊y ϑ} − mz − px (v + zM ϑ) + T y ) δv(x)
′ ′′ ′
=0
a

3.4.6 Differential Equations, boundary Conditions, Internal Forces

3.4.6.1 Axial Stressing

D.equ.: (EA u′ − Q̊y v ′ − Q̊z w′ )′ + px = 0 (3.87)

BC.: EA u′ − Q̊y v ′ − Q̊z w′ − L = 0 (3.88)

The resulting axial force in point x in the reference configuration follows from the boundary
terms:

L = EA u′ − Q̊y v ′ − Q̊z w′ (3.89)

The normal force in the actual configuration we obtain applying a constitutive law:

Ñ = EA ũ′ ≈ EA u′ (3.90)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 105
Thus we get for the longitudinal force:

L = Ñ − Q̊y v ′ − Q̊z w′ (3.91)

The normal force being relevant for the dimensioning we calculate from:

Ñ = L + Q̊y v ′ + Q̊z w′ (3.92)

Remark 1: For any point x the barred stress resultants represent internal forces of the actual
configuration, being transformed to the coordinate frame of the reference configuration.
Thus we omit the bar henceforth.
Remark 2: Internal forces of the actual configuration are relevant quantities for dimensioning.
Henceforth they are denoted by a tilde, for instance M̃y .
Remark 3: In accordance with many authors we denote by Grundsatz all internal forces
calculated according to a first order analysis and mark them by a small circle (for instance
M̊y ).

3.4.6.2 Torsion

h  i′
D.equ.: {EIω ϑ′′ }′′ − {GIT ϑ′ }′ − N̊ i2P M ϑ′ + zM v ′ − yM w′
 ′  ′  
+ M̊y v ′′ − M̊y by ϑ′ − M̊z w′′ − M̊z bz ϑ′ − M̊ω rM ω ϑ′
   
+ kv − zBM · v + zBM
2
ϑ + kw yBM · w + yBM
2
ϑ + kϑ ϑ

+ mz w′ − my v ′ + (py ypM + pz zpM ) ϑ − mx + py zpM − pz ypM = 0 (3.93)

Boundary conditions:

BC. 1: GIT ϑ′ − (EIω ϑ′′ )′


+ N̊ i2P M ϑ′ + zM v ′ − yM w′ ) − Q̊y w + Q̊z v

− M̊y (v ′ − by ϑ′ ) + M̊z (w′ + bz ϑ′ ) + M̊ω rM ω ϑ′ − M x = 0 (3.94)

BC. 2: EIω ϑ′′ + M ω = 0 (3.95)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 106
Resulting torsional moment in point x, related to x–axis of the coordinate frame of the unde-
formed configuration:
 
Mx = GIT ϑ′ − (EIω ϑ′′ )′ + N̊ i2P M ϑ′ + zM v ′ − yM w′ − Q̊y w + Q̊z v

− M̊y (v ′ by ϑ′ ) + M̊z (w′ + bz ϑ′ ) + M̊ω rM ω ϑ′ (3.96)

The torsional moment of the deformed configuration we get using the constitutive laws of
uniform of non–uniform torsion:

M̃xV = GIT ϑ̃′ ≈ GIT ϑ′ (3.97)

M̃xW = M̃ω′ = −EIω ϑ̃′′′ ≈ −EIω ϑ′′′ (3.98)

M̃x = M̃xV + M̃xW

Using these relations and neglecting the last term in eq.(3.96) we can rewrite for the torsional
moment in the reference frame of the initial configuration
 
Mx = M̃x + N̊ i2P M ϑ′ + zM v ′ − yM w′ − Q̊y w + Q̊z v

− M̊y (v ′ − by ϑ′ ) + M̊ (w′ + bz ϑ) (3.99)

The torsional moment of the actual configuration, which governs the design of the member is
obtained as:
 
M̃x = Mx − N̊ i2P M ϑ′ + zM v ′ − yM w′ + Q̊y w − Q̊z v

+ M̊y (v ′ − by ϑ′ ) − M̊z (w′ + bz ϑ) (3.100)

3.4.6.3 Bending in the x–z –Plane

D.equ.:
{EIy w′′ }′′ − {N̊ (w − yM ϑ)′ }′ − {M̊z ϑ}′′ − {M̊x v ′ }′′ + kw (w + yBM ϑ)

− my ′ − pz = 0 (3.101)

BC.1:
− N̊ (w − yM ϑ) − M̊x v ′ + EIy w′′ − M̊z ϑ + M y = 0 (3.102)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 107
 BC.2:
 ′
− N̊ (w − yM ϑ) − M̊x v ′ + EIy w′′ − M̊z ϑ − my − px (w − yM ϑ) + T z = 0 (3.103)

Resulting bending moment My and transverse force Tz in point x, related to the reference
frame of the initial state:

My = − EIy w′′ + N̊ (w − yM ϑ) + M̊x v ′ + M̊z ϑ (3.104)

Tz = {−EIy w′′ + N̊ (w − yM ϑ) + M̊x v ′ + M̊z ϑ}′ + my + px (w − yM ϑ)

= My′ + my + px (w − yM ϑ) (3.105)

Constitutive law of the bending moment in the actual state:

M̃y = −EIy w̃′′ ≈ −EIy w′′ (3.106)

Thus, we get:

My = M̃y + N̊ (w − yM ϑ) + M̊x v ′ + M̊z ϑ (3.107)

Tz = {M̃y + N̊ (w − yM ϑ) + M̊x v ′ + M̊z ϑ}′ + my + px (w − yM ϑ)

= My′ + my + N̊ ′ (w − yM ϑ) + px (w − yM ϑ) + N̊ (w − yM ϑ)′ +

+ M̊x′ v ′ + M̊x v ′′ + M̊z′ ϑ + M̊z ϑ′ (3.108)

= Q̊z + N̊ (w′ − yM ϑ′ ) + M̊x′ v ′ + M̊x v ′′ + Q̊y ϑ + M̊z ϑ′

Bending moment M̃y and shearing force Q̃z , (which are governing structural dimensioning and
design of members):

M̃y = My − N̊ (w − yM ϑ) − M̊x v ′ − M̊z ϑ (3.109)

Q̃z = Tz − N̊ (w′ − yM ϑ′ ) − M̊x′ v ′ − M̊x v ′′ − Q̊y ϑ − M̊z ϑ′ (3.110)

3.4.6.4 Bending in the x–y –Plane

D.equ.:
{EIz v ′′ }′′ − {N̊ (v + zM ϑ)′ }′ + {M̊y ϑ}′′ + {M̊x w′ }′′ + kv (v − zBM ϑ) − m′z − py = 0
(3.111)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 108
Boundary conditions:

BC.1:
− N̊ (v + zM ϑ) + M̊x w′ + M̊y ϑ + EIz v ′′ + M z = 0 (3.112)

BC.2:
{−N̊ (v + zM ϑ) + M̊x w′ + M̊y ϑ + EIz v ′′ }′ − mz − px (v + zM ϑ) + T y = 0 (3.113)

Resulting bending moment Mz and transverse force Ty in point x, related to the reference
frame of the initial state:

Mz = − EIz v ′′ + N̊ (v + zM ϑ) − M̊x w′ − M̊y ϑ (3.114)

Ty = {−EIz v ′′ + N̊ (v + zM ϑ) − M̊x w′ − M̊y ϑ}′ + pz (v + zM ϑ)}′ + mz + px (v + zM ϑ) + px yp

= Mz′ + mz + px (v + zM ϑ) (3.115)

Constitutive law of the bending moment in the actual state:

M̃z = −EIz ṽ ′′ ≈ −EIz v ′′ (3.116)

Thus, we get:

Mz = M̃z + N̊ (v − zM ϑ) − M̊x w′ + M̊y ϑ (3.117)

Ty = {M̃z + N̊ (v + zM ϑ) + M̊x w′ + M̊y ϑ}′ + mz + px (v + zM ϑ) + px yp

= Mz′ + mz + N̊ ′ (v + zM ϑ) + px (v + zM ϑ) + N̊ (v + zM ϑ)′ +

− M̊y′ ϑ − M̊y ϑ′ + M̊x′ w′ − M̊x w′′ (3.118)

= Q̊y + N̊ (v ′ + zM ϑ)′ − Q̊z ϑ − M̊y ϑ′ − M̊x′ w′ − M̊x w′′

Bending moment M̃z and shearing force Q̃y , (which are governing structural dimensioning and
design of members):

M̃z = Mz − N̊ (v + zM ϑ) + M̊x w′ + M̊y ϑ (3.119)

Q̃y = Ty − N̊ (v ′ + zM ϑ′ ) + Q̊z ϑ + M̊y ϑ′ + M̊x′ w′ + M̊x w′′ (3.120)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 109
Winter Term 2022/2023
Stability of Structures

3.4.6.5.1 Governing System of Differential Equations Illustrating Coupling Effects;

without Prestressing, without Imperfections

(EIz v ′′ )′′ + kv v

−(N̊ v ′ )′ +(M̊x w′ )′′ −[N̊ (zM ϑ)′ ]′ + (M̊y ϑ)′′ − kv zBM ϑ −m′z − py = 0

(EIy w′′ )′′ + kw w

−(M̊x v ′ )′′ −(N̊ w′ )′ +[N̊ (yM ϑ)′ ]′ − (M̊z ϑ)′′ + kw yBM ϑ −m′y − pz = 0

(EIω ϑ′′ )′′ − (GIT ϑ′ )′

−kv zBM · v +kw yBM · w +kv zBM

2
ϑ + kw yBM
2
ϑ + kϑ · ϑ

−(N̊ zM v ′ )′ +(N̊ yM w′ )′ −[(N̊ i2pM + M̊y by + M̊z bz + M̊ω rM ω ) ϑ′ ]′

3 Second Order Analysis and Stability of Space Frames

+M̊y v ′′ −M̊z w′′ +(py ypM + pz zpM ) ϑ +py zpM

−pz ypM

−my v ′ +mz w′ −mx = 0

110
Winter Term 2022/2023
Stability of Structures
3.4.6.5.2 Governing System of Differential Equations Illustrating Coupling Effects;
without Prestressing, including Imperfections v0 , w0 , ϑ0 (Imperfections in the Initial State do not produce Stresses.)

(EIz v ′′ )′′ + kv v

−(N̊ v ′ )′ +(M̊x w′ )′′ −[N̊ (zM ϑ)′ ]′ + (M̊y ϑ)′′ − kv zBM ϑ

−(N̊ v0 ′ )′ +(M̊x w0 ′ )′′ −[N̊ (zM ϑ0 )′ ]′ + (M̊y ϑ0 )′′ −m′z − py = 0

(EIy w′′ )′′ + kw w

−(M̊x v ′ )′′ −(N̊ w′ )′ +[N̊ (yM ϑ)′ ]′ − (M̊z ϑ)′′ + kw yBM ϑ

−(M̊x v0 ′ )′′ −(N̊ w0 ′ )′ +[N̊ (yM ϑ0 )′ ]′ − (M̊z ϑ0 )′′ −m′y − pz = 0

(EIω ϑ′′ )′′ − (GIT ϑ′ )′

−(N̊ zM v ′ )′ +(N̊ yM w′ )′ −[(N̊ i2P M + M̊y by + M̊z bz + M̊ω rM ω ) ϑ′ ]′

3 Second Order Analysis and Stability of Space Frames

+M̊y v ′′ −M̊z w′′

−kv zBM · v +kw yBM · w +kv zBM
2
ϑ + kw yBM
2
ϑ + kϑ ϑ
−my v ′ +mz w′ +(py ypM + pz zpM ) ϑ

−(N̊ zM v0 ′ )′ +(N̊ yM w0 ′ )′ −[(N̊ i2P M + M̊y by + M̊z bz + M̊ω rM ω ) ϑ0 ′ ]′

+M̊y v0 ′′ −M̊z w0 ′′ +(py ypM + pz zpM ) ϑ0 +py zpM

−pz ypM
−my v0 ′ +mz w0′ −mx = 0
111
Winter Term 2022/2023
Stability of Structures

3.4.6.5.3 Governing System of Differential Equations for Cross-Sections, Symmetric to the z-Axis;
without Prestressing, including Imperfections v0 , w0 , ϑ0 ; yM = bz = rM ω = 0

(EIz v ′′ )′′ + kv v

−(N̊ v ′ )′ +(M̊x w′ )′′ −[N̊ (zM ϑ)′ ]′ + (M̊y ϑ)′′ − kv zBM ϑ

−(N̊ v0 ′ )′ +(M̊x w0 ′ )′′ −[N̊ (zM ϑ0 )′ ]′ + (M̊y ϑ0 )′′ −m′z − py = 0

(EIy w′′ )′′ + kw w

−(M̊x v ′ )′′ −(N̊ w′ )′ −(M̊z ϑ)′′ + kw yBM ϑ

−(M̊x v0 ′ )′′ −(N̊ w0 ′ )′ −(M̊z ϑ0 )′′ −m′y − pz = 0

(EIω ϑ′′ )′′ − (GIT ϑ′ )′

−(N̊ zM v ′ )′ −[(N̊ i2P M + M̊y by + M̊ω rM ω ) ϑ′ ]′

+M̊y v ′′ −M̊z w′′

3 Second Order Analysis and Stability of Space Frames

−kv zBM · v +kw yBM · w +kv zBM

2
ϑ + kw yBM
2
ϑ + kϑ ϑ
−my v ′ +mz w′ +(py ypM + pz zpM ) ϑ

−(N̊ zM v0 ′ )′ −[(N̊ i2P M + M̊y by + M̊ω rM ω ) ϑ0 ′ ]′

+M̊y v0 ′′ −M̊z w0 ′′ +(py ypM + pz zpM ) ϑ0 +py zpM

−pz ypM
−my v0 ′ +mz w0′ −mx = 0
112
Winter Term 2022/2023
Stability of Structures
3.4.6.5.4 Governing System of Differential Equations for Double Symmetric Cross Sections;
without Prestressing, including Imperfections v0 , w0 , ϑ0 ; yM = zM = by = bz = rM ω = 0; iP M = iP S

(EIz v ′′ )′′ + kv v

−(N̊ v ′ )′ +(M̊x w′ )′′ +(M̊y ϑ)′′ − kv zBM ϑ

−(N̊ v0 ′ )′ +(M̊x w0 ′ )′′ +(M̊y ϑ0 )′′ −m′z − py = 0

(EIy w′′ )′′ + kw w

−(M̊x v ′ )′′ −(N̊ w′ )′ −(M̊z ϑ)′′ + kw yBM ϑ

−(M̊x v0 ′ )′′ −(N̊ w0 ′ )′ −(M̊z ϑ0 )′′ −m′y − pz = 0

(EIω ϑ′′ )′′ − (GIT ϑ′ )′

3 Second Order Analysis and Stability of Space Frames

+M̊y v ′′ −M̊z w′′ −(N̊ i2P S ϑ′ )′ − (N̊ i2P S ϑ0 ′ )′

−kv zBM · v +kw yBM · w +kv zBM

2
ϑ + kw yBM
2
ϑ + kϑ ϑ

−my v ′ +mz w′ +(py ypM + pz zpM ) ϑ

+M̊y v0 ′′ −M̊z w0 ′′ +(py ypM + pz zpM ) ϑ0 +py zpM

−pz ypM
−my v0 ′ +mz w0′ −mx = 0
113
3.4.7 Internal Forces in the Reference Frames of the Initial and of the
Actual States – Matrix Formulation

We can collect equations (3.92), (3.120), (3.110), (3.100), (3.109) and (3.119) in matrix notation
and get:
  
0 v′ w′ 0 0 0 N̊
   
Ñ L
      
−(v ′ + zM ϑ′ ) 0 0 0 0  Q̊
 
 Q̃y  ϑ  Ty 
   
  y


     
−(w′ − yM ϑ′ ) 0 0 0 0
 
 Q̃z  −ϑ   Q̊z  T
    
   z

= +
     
−z v ′ + y w ′ − i2 ϑ′ 0
    
M̃ 
 x  M M PM w −v v ′ − by ϑ′ ′ ′ 
−w − bz ϑ  M̊x  Mx 
  

      
M̃y  −(w − yM ϑ) 0 0 −v ′ 0 −ϑ M̊y  My 
      
  
    
−(v + zM ϑ) 0 0 0
 
′
M̃z w ϑ M̊z Mz

sG = TAG
⃗ s̊A + sA (3.121)

Remark: In contrary to eq.(3.83) in the above written equation all parts of the moment
Mxσ = N̊ i2P M ϑ′ + M̊y by ϑ′ + M̊z bz ϑ′ form eq. (3.80) are explicitly written down.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 114
3.4.8 Special Case with an Exact Solution:
Girder in Yoke Supports having a single symmetric Cross–section
and M y = const.

Until now very few analytical (exact) solutions of the system of differential equations presented
in Chapters 3.4.6.5.1–3.4.6.5.4 are known.

One of them holds for the girder and its loads, plotted in Fig. 3.3. The solution is established
in the following

mx p
My My λ py
M
y S
x
kv
B
l kϑ

z
py z

x
l

y
Figure 3.3

First order analysis: loads and stress resultants (1st order analysis)

p̊z = 0 (3.122) N̊ = 0 (3.125)

p̊x = 0 (3.123) M̊y = const. (3.126)

Q̊z = 0 (3.124) M̊x = 0 (3.127)

Loads for a second order analysis:

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 115
 πx
 
py = pym sin (3.128) ypM = 0 (3.130)
l
πx
 
mx = mxm sin (3.129)
l
Imperfections:

πx
 
v0 = v0m sin (3.131)
l
πx
 
ϑ0 = ϑ0m sin (3.132)
l

Thus the system of governing equations of Sections 3.4.6.5.1 - 3.4.6.5.4 may be rewritten:

EIω ϑ′′′′ − GIT ϑ′′ + kv (−zBM · v + zBM

2
ϑ) + kϑ ϑ + M̊y (v ′′ − by ϑ′′ )

= mx − py zpM − M̊y (v0 ′′ − by ϑ0 ′′ ) (3.133)

EIz v ′′′′ + M̊y ϑ′′ + kv (v − zBM ϑ) = py − M̊y ϑ0 ′′ (3.134)

These equations only contain derivatives of order 0, 2 and 4 of the unknown displacement
functions. The trial functions

πx πx
v = vm sin (3.135) ϑ = ϑm sin (3.136)
l l
end up in a system of two algebraic equations to determine the constant values vm and ϑm :

A vm = b (3.137)
π2 π4 π2 π2
+ GIT + kv zBM + kϑ + M̊y by
 
2
−kv · zBM − M̊y l2
EIω l4 l2 l2 
A= 


 
π4 π2
EIz l4
+ kv −kv · zBM − M̊y l2

2
mxm − zpM pym + M̊y [v0m − by ϑ0m ] πl2
 
vm
 

vm =   b=
   

 
π2
ϑm pym + M̊y ϑ0m l2

Geometric boundary conditions:

v(0) = v(l) = 0 (3.138)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 116
 ϑ(0) = ϑ(l) = 0 (3.139)

Boundary conditions of forces (in general to be established using equations (3.121))

Mz (0) =M̃z (0) + N̊ (0) (v(0) + zM ϑ(0)) − M̊x (0) w′ (0) − M̊y (0) ϑ(0) (3.140)

= − EIz v ′′ (0) − M̊y (0) ϑ(0) = 0 (3.141)

Analogously

Mz (l) = − EIz v ′′ (l) − M̊y (l) ϑ(l) = 0 (3.142)

Obviously all boundary conditions are met.

Abbreviations:

π4 π2
B = EIz + kv (3.143) D = Ts + TB + M̊y by (3.146)
l4 l2
 π2  π2 π2
TS = EIω + GI T (3.144) M̊ = M̊y + kv zBM (3.147)
l2 l2 l2
TB = kv zBM
2
+ kϑ (3.145)

Solution of the system of algebraic equations:

1
" #
π2 π2
  
vm = mxm − zpM pym + M̊y (v0m − by ϑ0m ) 2 M̊ + pym + M̊y ϑ0m 2 D
−M̊ 2 + B D l l
(3.148)

1
" #
π2 π2
  
ϑm = mxm − zpM pym + M̊y (v0m − by ϑ0m ) 2 B + pym + M̊y ϑ0m 2 M̊
−M̊ 2 + B D l l
(3.149)

Internal forces and moments, relevant for design (see section 3.3 ”Bending and Torsion in a
1st Order Analysis”):

π πx
M̃xV = GIT ϑ′ = GIT ϑm cos (3.150)
l l
π3 πx
M̃xW = −EIω ϑ′′′ = EIω ϑm cos (3.151)
l3 l
M̃y ≈ My = M̊y = const. (3.152)

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 117
 π2 πx
M̃z = −EIz v ′′ = EIz vm 2
sin (3.153)
l l
π2 πx
M̃ω = −EIω ϑ′′ = EIω ϑm 2
sin (3.154)
l l

Forces in the spring bed:

pyB = − (v − zBM ϑ) kv (3.155)

mxB = − ϑ kϑ (3.156)

Remark: To give proof of a correct solution, stress resultants in the coordinate system of the
initial state are calculated using equations (3.121) and (3.83).

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 118
 Example 3.1

p
My My λ py
M
S
x y

z
z
λ py

Cross–sectional quantities and cross–

x sectional stiffnesses:
l
EIy = 16923,9 kNm2

y EIz = 1576,1 kNm2

GIT = 9,91 kNm

zM = −8,8 cm

by = 22,2 cm

EIω = 14,95 kNm4

Span: l = 6 m kv = kϑ = 0

Loads:

M̊y = 30 kNm
π·x
 
py = 1,5 sin kN
zpM = 0
6 m

First order analysis (selected results):

1,5 · 64
max v = = 0,01267 m
π 4 1575

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 119
 1,5 · 62
max Mz = = 5.4713 kNm
π2
ϑ=0

1,5 · 6
Ay = = 2,865 kN
π
Mx = 0

Second order analysis:

Matrix A:

π2
a11 = −30 = −8,228
62
π4 π2 π2
a12 = 14,95 4 + 9,91 2 − 30 (−0,222) 2 = 1,12 + 2,72 + 1,89 = 5,668
6 6 6
π4
a21 = 1576,134 = 118,5
64
π2
a22 = −30 = −8,228
62
Load vector of the system of linear equations A v0 = b

b1 = 0
b2 = 1,5

Maxima of the displacement functions:

vm = 0,0148m
ϑm = 0,02044

Displacements and internal forces:

π
 
v(x) = 0,01408 sin x
6
π
 
ϑ(x) = 0,02044 sin x
6
π π π
   
M̃xV = 0,02044 · 9,91 cos x = 0,1061 · x
6 6 6

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 120
 π3 π π
   
M̃xW = 0,02044 · 14,95 cos x = 0,04388 · cos x
63 6 6
π
 
M̃x = 0,1500 · cos x
6
My = M̊y = 30 kNm
π2 π π
   
M̃z = 0,01408 · 1576,1 2 sin x = 6,806 · sin x
6 6 6
π2 π π
   
M̃ω = 0,0244 · 14,95 2 sin x = 0,08381 · sin x
6 6 6

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 121
3.5 Torsional Buckling and Lateral–Torsional Buckling of
Beam Columns and Girders of Uniform Cross–Section;
Governing Equations, Analytical Solutions and Structural
Behaviour

Throughout this chapter the following assumptions hold:

1. (EIω )′ = (EIy )′ = (EIz )′ = (GIT )′ = (EA)′ = yM ′ = zM ′ = 0

2. px = 0 → N̊ ′ = 0 → N̊ = const.
3. my = mz = mω = 0
4. M̊ω = 0

Governing differential equations, boundary conditions, first order internal forces (circled quan-
tities) Rem.: First order internal forces (circled quantities, henceforth: f.o.a) are determined in
any special case and depend upon the shape of the cross-section, the supports and the loads.

Torsion:

EIω ϑ′′′′ − GIT ϑ′′ − kv zBM (v − zBM ϑ) + kw yBM (w + yBM ϑ) + kϑ ϑ

− N̊ (i2P M ϑ′′ + zM v ′′ − yM w′′ ) + M̊y (v ′ − by ϑ′ )′ − M̊z (w′ + bz ϑ′ )′

+ (py ypM + pz zpM ) ϑ − mx + py zpM − pz ypM = 0

BC. 1: Mx = GIT ϑ′ − EIω ϑ′′′

+ N̊ (i2P M ϑ′ + zM v ′ − yM w′ ) − Q̊y w + Q̊z v

− M̊y [v ′ − by ϑ′ ] + M̊z [w′ + bz ϑ]

BC. 2: Mω = − EIω ϑ′′

Bending in the x–z–plane:

EIw ′′′′ + kw (w + yBM ϑ) − pz − N̊ (w − yM ϑ)′′ − (M̊z ϑ)′′ − (M̊x v ′ )′′ = 0

BC. 1: My = −EIy w′′ + N̊ (w − yM ϑ) + M̊x v ′ + M̊z ϑ

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 122
 BC. 2: Tz = −EIy w′′′ + N̊ (w − yM ϑ)′ + {M̊x v ′ }′ + {M̊z ϑ}′

Bending in the x–y–plane:

EIv ′′′′ + kv (v − zBM ϑ) − py − N̊ (v + zM ϑ)′′ + (M̊y ϑ)′′ + (M̊x w′ )′′ = 0

BC. 1: Mz = −EIz v ′′ + N̊ (v + zM ϑ) − M̊x w′ − M̊y ϑ

BC. 2: Ty = −EIz v ′′′ + N̊ (v + zM ϑ)′ − {M̊x w′ }′ − {M̊y ϑ}′

3.5.1 Buckling and torsional Buckling of Thin–Walled Bars

of Double Symmetric Cross–Sections
with a Centric Concentrated Compressive Force

Additional Assumptions:

Cross–section yM = zM = 0 b y = bz = 0 EIω

First order analysis: N̊ < 0 pz = M̊y = Q̊z = 0

py = M̊z = Q̊y = 0

K̊T σ = N̊ i2P S mx = M̊x = 0

Governing equations and boundary conditions:

Torsion:

EIω ϑ′′′′ − GIT ϑ′′ − N̊ i2P S ϑ′′ = 0 BC.1: Mx = GIT ϑ′ − EIω ϑ′′′ + N̊ i2P S ϑ′
BC.2: Mω = −Eω ϑ′′

Bending in the x–z–plane:

EIy w′′′′ − N̊ w′′ = 0 BC.1: My = −EIy w′′ + N̊ w

BC.2: Tz = −EIy w′′′ + N̊ w′

Bending in the x–y–plane:

EIz v ′′′′ − N̊ v ′′ = 0 BC.1: Mz = −EIz v ′′ + N̊ v

BC.2: Ty = −EIz v ′′′ + N̊ v ′

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 123
 Reference N̊
state
P
y

1. Neighbouring state
Buckling in the x–y–plane
z

P x
v 2. Neighbouring state
Mz
Buckling in the x–z–plane

w
My
ϑ 3. Neighbouring state:
Mx Torsional Buckling without side sway

1
  2 
1. P Di = · GIT + l EIωπ
i2P S
1 π 2 EIz
!
GIT l2 Iω
= 2 · · · + ; Remark: Iz < Iy
iP S l2 EIz π 2 Iz
GIT l2 Iω
c2 = · 2+ ; Radius of gyration
EIz π Iz
1 π 2 EIz 2
⇒ P Di = 2 · ·c
iP S l2
π2
2. P Kiy = · EIz
l2
π2 c2 π 2 EIz
3. P Kiy ≥ P Di : · EI z ≥ · iP S > c
l2 i2P S l2

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 124
 4. Remark concerning the notion (or basic concept) of slenderness:

π 2 EI
For Euler columns we get: Pki =
s2k
Pki π 2 EI
Respective buckling stress: σki = =
A A s2k
s
I
Radius of gyration i=
A
sk
Slenderness ratio: λ=
i
π 2 E i2
Hence: σki =
s2k
π2 E π2 E
σki = 2 ≤ σyield ; λ2 =
λ σki
Remark: Bars, having equal slenderness ratios (but possibly different supports) show equal
buckling stresses (at least in the elastic range).

Analogously, we define for bars in torsion:

π2 E π2 E i2P S
λ2V = = =
1 1 GIT Iω
!
π 2 EIω
 
σDi
GIT + + 2
A i2P S l2 A π2 E l

π2 E
σDi =
λ2V

Bars with similar comparing slenderness ratios have similar torsional buckling loads.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 125
3.5.2 Buckling and Lateral Torsional Buckling of Bars
of Arbitrary Thin–Walled Cross–Section
with a Centric Concentrated Compressive Force

Additional assumptions:

Cross–section: yM , z M
by , b z
EIω
First order analyis: N̊ < 0
pz = M̊y = Q̊z = 0
py = M̊z = Q̊y = 0
mx = M̊x = 0
K̊T σ = N̊ i2P M

Governing equations and boundary conditions (supports):

Torsion:

EIω ϑ′′′′ − GIT ϑ′′ − N̊ (i2P M ϑ′′ + zM v ′′ − yM w′′ ) = 0

BC.1: Mx = GIT ϑ′ − EIω ϑ′′′ + N̊ (i2P M ϑ′ + zM v ′ − yM w′ )

BC.2: Mω = −EIω ϑ′′

Bending in the x–z–plane:

EIω w′′′′ − N̊ (w − yM ϑ)′′ = 0

BC.1: My = −EIy w′′ + N̊ (w − yM ϑ)

BC.1: Tz = −EIy w′′ + N̊ (w − yM ϑ)′

Bending in the x–y–plane:

EIω v ′′′′ − N̊ (v + zM ϑ)′′ = 0

BC.1: Mz = −EIz v ′′ + N̊ (v + zM ϑ)

BC.1: Ty = −EIz v ′′ + N̊ (v + zM ϑ)′

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 126
Solution of problem:

P P
x

Boundary conditions (supports):

position x = 0 : position x = l :

w=0 w=0

My (0) = −EIy w′′ + N̊ (w − yM ϑ) = 0 My (l) = −EIy w′′ + N̊ (w − yM ϑ) = 0

v=0 v=0

Mz (0) = −EIz v ′′ + N̊ (v + zM ϑ) = 0 Mz (l) = −EIz v ′′ + N̊ (v + zM ϑ) = 0

ϑ=0 ϑ=0

Mω (0) = −EIω · ϑ′′ = 0 Mω (l) = −EIω · ϑ′′ = 0

Trial function:
π
v = A · sin ·x
l
 3
π π π π
v′ = · A · cos · x v ′′′ = − · A · cos ·x
l l l l
 2  4
π π π π
v =−
′′
· A · sin · x v IV
= · A · sin ·x
l l l l

π
w = B · sin ·x
l
π π
w′ = · B · sin · x
l l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 127
  2
π π
w =−′′
· B · sin ·x
l l

π
ϑ = C · sin ·x
l
π π
ϑ′ = · C · cos · x
l l
 2
π π
ϑ′′ = − · C · sin ·x
l l

The trial functions, chosen, satisfy all boundary conditions (see eqs. (3.121)). Substituting
the approach into the differential equations yields:
 4  2
π π π π
1. EIω · · C · sin x + GIT · C · sin x+
l l l l
h i  π 2 π
+ N̊ · iP M 2 · C + zM · A − yM · B · · sin x = 0
l l
 2
π  
EIω C + GIT · C + N̊ iP M 2 C + zM A − yM B = 0
l
 4  2
π π π π
2. EIy · · B · sin x + N̊ · (B − yM · C) · · sin x = 0
l l l l
 2
π
EIy · · B + N̊ · (B − yM · C) = 0
l
 2
π
3. EIz · · A + N̊ · (A + zM · C) = 0
l

Algebraic system of equations to calculate A, B and C:

 2
π
 
N̊ zM −N̊ yM EIω + GIT + N̊ iP M   A 
2
0
 
l

  2 
 π    
0 EIy + N̊ −N̊ yM  B  = 0
    

 l    
0
 2
π
 
 C
+ N̊ 0

EIz N̊ zM
l

Determinant of the coefficient matrix:

"  2 # "  2 #
π π
2
N̊ · zM 2
· EIy · + N̊ + N̊ · yM 2 2
· EIz · + N̊ −
l l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 128
 "  2 #"  2 #"  2 #
π π π
− EIz · + N̊ EIy · + N̊ EIω + GIT + N̊ · iP M 2
=0
l l l

Yields a cubic equation for the critical loads

3.5.3 Buckling and Lateral Torsional Buckling of Bars

of Thin–Walled Monosymmetric Cross–Section,
with a Concentrated Compressive Force, Acting Eccentrically on
the z–Axis

Cross–section: P
y
yM = 0 , z M by , b z = 0
v
First order analyis: ϑ
Mz
z Mx
N̊ < 0 P

pz = 0, M̊y = const. Q̊z = 0 x

Neighbouring
py = M̊z = Q̊y = 0 state
lateral–torsional
iP M 2 = iP S 2 + zM 2 buckling

mx = M̊x = 0 ẘ
N̊ Reference
state
M̊y
See diagrams 7.2 and 7.3 in: [Petersen
1982, chapter 7: Biegedrillknicken und
Kippen]

Governing equations and boundary conditions (supports):

Torsion:
h i
EIω ϑ′′′′ − GIT ϑ′′ − N̊ iP M 2 ϑ′′ + zM v ′′ + M̊y (v ′′ − by ϑ′′ ) = 0

BC.1: Mx = GIT ϑ′ − EIω ϑ′′′ + N̊ (iP M 2 ϑ′ + zM v ′ ) − M̊y (v ′ − by ϑ′ )

BC.2: Mω = −EIω ϑ′′

Bending in the x–z–plane

EIy w′′′′ − N̊ w′′ = 0 BC.1: My = −EIy w′′ + N̊ w

BC.2: Tz = −EIy w′′′ + N̊ w′

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 129
Remark: In general the deformed state is computed regarding first order theory. Thus the N̊ –
terms in the equation above as well as the respective boundary conditions can be neglected.

Bending in the x–y–plane:

EIz v ′′′′ − N̊ (v + zM ϑ)′′ + M̊y ϑ′′ = 0 BC.1: Mz = −EIz v ′′ + N̊ (v + zM ϑ) − M̊y ϑ

BC.2: Ty = −EIz v ′′′ + N̊ (v + zM ϑ)′

3.5.4 Lateral Buckling of Deep Girders

of Thin–Walled Monosymmetric Cross–Section
(f. i. steel work cross–sections)

pz v
ϑ
Mx
z Mz

x Neighbouring
state Additional assumptions:

ẘ
Q̊z
Reference
state
M̊y

Cross–section:
yM = 0 , zM by , b z = 0
First order analysis:

N̊ = 0 pz , M̊y , Q̊z ; ypM = 0 mx = M̊x = 0

py = M̊z = Q̊y = 0 iP M 2 = iP S 2 + zM 2

Governing equations and boundary conditions (supports):

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 130
 Torsion:

EIω ϑ′′′′ − GIT ϑ′′ + M̊y (v ′′ − by ϑ′′ ) + pz zpM ϑ = 0

BC.1: Mx = GIT ϑ′ − EIω ϑ′′′ + Q̊z v − M̊y (v ′ − by ϑ′ )

BC.2: Mω = −EIω ϑ′′

Bending in the x–z–plane

EIy w′′′′ − pz = 0 BC.1: My = −EIy w′′
BC.2: Tz = −EIy w′′′

Bending in the x–y–plane

EIz v ′′′′ + (M̊y ϑ)′′ = 0 BC.1: Mz = −EIz v ′′ − M̊y ϑ

BC.2: Ty = −EIz v ′′′ − (M̊y ϑ)′

See diagrams 7.2 and 7.3 in: [Petersen 1982, chapter 7: Biegedrillknicken und Kippen]

3.5.5 Lateral Buckling of Deep Girders

of Thin–Walled monosymmetric Cross–Section
being Stressed by M̊y = const.

In Section 3.4.8, we discussed the case of a yoke–supported girder with respect to a second
order analysis.

My My
M
S
y
kv
B
kϑ

z
z

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 131
Governing equations, see section 3.4.8.

Setting the determinant of eq. (3.137) equal to zero, and solving the quadratic equation
emerging, for M̊y , we receive (using the abbreviations, introduced in Section 3.4.8):

1
!
q
l2
M̊y1,2 = − 2 kv zBM + B by ± B (−4 kv zBM by + B by 2 + 4 TS + 4 TB ) 2
2 π

if we exclude an elastic continuous support, we can rewrite this formula, using cross–sectional
stiffness quantities:
s
1 by π 2 EIz π4 π4 π2
M̊y1,2 = ± EIz 2 by 2 + 4 EIz k ϑ + 4 EIz EIω + 4EIz GI T (3.157)
2 l2 l4 l4 l2

If we assume double symmetric cross–sections without a torsional spring bed we finally get:
s
π2 l2

M̊y,crit = 2 EIω + 2 GIT EIz
l π

Example 3.2: Continuing Example 3.1

Critical Moment of the girder

1. Computation out of the roots of the characteristic polynomial of the coeff-matrix of (3.137)

π2 π4 π2 π2
 
 −M̊y 2 14,949 4 + 9,908 2 + M̊y 0,22253 2 
A= 6 l l 6 
4 2
π π
 
1576,134 4
 
−M̊y 2
6 6

−0,2742 M̊y 3,8399 + M̊y 0,061

 

= 
118,4643 −0,2742 M̊y

det A = 0,07516 M̊y2 − 7,2272 M̊y − 454,891 = 0 M̊y,crit = 139,53 kNm

M̊y,crit,2 = −43,37 kNm

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 132
 2. Evaluating equation(3.157) with kv = 0 and kϑ = 0.
s
1 by π 2 EIz π4 π4 π2
M̊y1,2 = ± EIz 2 by 2 + 4 EIz k ϑ + 4 EIz EIω + 4EIz GI T
2 l2 l4 l4 l2

M
S
y

Rem.: The difference in the absolute value of the crit. moments results exclusively out of
π2
the expression M̊y by in the coeff-matrix A.
l2
For a double-symmetric cross section with (the same) stiffnesses EIω , EIz and GIT one
would obtain:

454,8905
s
M̊y,crit = ± = ±77,796 kNm
0,07516

3.5.6 Lateral–Torsional Buckling of Glued Laminated Beams of

Rectangular Cross–Section

Additional assumptions:

Cross–section
yM = zM = 0
by = b z = 0
EIω = 0

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 133
 First order analysis
N̊ = 0
pz , M̊y , Q̊z ypM = 0
py = M̊z = Q̊y = 0 zpM = 0
mx = M̊x = 0
K̊T σ = 0

Governing equations and boundary conditions (supports):

Torsion:
GIT ϑ′′ − M̊y v ′′ = 0
BC.1: Mx = GIT ϑ′ + Q̊z v − M̊y v ′
BC.2: not applicable

Bending in the x–z–plane:

EIy w′′′′ − pz = 0
BC.1: My = −EIy w′′
BC.2: Tz = −EIy w′′′

Bending in the x–y–plane:

EIz v ′′′′ + (M̊y ϑ)′′ = 0
BC.1: Mz = −EIz v ′′ − M̊y ϑ
BC.2: Ty = −EIz v ′′′ − (M̊y ϑ)′

Yoke–supported Beam loaded by M̊y = const.

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 134
 y

My

My
z

v
ϑ Neighbouring
Qy
Mz state
Mx

Reference
ẘ state
M̊y

Governing equation and internal forces and moments:

−GIT ϑ′′ + M̊y v ′′ = 0 BC.1: Mx = GIT ϑ′ − M̊y v ′

EIz v ′′′′ + M̊y ϑ′′ = 0 BC.1: Mz = −EIz v ′′ − M̊y ϑ

BC.2: Ty = −EIz v ′′′ − M̊y ϑ′

Boundary conditions:
position x = 0: position x = l:
ϑ(0) = 0 ϑ(l) = 0
v(0) = 0 v(l) = 0

Mz (0) = −EIz v ′′ (0) − M̊y ϑ(0) = 0 Mz (l) = −EIz v ′′ (l) − M̊y ϑ(l) = 0

Basic approach for the displacement functions:

x x
v = C1 sin π ϑ = C2 sin π
l l
π x π x
v = C1 cos π
′
ϑ = C2 cos π
′
l l l l
 2  2
π x π x
v ′′ = −C1 sin π ϑ′′ = −C2 sin π
l l l l
 3
π x
v ′′′ = −C1 cos π
l l
 4
π x
v ′′′′ = C1 sin π
l l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 135
All boundary conditions are satisfied.
Substitution of the trial functions into the differential equations yields:
 2  2
π x π π
− GIT C2 sin π + M̊y C1 sin π =0 (3.158)
l l l l
 4  2
π x π x
EIz C1 sin π − C2 sin π M̊y = 0
l l l l

In matrix notation:
 
M̊y −GIT  
C1 0
 

= 
 
 2 · 
π

−M̊y C2 0
 
EIz
l

Setting the determinant equal to zero results in:

 2
π πq
−M̊y2 + EIz GIT = 0 M̊y,crit = EIz GIT (3.159)
l l

Using (3.159) we get from (3.158)

√ s
πq EIz GIT π EIz
EIz GIT C1 − GIT C2 = 0 C2 = C1 π = C1
l l GIT l GIT

Thus we get for an arbitrary C1 :

s
x π EIz x
v = C1 sin π ϑ = C1 sin π
l l GIT l
 2 s
π x πq π EIz x
Mz = −EIz v ′′ − M̊y ϑ = C1 EIz sin π − C1 EIz GIT sin π
l l l l GIT l
 2  2 !
π π x
= C1 EIz − EIz sin π =0
l l l
 2 q
π x πq π x
Mx = GIT ϑ′ − M̊y v ′ = C1 EIz GIT cos π − C1 EIz GIT cos π
l l l l l
 2 q
π x
= C1 EIz GIT (1 − 1) cos π =0
l l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 136
If we adopt any C1 , we get internal moments in a neighbouring state.
As M̊x = N̊ = Q̊y = 0 and with M̊y = const. → Q̊z = M̊y′ = 0 as well as by = bz = 0 follows
from (3.119) and (3.100):
 2
π x
M̃z = −EIz v = M̊y ϑ = C1
′′
EIz sin π
l l
πq π x
M̃x = GIT ϑ′ = M̊y v ′ = C1 EIz GIT cos π
l l l

Stability of Structures 3 Second Order Analysis and Stability of Space Frames

Winter Term 2022/2023 137
Bibliography
[Kiener 2004] Kiener, Gunter: Stability of Structures, Lecture Notes. 2004

[Petersen 1982] Petersen, Ch.: Statik und Stabilität der Baukonstruktionen. 2. Vieweg
Verlagsgesellschaft, 1982

[Wunderlich and Kiener 2004] Wunderlich, W. ; Kiener, G.: Statik der Stabtragwerke.
Vieweg+Teubner, 2004

Stability of Structures
Winter Term 2022/2023
138