"7

Guei

FINITE

ELEMENT STABILITY

OF

STEEL STRUCTURES

THIN-WALLED

A Thesis
the

ANALYSIS

for

submitted
Degree

of

DOCTOR OF PHILOSOPHY
by
MOHAMED TAHER MOHAMEDNEMIR

Department

of

University

Civil
of

Engineering
Salford

ME"" OF
DD-f-ft,
U L::.,
,
%..

r, I -,

t"..

!-).

DI
,:
r
-1
lr*,,
J, k a"IZ,

September

1985

CONTENTS
PAGE No.
DECLARATION

ABSTRACT

ii

ACKNOWLEDGEMENTS

iv

LIST

OF SYMBOLS,

LIST

OF FIGURES AND TABLES

vii

LIST

OF PLATES

xvi

PART (I)
1

CHAPTER 1.

INTRODUCTION

CHAPTER 2.

TORSIONAL-FLEXURAL

BUCKLING OF THIN-

WALLED STRUCTURES
2.1.

Uniform

2.2.

Combined
Prismatic
2.2.1.
2.2.2.
2.2.3.

2.2-4.
2.2-5.
2.3.

Methods

and Nonuniform

4
4

Torsion
Behaviour

Torsional-Fiexural
Members

of

5
5

Basic

assumptions
Torsional-flexural
behaviour
prismatic
member
_
Sectorial
of the
properties
2.2.3.1.

Sectorial

2.2.3.2.

Sectorial

2.2.3.3.

Second sectorial
(warping
constant)

of

Analysing

2.3.1.

General

2.3.2.

Equilibrium

Elastic

(w)
of

area

moment

of

area

Stability

7
7

moment

First
equations
order
equilibrium
bending
and torsion
combined
Basic
theory
of torsional-flexural
buckling
of

section

cross

co-ordinate
static

of

7
8
8
9

Problems

10
10

methods

10

PAGE No.
(closed

2.3.2.1.

Exact

2.3.2.2.

Approximate
differential
a.
b.

form)
solutions
equations

11

solutions
of the

13

Infinite

13

series
solution
Iterative
integration
method
difference
C. Finite
solution
d. Finite
integral
solution
2.3.3.

2.3.4.

2.4.

Energy

2.4.2.
2.4.3.
2.4-4.
CHAPTER

3.

14
17
19

methods

2.3.3.1.

General

2.3-3.2.

The Ritz

19
20

method

Numerical
techniques
displacement
method

based

on the

21
21

2.3-4.1.

General

2.3-4.2.

Member

stiffness-matrix

2.3-4-3.

Finite

element

Literature
Review
Torsional-Flexural
2.4-1.

14

method

25

method

Studies
of Previous
Buckling
Problems

of

28
28

Single

span elements
beams
Continuous
Three-dimensional
and space
plane
The contribution
FINITE

buckling
frames

BUCKLING

of

analysis

made by this

THREE-DIMENSIONAL
THIN-WALLED

30

FORMULATION

ELEMENT

22

34
38

study

OF THE ELASTIC
BEHAVIOUR

OF
40

SYSTEMS

3.1.

General

40

3.2.

Bimoment

40

3.3.

Strain

Energy

42

3.3-1.
3.3.2.

Strain
Strain

3.4.

The Potential

3.5.

Potential
Loads

3.6.

Derivation

energy
energy

due to normal stresses

due to shear stresses

of the Applied

Energy

43

Load

in Terms of the External

of the Element

42

Matrices

43
Joint
47
48

PAGE No.
3.7.

Evaluation
Geometric

of the
Matrix

3.8.

Stiffness

Matrix

53

3.9.

Geometric

Matrix

55

3.10.

Transformation

CHAPTER 4.

The Buckling

4.2.

Solution

4.2.2.
4.2.3.
4.3.

the

61

OF THE BUCKLING LOAD

67
67

Stability

Southwell

68

Equation

68
buckling
load
curve

70
72
IN THIN-WALLED

Calculating

for

the

5.2.1.2.

Simplified

Continuous

beams

83

5.2.2.2.

Analogy
bending

5.3-3.

Test

81

solutions

Bimoment-distribution

Test

80

solutions

5.2.2.1.

5.3.2.

80
80

Closed-form

Object

Bimoments

beams

5.2.1.1.

5.3-1.

MEMBERS 78
78

Single-span

Experimental
to Combined

70

Program

used

Finite

from

method

BIMOMENT DISTRIBUTION

Methods

5.2-3.

Axes

solution
Prediction
of the
the load-displacement

5.2.

5.2.2.

the

Eigenvalue

Introduction

5.2.1.

in

Criterion

of

5.1.

5.3.

of

The Computer

CHAPTER 5.

Terms

51

PREDICTION

4.1.

4.2-1.

Bimoment

element
Study
Bending

method

with
second-order
technique

84
85

method

of Cold-Formed
and Torsion

83

Z-Beams

Subjected
86
86

program

87
87

5.3.4.

rig
Instrumentation

88

5.3-5.

Test

88

results

PAGE No.
CHAPTER 6.

OF THE FINITE

APPLICATION

ELEMENT METHOD
91

TO BUCKLING PROBLEMS

91

6.1.

General

6.2.

Conventional
Pure

6.2.2.

Lateral
buckling
of
bending
uniform
buckling
Lateral
of
concentrated
central
Lateral
buckling
of
load
concentrated

6.2-4.

6.4.

Lateral

6.3.1.

Elastically

6.3.2.

Interaction

6.4.2.
6.4.3.
6.4.4.
6-4.5.
6.. 5.

6.5.2.

a simple
load

beam by

92
92

a cantilever
at the free

of Continuous

restrained
buckling

beam by
end

93

Beams

94
94

beams

Beams

103

The effect
of monosymmetry
beams by uniform
Simply supported
moment
beams under central
Simply supported
load
concentrated
loaded with transverse
Cantilevers
free
the
load
end
at
concentrated
Comparison between the finite
element
results
and experimental
solutions

103

Buckling

of continuous

of Monosymmetric

Buckling

of Plane

Frames

buckling
Torsional-flexural
of elastically
frames
plane
restrained
narrow rectangular
buckling
Interaction
of doubly symmetric
I-portal
frames

6.6.

Three-Dimensional
Frames

6.7.

Summary and Conclusions

CHAPTER 7.

beam by

98

Torsional-Flexural
6.5-1.

Buckling

a simple

beams

Elastic
Lateral
and cantilevers
6.4.1.

91

buckling

torsional

Elastic

91

Problems

6.2.1.

6.2.3.

6.3.

Stability

Buckling

Analysis

of Space

SECOND-ORDERTORSIONAL-FLEXURAL BEHAVIOUR
OF Z-BEAMS

7.1.

Introduction

7.2.

Test

Program

106
107
107
107
109
log
112
113
115

118
118
119

.i

PAGE No.,
Testing

7.4.

Discussion

7.5.

Conclusions

CHAPTER 8.

119

Procedure

7.3.

of

120

the, Results

122

CONCLUSION

8.1.

Principal

Conclusions

8.2.

Suggested

Further

124
126

Work

(II)

PART

CHAPTER 9.

DIAPHRAGM-ACTION

IN TRUNCATED PYRAMID
128

STRUCTURES WITH FOLDED ROOFS

9.1.

Introduction

9.2.

Stressed

128
Skin

Diaphragm

130

9.2.1.

General

9.2.2.

Historical

9.2-3.

Diaphragm

9.2-4.

Diaphragm

9.3.

Loading
Structure

9.4.

action

132

arrangements

133

and

components

9.2-4.2.

arrangements
Components
of a diaphragm

9.2-4-3.

Failure

Structural
structures
Tests

on the

9.3.3.

The two-bay

9.3.4.

Loading

9.4.1.

General

9.4.2.

Finite

133

of

MACE Unit

133

hipped

roof

and the

134

Pyradome
135
135

tests

135

structure
two-bay

137

loading
Pyradome

tests
of

panel

modes

behaviour

9.3.2.

The Behaviour

133

Basic

The MACE unit

The MACE unit

9.3.1.

130

background

9.2-4.1.

9.2-5.

130

Action

the

on the
Plane

Trapezoidal

Pyradome

137

Panel

138
138

element

modeling

139

PAGE No.
9.4.2.1.

Design
of
elements
Orthotropic

9.4.2.2.
9.4.2-3.
9.4.2-4.
9.4.2-5.

9.4-3.

Plane

9.4,. 3.3.
9.4-3-4.
9.4-3-5.

for

9.4.5.

Comparison between
three models

9.5.2.

of

truss

the

9.5.2.2.

9.5-3.

used
and

in

141

seam fasteners

142

142
143
143
143
144
144
144

the

results

and

the

of the

145

Two-Bay
148
148

the

analysis
discussion
of

the

results

deflections
The MACE unit
forces
Critical
fastener
calculated
from the model of the MACE unit

The two-bay

141

to

analysis

MACE Unit

The model
Comparison
9.5.2.1.

sheet

simulation

Simple

9.5.1.

9.6.

Spring

9.4-4.

Analysis
Pyradome

elements

141
for

elements

140

Beam elements
Diagonal
truss members
the sheeting
representing
Vertical
truss
members
Prismatic
member representing
the seam fasteners
Spring elements for sheet to
frame fasteners

9.4-3.2.

9.5.

plate

triangular

Beam elements
Spring
elements
frame fasteners

frame

9.4-3-1.

rectangular

Pyradome deflections

149
149
151

153
153

Conclusions
APPENDIX A. 2.1.

155

APPENDIX A-3-1.

157

APPENDIX A-3.2.

159

APPENDIX A-4-1.

160

APPENDIX A. 4.2.

166

APPENDIX A-4-3.

171

APPENDIX A. 5.1.

188-

APPENDIX A-5.2.

189

APPENDIX A. 9.1.

190

APPENDIX A. 9.2.
REFERENCES (PART I)

192

REFERENCES

(PART 11)

193
202

DECLARATION

None

in this
thesis
of the material
contained
been submitted
in support
for another
of an application
degree
or qualification
of this
or any other
university
institution
of learning.

MOHAMED TAHER
September

1985

has

or

ii

ABSTRACT

Recent
members

have

systems

built

explicit
different

entirely

analysis

of

phenomena
In particular,

loading.
important
since

the

close

upper

The first
Chaptersl-8)

has

to

part
of this
been devoted

scale

may be prone

to

during

phenomena

is

an

of

such
often

occurs

carrying

of

capacity
two-part

structures
a
provides
the structure.
(Part

thesis

to

of

methods
general
buckling
of thin-walled

of the torsional-flexural
A review
investigations
of previous
is presented.
A general
of solution

and the
finite

also

includes

the

effect

of

I,
analysis

structures.

available

methods
formulation

element
thin-walled
structures

buckling
of the torsional-flexural
of
has been derived.
The resulting
geometric
elastic
be used to analyse
structures
monosymmetrical
with
It

steel

An
members.
by the
complicated

is

design

buckling

the

large

buckling

elastic
in the

gauge

steel

cold-formed

such structures
that
the structure

consideration
load at which
bound

use of
developing

with

from

light

the

in

applications
been concerned

matrix

can

members.
for
osymmetry

sectorial-mon

A
transcross-sections
general
without
symmetry.
any axis
of
formation
for the applicato allow
has been developed
matrix
tion
of the finite
element
method to the three-dimensional
The
frames.
elastic
and
portal
stability
analysis
of space
formulation,
finite
the
validity
element
and accuracy
of
new
have

been

lateral
solutions

by analysing
checked
buckling
for
problems
by other
techniques

An experimental
supported
program
element
torsion.
buckling

a number
which
are

of

exact

different
or

highly

accurate

available.

out
was carried
The first
z-beams.

program

elastic

on simply

part
of this
steel
the
the
finite
to
validity
of
check
undertaken
was
caused by nonuniform
of the bimoments
calculations
to elastic
lateral
The second part
was devoted
bending
combined
and torsion.
of z-beams under
cold-formed

iii

this

The
deals

of
second part
the analysis
of hipped

with

corrugated
has been
elastic
hipped

steel

suggested.
linear
analysis

roof

compared
structures.

roof

with

structures.
previous

sheeting.
The model

thesis

(Part

II,

Chapter

9)

roof

structures
with
A simple
theoretical
model
has been used to perform
an

of the behaviour
The theoretical
experimental

of

two

results
for
results

types

of

are
these

two

the

iv

ACKNOWLEDGMENTS

I
Professor
gave
E. R.

would like
J. M. Davies,

valuable
advice
Bryan,
for his

throughout

the
I

Laboratory
Mr.

S.

to

Mrs.

also
staff

Abed

my gratitude
express
me to
who introduced
throughout
the project,
continued

period
wish
and

Elhafiz

L. C. Rycroft

to

to

of

interest

my thanks
to Mr.

express

in

particular
his preparation

for

the

and
subject
and to Professor

encouragement

study.

to
for

and

my supervisor,

typing

the

of

to

the

Structural

W. Deakin,
the

manuscript.

graphs,

and

v
LIST

OF MAIN

Static

moment

of

area

about

y-axis

Static

moment

of

area

about

z-axis

Second

moment

of

area

about

y-axis

Second

moment

of

area

about

z-axis

S
I

z
y

SYMBOLS

z
Io

Polar

Area

Z5
m
SW

Sectorial

co-ordinate

Sectorial

static

IW

Second

Torsion

constant

Modulus

of

Shear

Displacement

in

Shear

center

displacement

in

y-direction

Shear

center

displacement

in

z-direction

6x

Angle

of

twist

4)

Angle

of

rotation

about

z-axis

Angle

of

rotation

about

y-axis

moment
of

cross

Local

X, Y, Z

Global

ml, n 1

m
n2
2' 2'

moment

of

the

area
area

cross

(warping

constant)

section

of

the

cross

cross

the

the

the

system
of
axes X,

cosines

cosines
axes

global

cosines
axes

global

Member

elastic

KG

Member

geometric

Span

the

of

system

global

stiffness
matrix

member

section

section

co-ordinate

KE

of

the

co-ordinate

Direction
to

x-direction

of

Direction
to

k 3, m3 n3

of

elasticity

Direction
to

moment

of

point

at

modulus

x, y, z

tl,

section

sectorial

Warping

inertia

of

of
the

X,

X,

local

x-axis

with

respect

local

y-axis

with

respect

Y and Z respectively
the

of

axes

Y and Z respectively
the

of

axes

local

z-axis

Y and Z respectively

matrix

with

respect

vi

t.

transformation

Member

(warping

matrix

Bimoment

My

Bending

moment about

y-axis

Mz

Bending

moment

z-axis

Qy

Shear

force

in

y-direction

QZ

Shear

force

in

x-direction

Px

Compressive

axial

0y

Coefficient

of

monosymmetry

about

y-axis

Oz

Coefficient

of

monosymmetry

about

z-axis

OW

Coefficient

of

sectorial

Bending

moment)

about

twisting
-

force

monosymmetry

parameter

vii

LIST

(Figures

and

OF FIGURES

tables

are

AND TABLES

at

presented

the

end

of

the

relevant

chapter).
CHAPTER

Fig.

2.1.

Warping

of

2.2.

Warping

of

2.3.
2.4.

The prismatic
Torsional-flexural

2.5.

Normal

Sectorial

2.7.

Distribution
for

I-beam

symmetrical
beam
a cantilever
member
displacement

tangential

and
displacement

2.6.

at

components

point

of

point
of the

co-ordinates
of

the

sectorial

co-ordinates

Z-section

2. -8.

Biaxially

2.9.

Finite

2.10.

a doubly

loaded
difference

beam-column
for

approximation

the

angle

of twist e

Approximate

interaction

curve

of the lateral
beams (Ref.
50).

calculations
continuous

for

rapid
load
buckling

of

CHAPTER
Fig.
3.1.

Analysis

of

3.2.

Element

end

3.3.

Bending

My

3.4.

Bending

Mz

3.5.

Direction

3.6.

Misalignment
joint

the

eccentric
forces

cosines

positions

of

between

Px

axial

load

local

and global

member ends

from

specified

axes

viii

CHAPTER
Fig.
4.1.

Standard

Southwell

plot

4.2.

Modified

Southwell

plot

4.3.

The

4.4.

Flow

4.5.
4.6.
4.7.
4.8.

inner

iteration

procedure

diagram

of the computer
program
Flow chart
to find
of the computer
operations
the value
load factor
of A
at a given
cr
Flow chart
from the
of the prediction
of X
cr
Southwell
modified
plot
Joints
connecting
part
of the structure
Imaginary
joint
connections
after
eliminating

CHAPTER
Fig.
5.1.

Correction

5.2.

subjected
Correction

5.3.

simply

to

beams

supported
torque

(81).

The distribution
by Walker

5.5.

F for

central
concentrated
factors
F for fixed
end beam
to central
subjected
torque
concentrated
The three-span
by
conti: auous beam solved
Walker

5.4.

factors

Bimoment

(81)

and

diagram

of analysis,
(82),
bution
bending

of

(84)

by the

c)

a)

calculated

as

finite

calculated

namely,
b) the
and

bimoment

the

the

method
element
by three
methods
distri-

bimoment

with
second-order
analogy
the finite
element.

Table
5.1.

Bimoment

values

of

the

beam shown

in

5.5.

fig.

Fig.
5.6.

The

5.7.

Positions

tested

beam and its

of the applied

cross-section
loads
in the

three

tests
5.8.
5.9.

frame
The end support's
ILI brackets
The light
gauge steel
the torsional
and warping
simulate

used

to

conditions

ix

5.10.

Dimensions

5.11.

Positions

the

of
of

plate
for
sections

reinforcement

the

strain

gauges

1-1

5.13.

and 2-2.
Maximum compressive,
Longitudinal.
strains

5.14.

under static
Longitudinal

5.12.

under
CHAPTER

torsional

flange
1-1

section
at cross
strains
load = 40.0 kg/hanger.

2-2

buckling

6.1.

Distribution

6.2.

Modified
applied
Lateral

ex
of the angle of twist
(8
Southwell
plot
elements-load
=P cr)

buckling

6.3.

Distribution

6.4.

Modified
applied

6.6.

top

Pure

...
6.5.

the

section
at cross
= 40.0 kg/hanger

load

static

in

strain

Lateral

by uniform
of the

Southwell

bending

lateral
plot

displacement

(8 elements-load

= Pcr)

buckling

by concentrated

load

Distribution
Modified
applied
Lateral

displacement
of the lateral
(8
Southwell
elements-load
plot
=P cr)

buckling

6.7.

Distribution

6.8.

Modified

6.9.

applied
Details
Hartmann

of

of
Southwell

displacement

lateral
plot

(8 elements-load

=P

cr)
of the
(54)

by

beams studied

continuous

Table
6.1.

load

by concentrated

a cantilever
the

Critical
continuous
the middle

load

(a)

parameter
beams - one span
(case a)

for
is

two-span
loaded

at

Fig.
6.1o.

Convergence
for

6.11.

6.12.

kZ=

the

of

finite

Case

2.0.

of

solutions
element
two-span
continuous
(case a)
the middle

at
one span is loaded
Effect
flexural
of strong-axis
bracing
lateral
on the critical
span beam - one span is loaded
(case a)
Effect

of
a two-

stiffness
load of
at

flexural
strong-axis
lateral
bracing
on the critical
span beams - the two spans are
of

the

beam

middle

of
stiffness
load of two(case b)
loaded

Table
6.2.

Critical

(a)

load-parameter

beams - the central

(case
middle
c).

continuous
at

the

three-span

for

span

is

loaded

Fig.
6.13.

Two-span

continuous

6.14.

Two-span

continuous

beam with
beam with

one
the

span loaded
two spans

loaded
6.15.
6.16.

beam
of a two-span
parameters
(Ref.
50)
5.0
with k 2/k,
=
The continuous
beam A. 20.20. '20.2
presented

Critical

in

ref.

load

51

Table

6.3.

Comparison

the

between

critical
experimental
loads and the finite
element results
(4 elements/span)
for beam A. 20.20.20.2

Fig.
6.17.

Convergence

6.18.

of test'7
Critical

...

of
(Ref.
loads

finite

the

element

solutions

51)
of

beam A. 20.20.20.2

(Ref-51)

Beam A. 20.20.20.2.

6.19.

Distribution

of

the

lateral

6.20.

Distribution

of

the

angle

displacement
of

twist

6x

xi

6.21.

(load
The modified
Southwell
plot
)
(4
for
test
1
=P cr
elements/span)
Monosymmetric
I-cross
section

6.22.

applied

Table

6-4.

Finite

element

solutions

supported

6-5.

simply
Finite

beams under
for
solutions

6.6.

simply
Finite

element

for

monosymmetric
uniform

monosymmetric
load
central

beams under
for
solutions

supported

moment
P

element
loaded
cantilevers
with
Finite
element
solutions

6-7.

cantilevers

monosymmetric
P at the free
end
y
of t he monosymmetric
by Anderso n and_Trahair(40)

tested

Fig.
6.23.

Effect

of axial
stiffness
on the buckling
load of
(comparison
between
the

the

of

frames

portal
finite

(comparison

between

6.25.

55)
ref.
Interaction

6.26.

Interaction

6.27.

The dimensions
studied

buckling
buckling

the

of
of

finite

for

frames
and

element

frame

and loading
by Razzaq and Naim

knee

the
portal

frame

for

and

element

55)
ref.
Effect
of torsional
stiffness
bracing
load
on the buckling

6.24.

bracing

knee

system
(11)

1 of
1 of
of

ref.

60

ref.
the

60
frames

Table
6.8.

Finite

6.9.

studied
Convergence
for

for the
element
solutions
by Razzaq and Naim (11)

the

the

of

space

frames

cases

of

the

the

dial

finite

element

loaded

with

tested

beams

loading.
CHAPTER

Fig.
7.1.

Loading

7.2.

Locations

of

gauges

space

frames

solutions
case 'at

of

xii

7.3.

Test

7.4.

Test

7.5.

Test

7.6.

Test

B-1, angle
of
B-2, vertical

7.7.

Test

B-2, horizontal

7.8.

Test

7.9.

Test

B-2, angle
of
B-3, vertical

7.10.

Test

B-3, horizontal

7.11.

Test

7.12.

Test

B-3, angle
of
B-4, vertical

7.13.

Test

B-4, horizontal

7.14.

Test

7.15.

Test

B-4, angle
of
B-5, vertical

7.16.

Test

B-5, horizontal

7.17.

Test

B-5, angle

B-1, vertical
B-1, horizontal

of

deflection

at

mid-span

movement at mid-span
twist
at mid-span
deflection

mid-span

at

movement at mid-span
twist
at mid-span
deflection

at

mid-span

movement at mid-span
twist
at mid-span
deflection

mid-span

at

movement at mid-span
twist
at mid-span
deflection

mid-span

at

movement at mid-span
twist
at mid-span

Table
7.1.

Finit

solutions
e element
in comparison
loads
with
failu re loads

of

the

buckling

experimental

CHAPTER
Fig.
9.1.

Diaphragm

9.2.

f rames under vertical

in
Diaphragm
action

9.3.

under sway loading

Diaphragm
in
action

9.4.

Basic

9.5.

Basic

9.6.

Profile

9.7.

chapter)
Finite
element
diaphragm

9.8.

Full

9.9.

Finite

9.10.

Details

9.11.

Details

action

in

roof

pitched
loads
flat

roof

a folded

portal
frames

portal
plate

roof

diaphragm

arrangements
structure
of MACE type
(presented
dimensions

frame

model

for

simulation

element
simulation
B
of joint
of

joint

of

30 unit
within

the

trapezoidal

the

diaphragm

of

fastener

the

xiii

9.12.

Simplified
the

9.13.

Wind load

case.
bottom flange

Wind

load

case.
bottom flange

the
9.15.

Wind

load

the

9.16.
9.17.

top

9.23.

9.24.

In-plane
(corner

joints

are

displacements
joints

are
displacements

of
hinged)
of
fixed)
of

10.0 kN each,
loads,
at
vertical
displacements
In-plane
of the
(corner
flange
joints
are fixed)

flange

Case of two
apex joints.
flange

Case of
joint.

10.0 kN each,
loads,
at
vertical
In-plane
displacements
of the
(corner
joints
are hinged)
10.0 kN each,
loads,
at
vertical
In-plane,
displacements
of the
(corner
joints,
are fixed)

10.0 kN atapex
load
one vertical
displacements
In-plane
of the bottom
(corner
joints
are hinged)

10.0 kN at
load
one vertical
displacements
In-plane
of the
(corner
flange
joints
are-fixed)
,10.0 kN at
load
Case of one vertical
joint.
displacements
In-plane
of the
(corner
flange
joints
are hinged)
Case of
joint.

Case

of

joint.
flange
9.25.

(corner

displacements

10.0 kN each,
loads,
at
vertical
displacements
In-plane
of the
(corner
flange
joints
are hinged)

flange
9.22.

In-plane

In-plane
case.
(corner
flange
joints

Case of two
apex joints.

top
9.21.

of

Case of two
apex joints.

top
9.20.

analysis

Case of two
apex joints.

bottom
9.19.

the

are hinged)
Wind load case.
displacements
In-plane
of
(corner
the top flange
joints
are fixed)

bottom
9.18.

for

diaphragm

the
9.14.

truss

Case

of
In-plane
(corner

load.

10.0

kN at
one vertical
displacements
In-plane
of the
(corner
joints
are fixed)

apex
bottom

apex
top

apex
top

a side load 10.0 kN at apex joint.

displacements
flange
of the bottom
joints
are hinged)

xiv

9.26.

9.27.

9.28.

9.29.

Case of a side load 10.0

In-plane
displacement
of
(corner j ints are fixed)
Case of a side load 10.0
In-plane
displacements
of
(corner joints
are hinged)

M at apex joint.
the bottom flange
M at apex joint.
the top flange

Case of a side load 10.0 kN at apex joint.

In-plane
displacements
of the top flange
(corner joints
are fixed)
Fastener forces due to wind load (hinged
corners model)

Table

9.1.

Critical
fixed

for

forces

fastener

hinged

and

models

Fig.
9.30.
9.31.

The model used to analyse the MACE unit

Cases of loading
on the MACE structure

Table
9.2.

Experimental

9.3.

under U. D. L.
Experimental

and theoretical
2
0.81 kN/m

displacements

9.4.

displacements
and theoretical
under wind load cases
Comparison
between experimental
and
results
the results
model
of the X-diagonals
2)
O. D. L. 0.81 kN/m

9.5.

Comparison
the
load

between

results

of

the

results

experimental
X-diagonals

model

and
(wind

cases)

Fig.
9.32.
9.33.
9.34.

Comparison

and theoretical
experimental
load
displacements
in the case of asymmetric
(in kN) due to U. D. L. of
Fastener
forces
0.81 kN/m2
Fastener
of

1.215

between

forces
kN/m2

(in

kN)

due

to

asymmetric

load

xv

9.35.

Fastener

forces

(in

kN)

due

to

wind

load

9.36.

case Ic'
Fastener

forces

(in

kN) due

to

wind

load

9.37.
9.38.
9.39.

case Id'
Details
of the
The model used
Pyradome

two-bay
to

Pyradome

analyse

Vertical

loads

Vertical
load for

deflections
the two-bay

on the

the

structure
two-bay

Pyradome

Table

9.6.

U.
D.
L.
and asymmetric
Pyradome

xvi

LIST

OF PHOTOGRAPHIC

PLATES

CHAPTER
5.1.

Restraining

5.2.

Strain

the

support
1-1

measurements

during

arrangements

at

5.3.

gauges at cross-section
Loading
3
of test

5.4.

Longitudinal

CHAPTER

7.1.
7.2.
7.3.
7.4.

strain

Beam B-2

at the moment of
Web and flange
failure
of
Beam B-3 at the moment of
Web and flange
failure
of

failure
beam B-2
failure
beam B-3.

test

PART
-------------

(I)

1.

CHAPTER ONE

INTRODUCTION
The conventional
structures
carrying
the assumption
that

static

forces.
and external
due
of the stiffness
However,
the

the

with

analysis
loads

linearly

of
is

under
performed
between
internal
exists

always

stable
This

equilibrium
includes
the reduction
neglecting
the change
of the structure-geometry.

to

increase

in

the

of

value

may be prone to the effect

lity
A condition
of instability
phenomena.
to lose
its
structure
starts
stiffness.
by the fact
that
the deformations
of the
ding

structure

a given

arbitrarily

structure
commences, this

remains
type of

entire

The load

level

buckling

load

the

or
the

In

this

at

stage
elastic

fundamental
long

that

a thin-walled
by flexure
in the
under

perfectly
buckling

plane

of

the

load

static

one of

the

stabiwhen the

exists
This is characterised

structure
corresponinfinite
for
values

load factor
can reach
(infinitesimal)
load
small

to

elastic

increments.

If
buckling

until

elastic

the

buckling.
called
elastic
is defined
as the elastic
load.
critical
is

case

column
of the

of

with
least

buckling
open

it

cross

rigidity.

is

assumed
buckles
section
However,

uniform

axial
compression
cross
a column with
cruciform
torsionally
longitudinal
its
sectional
shape buckles
while
takes
In general,
buckling
axis
straight.
of columns
remains
in a combined
torsional
and flexural
mode. -The in-plane
place
displacements
lation

of
the

of

the

shear

cross
center

A beam bent

be analysed
and a rotation
about it.
section

can

as a trans-

flexural
of the greatest
plane
The
in a similar
laterally
rigidity
manner.
may buckle
flexural-torsional
load may represent
the ultimate
buckling
beams.
strength
of thin-walled
unbraced
For
members

in

more
building

systems
such
frameworks.

in

the

the
25 years
has been almost

than

common use
limited
to

and the
purlins
as the roof
The use of cold-formed
steel

of

cold-formed

secondary
side beams of. steel
members for primary

2.

has been comparatively

such as portal
and space frames
This may be due to the lack
rare.
of information
concerning
the behaviour
of such structures
and the need for theoretical
techniques
to analyse
the different
may occur
phenomena that
during
loading.
have been
However,
applications
recent
systems,

concerned
entirely

developing

with
of

cold-formed

is

of
based

often

and

portal

space

frames

built

members.

The design
members

steel

frames built
of hot-rolled
plane
behaviour
alone.
on the in-plane

to be valid,
such consideration
to the out of plane displacements

For

the

resistance
of the
must be sufficiently

frame
high.

Light

however,
have a high tendency
frames,
gauge steel
portal
to twist,
laterally
in-plane
loading.
under
warp, and buckle
buckling
Torsional-flexural
is an important
in
consideration
the design
of such frames.
Although

the

torsional-flexural

buckling

beams has been extensively

span and continuous
has been reported
about the torsional-flexural
Most of the studies
and space frames.
plane
this

field

loading

limited

were

out

carried

shapes

and

in

special

element
method is
to be used for the

technique

powerful

complex
and irregular
15 years,
the method
the

frame

certain

single
little
studied,
buckling
of

conditions.
The finite

to

to

of

deal

torsional-flexural

with

application

single

span

structural
has been

the

of

or

well
recognised
linear
analysis
the
During
systems.

as a
of
last

by many investigators
extended
However,
buckling
problems.
to
been limited
has generally

method
beams.
continuous

thesis
in the first
of this
part
reported
(Part
to establish
I) was undertaken
in order
element
a finite
formulation
buckling
for the torsional-flexural
of thinbeams, columns
The new formulation
walled
was
and frames.
The study

aimed

to

be applicable

to

The first
chapters.
chapter.

part
An introduction

Chapter

two

any

of

this

thesis

shape.

sectional
(Part

I)

in the
presented
to three
devoted
main
is

is

cross

contains
present
items,

eight
namely,

3.

a)

Review

of

flexural
b)

Methods

theory

the

of
thin-walled

general
behaviour
of

used

to

torsional-

structures.

torsional-flexural

the

analyse

the

buckling.
c)

Review

of

flexural

The derivation

on the

studies

previous
buckling.

torsional-

the

elastic
new finite
element
in chapter
is presented
matrices
stiffness
and geometric
(1)
The derivation
is based on Vlasov's
of the
concept
behaviour
torsional-flexural
of thin-walled
structures.
of

a new transformation
also includes
matrix
chapter
buckling
three-dimensional
analysis
of plane
and
four

Chapter
techniques

used

instability
the

problems
solutions

five

is

a review
critical
includes

also

computer

element
Chapter

structure.

It

equation.

finite

bimoments

to

presents
the
predict

program

devoted

to

the

The

the
frames.

space

of the different
load from the
the

used

for

three.

elastic

illustration

in

this

of

study.

of the
a thin-walled

calculations

torsion
of
caused by the nonuniform
The finite
for a number
element
solutions
form,
highly
accurate
are compared to closed
determined
bimoments.
and to experimentally

of

the results
of a theoretical
six presents
the
element
method to examine
study made by the finite
Finite
element
of the new formulation.
validity
and accuracy
for a number of previously
are
presented
problems
solutions
Chapter

given

the

with

comparison

other

with

of

solutions

these

problems.
buckling
the
lateral
of
with
seven
free.
Z-beams
end
warping
with
simply
supported
cold-formed
these
five
to
test
An experimental
of
out
carried
program
was
loading.
torsional
beams under different
bending
types
and
of
loads
The measured values
displacements
the
and
critical
of
Chapter

are

compared
The

study

are

to

deals

the

finite

corresponding

observations
in chapter
given

and

conclusions

eight.

element
of

the

solutions.
present

4.

CHAPTER

TWO

To rs ional

-Flexural

2.1.

UNIFORM AND NONUNIFORM TORSION

Buckling

A thin-walled
when it

is

twisted

Thin-Walled

of

member exhibits
by uniform
torque

Structures,

displacements
warping
if the flanges
at the

Under
no longitudinal
restraint.
is the same for all
such conditions
and
cross
sections
warping
the only stresses
stresses
at each
are the shearing
produced
The warping
cross
section
of the member.
of the cross
section
end

of

cross

a twisted

I-beam

while

If

shown in

some longitudinal

flanges

any

the

of

in

is

at
length

up a curvature
2.2,
fig.
where

trated

in

the

flanges

torque

is

to
applied
torque
varies

then-be
will
direction.

longitudinal

a cantilever
T applied
at the

twisting,

only the web remains

directions.
two opposite

restraint
or if the

cross
section,
the member, the

During

2.1.

fig.

do not remain plane,

sections
the flanges
bodily
in
rotate

plane

take

have

sections

beam is

twisted

plane

the

along
to
forced
As shown

by a concenthe
curvature-of

end, the
flanges
the member and the flanges
varies
along
appear
as
being under two equal, but opposite
bending
in
moments, acting
The combination
their
of the two bending
moments
own plane'.
torque
induced
in the flanges
as a result
of the nonuniform
free

The longitudinal
called
stresses
caused by
a bimoment.
in the
the bimoment
and must be considered
can be very large
Tc, an be
the acting
torque
At any cross
analysis.
section
is

divided

into

two

parts:

a)

T.

due

to

St-venant

b)

Tw due

to

the

The present

shear

stresses

normal

chapter

stresses,

contains

and

induced
three

by the

main

parts,

namely:
Review
behaviour

of

the
of

general
thin-walled

theory

of

torsional-flexural

structures.

bimoment.

5.

2.

which can be used to analyse

buckling
Of thin-walled

The different
the

methods
torsional-flexural
and the

structures
Review
the

of

validity

the

each

out to
carried
of thin-walled

studies
buckling

of
previous
torsional-flexural

method.
analyse

structures.
2.2.

Basic

assumptions

The basic
behaviour

flexural
by Vlasov

(1)

assumptions
of thin-walled

the

of

theory

torsionalas given

members

are:

The material

b)

Small

c)

The member retains

its
cross-sectional
deformations
during
all
undergoing

d)

of

prismatic

a)

warp

MEMBERS

BEHAVIOUR OF PRISMATIC

COMBINED TORSIONAL-FLEXURAL

of

deflection

the

theory

perpendicular

The shear

is

structure

to

deformations

is

the
of

elastic.

perfectly

adopted.

the

of

plane
the

shape while
but may
loading

middle

cross

surface

section.
of

the

member can be neglected.

2.2.2.

Torsional-flexural

The subject
chosen
defined

cross

behaviour

prismatic
is shown

section

of

member with
in fig.
2.3.

prismatic

member

an arbitrarily
The member is

to a rectangular
system
co-ordinate
respect
the
two
Axes
is
handed.
and
coincide
with
which
right
y
z
with
while
section
principal
axes of the cross
x coincides
the
the longitudinal
of
member.
axis
centroidal
with

The in-plane

displacement
point
m
of an arbitrary
(fig.
by
be
the
2-4)
can
represented
with
co-ordinatesy
and z
two components
respectively.
z directions
y and
vM and wm in
by,
These two components
given
are

v-

(Z

(y

(2.1)

0(2.2)

0-

6.

in

the

co-ordinates,
shear center
v and
and 0x is the
of the shear center
The
two in-plane
twist
the
section.
angle
of
of
cross
in fig.
2.5 by another
displacements
vm and wm are replaced
to the cross
tangential
two components,
tm in the direction
to, the
section
perpendicular
at
m and nm in the direction
which, y and z0 are
0
w are the displacements

tangent.
The tangential
tIn =v

tM is

component
a+w

sin

cos

by,

given

a+H8x

(2-3)

(zo-z)

(2-4)

where,
(yo-

H
The

shear

strain

Y)

at

point

atM+

UM is
where,
Applying
the

fourth

au

the

middle

is

surface

displacement

zero,

of

the

at

theory,

m.
point
the shear strain

hence,

at

M`m
37 , -ax

Substituting
to
respect
longitudinal

for
s

(2.6)

tm from
from

s=o
displacement
u

uo is

given

(2-5)

assumption

to

in

in
as

longitudinal

the

be set
can
sh

sin

au

33C

sh

Cos a-

the

o-

equation
to s,

2.3
the

integrating

)and
expression

of

with
the

um becomes,

i z-y-

displacement,

in

s
fHds

(2.7)

x-direction,
of point
p
the
Cartesian
is
and
are
z
y
s
measured,
the
first
derivatives
v'
are
and
w
m,
co-ordinates
at point
to
displacements
the
with
respect
axis,
x
of
shear center

where,
from which

by,

7.

e. is

and

the

to
respect
the effect

first

derivative.
The first

x.

the

of

The

point

of

the

p
is

in

fig.

cross

the

as

sectorial

2.6

sectorial

sectorial

is

to

is

section

m.

The

radius

given

in

shown

moment at
by,

sectorial
is rotating

counterclockwise
of the sectorial

moment of

static

static
is

point
when the

say
The distribution

Sectorial

sww
m

the

section,

(w)

up to

s=o

positive
direction,
that

section

cross

co-ordinate
wm
double
the area swept by the radius
line
from
middle
of the cross
section

where
taken

2.2-3.2.
The

equation

known

co-ordinate

shear center
for a Z-cross

co-ordinates

eX with
2.7 represent

m.

represents
the
along

co-ordinate
in the positive

the

is

Sectorial

when moving
the origin'

of

ds

2.2-3-1.

the

s
jH

proporties

point

about

Px,

Sectorial

As shown
of

load

twist

moment My and bending

The fourth
term of
cross
section.
displacement
the warping
caused by

axial
at the

integration

of

2.2.3.
.

of
bending

of

torsion.

nonuniform

co-ordinate

angle

terms

three

moment Mz acting
2.7 represents
equation
the

the

of

fig.

2-7.

area

on the

middle

line

(2.8)

dA
of

In

which$

As

for

the

the

sectorial

the

cross

sectorial
origin.

A is

the

area

Cartesian

of

co-or,

co-ordinates

at

which
can

cross

dinates.,
is

co-ordinates
section

the

the

%=o.

section.

the

principal
on the
point
The actual

be calculated

with

origin
middle
values
respect

g
line
of
to

of
of
the
this

8.

2.2-3-3.

Second
(warpinR

The warping
of

the

cross

=f;

of

area

constant)

is a geometrical
constant
by-,
and is given

section

Iw

moment

sectorial

characteristic

(2.9)

dA

A
i
is calculated
where,,
with
the sectorial
co-ordinates
It

be concluded

can

structures,
is

section

that,

to

respect

the

origin

principal

of

g.
in

the

theory

any point
m on the
defined
by the three

middle

of thin-walled
line
of the

cross
Z5.
z and

The
y,
statical
the
moments of area which are required
to calculate
shear stresses
at
m are Sy, Sz and SW. The corresponding
second moments of area for the cross
Iz and IW
are Iy,
section
2.2.4.

First

order
equilibrium
equations
torsional-flexural
behaviour.

The differential
the

co-ordin'ates

first

walled

order
prismatic

equations

torsional

and

flexural

combined

equilibrium
behaviour

describing
of

a thin-

member are,
-EA

EI

-EI

Gj

of

of

!* 11
dx =P
d4v
Z dx4

d4w
Y dx 4z

(2.10)
x

=q

(2.12)

4e

2dx
d ex
-E14x
dX2

(2.13)
w dx

in x-direction,
force
normal
qy and qz are
loads
in y and z directions
distributed
the uniformly
respectorque
length.
is
the
tively
per
unit
acting
and mx

In

which,

Px is

the

2.2-5.

Basic

theory

flexural

torsional-

of

The beam-column

with doubly
2.8 is loaded

shown in fig.
biaxial
with
eccentricities
the length
Z of the beam.
section

the

buckling
I

symmetrical
by a central

ez and ey which
If the initial

cross
thrust
Px

are constant
along
deflection,
due to

bending

the secondcouplespis
as very small,
considered
Px on the bending
thrust
order
effect
of the central
stresses
is given
can be neglected
and the normal
stress
at any point

by,

PX eZ z
=-PxA

In
deflected
of

investigating

beam,

buckling

zy

assumed

beam is

At

the

in
essentially
moment of buckling,

of
that

produced
and the beam passes
which is flexural-torsional.
By calculating
loads
and torque

the

intensities

produced
on the slightly

the

initially
the

up to

to

of the
by the initial

moment

flexural

a state
of
however,

are

equilibrium

lateral

stability

(2)

Timoshenko

the

equilibrium.
deflections

the

additional
a new form

of

distributed
compressive

displaced
stresses
when acting
section,
cross
(2) presented
Timoshenko
the differential
of
equations
for the flexural-torsional
equilibrium
buckling
of the beam.
These equations
are,

E jw

+ped20x=0
2xZ
2
dx
dx

d4w
EI
+p
Y dx 4X

dLw
2xy-Ped2ax=0
2
dx
dx

(2.16)

d4e.

io)e.

dx

where,

in

constant
the shear

L4v,
Z dx
.4x

(2-15)

EI

pd,

(GJ -P

_
4x-2xz2xy2
to

addition
10 represents
center.

dx

dx

the
the

(2.17)

+Ped2v-Ped2w=o

previously
polar

dx

given
moment

of

notations,
inertia

the
about

10.

(1)

Vlasov
is

to

the cross
section
M is not equal
to
He added the normal

applied

co-ordinate
bimoments.
bimoment

the

to

the

torsional-flexural

form

of
the

for

showed

in

terms

three

buckling
is

equilibrium

at

a point
this
zero,
stresses

three

chapter

2.3.

Px

sectorial

can produce
by the

Vlasov's

concept
is the

buckling

torsional-flexural
formulation

element

prismatic

torsional-

the

of

presented

thesis.

this

of

the

where
force

torsional-flexural.

of

force

caused
(2.14)
and studied
beam when the initial

equation
of the

case
general
basis
of the new finite
element
flexural
buckling
of thin-walled
in

a longitudinal

if

that

METHODS OF ANALYSING ELASTIC

STABILITY

PROBLEMS

General
Methods
problems

to

used
thin-walled

of

the

analyse
structures

as

methods,

energy
methods.
methods and numerical
methods are based on the solution
of the differenbuckled
the
of equilibrium
equations
which represent
On the other
hand, energy
of the structure.
methods and

equilibrium
Equilibrium
tial
form

numerical
methods do not require
tial
equilibrium
equations.
2.3-2.

Equilibrium

lateral

elastic
linear

the

of

solution

of

equilibrium
thin-walled
a given

equations
buckling

of
The

of
and homogeneous.
coefficients
on the geometric
characteristics
and elastic
structure
and on the load factors.

are
depend

There
equations
a)

of

are

the

differen-

methods

The differential
the

stability
elastic
may be classified

two

procedures

to

derive

the

representing
structure
these

equations
of the

differential

equilibrium:

forces
the internal
By calculating
caused by the
loading
the
system when acting
initial
of
stresses
displaced
the
member
the
and
considering
slightly
on
between

equilibrium
at the moment

of

the

external
(1,2).
buckling

and

internal

forces

ii.

By applying
the principle
of stationary
the calculus
using
concept
of variations
the differential
of equilibrium
equations
(3).
energy
expression

b)

The methods

be used to
may be classified

whichcan

equations
of equilibrium
form)
methods,
and approximate
2.3.2.1.
If

the

the

differential
(closed
as exact

solve

form)

solutions

is considered,
the
parameter
differential
equations
can be expressed
the
This parameter,
together
with
parameter.

this

load

the

are

The exact
is based

unknown

solution

of

quantities
of

the

equations.

differential

the

functions
on choosing
suitable
the deformed
represent
state
of the structure.
the boundary
conditions
must satisfy
and loading

equations

from

methods.

Exact-(closed

a single
of the

coefficients
in terms of
deformations,

energy
and
to derive

equilibrium
to
These
of

functions

the

structure.
After
equilibrium

which
differential
from

the

of

of
is

few
that

beams

assumed

the

of the
matrix
is considered
matrix
the

as

constant
equilibrium

can

be

solved

torsional-flexural
is

exactly,

that

of

PX
thrust
with
eccentricity
a
carrying
The
be.
k
the
length
the
m.
of
along
be
derived
this
for
can
case
equations
(section

2.15-2.17

equations

of

examples

elastic

I-beam

supported

ey

in

functions

criterion.

One

simply

substituting

and constructing
equations
the determinant
of this

coefficients,
the stability

buckling

the

2.2-5),

and

these

equations

become,

EI

d4v
Z

dx4

(2.18)

+Pdv=0
X
2
dx

d4-w
EI
+p
Y4x2xy
dx

d2w

dx

ped2

ex =0

dx2

(2.19)

12.

d4x

EI

(GJ

dx4

Io)d 26
-P xA

The end conditions

the

dx

at

beam are:

supported
z=k

and

z=o

(2.21)

(2.22)

z=Z

by taking

satisfied

v,

w, and

form,
A sin 2-x ,w=A
1z2z3P,
Substituting

(2.18),

are

(2.20)

=0

and

z=o

at

2-0

conditions

dx

a simply

ex

d2vd2wd2
2=
2dx
dx

ex in

x-Pod
2xy

for

OX =o

w=

These

dx

2w

(2.19),

and
become,

equilibrium

jLx-

sin

Ox
=A
,

by these
functions
(2.20)
the differential

sin

into

2x

(2.23)

equations

equations

of

2
(EI

Z-2-Px)A1=0
9

(EI

Px-e

y t2

-px)A2+

buckling

Px. e

Y'

in

of

these

(2.25)

A3=0,

(El
+
GJ -P
-A
-+
_12
2xA3
y2P,

The first
the

(2.24)

,
-L

O)A

(eq.

equations

2.24)

shows that
independent
and the

is
of symmetry
plane
buckling
load is the same as the Euler
load.
corresponding
(eq.
The second and third
2.25 and 2.26)
equations
show that
the lateral
buckling
in the xy plane
buckling
and the torsional
The corresponding
load can be obtained
are coupled.
critical
by equating
to
This
condition

the

(2.26)

zero the determinant

is given
by,

of

these

two

equations.

13.

(F

y-p

A==0

pxey

X)

(2.27)
(F

pxeyIo

in

e-p

X)

which,
22
Tr
Fy = EI
y2e10w

which

FA

(2.28)

By expanding
the determinant
from
A, the equation
the critical
is given
by,
load can be calculated
2_ Rp (F
px
+F )+
x
y

0220
2_
2)

where

The smallest
gives

Tr
GJ
2
21

(EI

the

critical

exact

equations
to simple

and limited

and

(2.29

of

(2-30)

10A

(closed

equilibrium

structures.
to solve
the

approximate
methods
be explained
these
methods will
2.3-2.2.

7F=0

positive
solution
buckling
load.

Unfortunately,
differential

KF

Approximate

of

form)

equation

2.29

solutions

of-the

are comparatively
There are,
however,

equilibrium

equations

rare
some
and

now.
solutions

of

the

differential

equations
a)

Infinite

series

In

differengoverning
equilibrium
can be carried
out by assuming
This
deformations.
the
to
series
represent
the loading
and constraint
of
conditions

tial

some cases

equations
of
finite
a suitable
series
must
the problem.
number

of

solution
the

satisfy
The accuracy
terms taken
from

solution

of

the

on the
of the method depends
(2) used
Timoshenko
the series.

14.

the

to

method

buckling

lateral

the

study

of

He used
load.
a concentrated
the angle
represent
twist ex.
of
(4) showed that
Trahair
a Taylor
to

subjected
to

series

application

can be used
beam.
b)

differential
gration
method

to

the

express

Iterative

integration

Another

approximate

twisted

an I-beam
a trigonometric
In

another

series
OX of the

shape

expansion
buckled

method

the

for

method

the

of

solution
the iterative

is
inteof equilibrium
equations
The method is known as the Stodola-Vianello
method.
(3) and is sometimes
the successive
approximation
called

method.
The method
differential

equations
for the

approximation

shape must
integration

assumed
numerical
equilibrium
deformed
repeated
In this
each
load.

based

is

of equilibrium
deformed
shape
the

satisfy
of

the

boundary

differential

equations

in

to

corresponds

a certain

The procedure
can be continued
The applications
is obtained.
accuracy
integration
method have shown excellent

form solutions
(2,3,4).
problems
closed

Finite

for

beam and

difference

The finite

difference

for

differential

equations
form.

value
until

the

of

the

of
the

the

can be
function.
where
buckling

desired
iterative

agreement

column

of
for

with

the

stability

solution
method
differential

solving
complex
to stability
method can be applied
for buckling
loads
in
mate values
closed

The

conditions.

a new improved
representation
Then the procedure
shape of the structure.
to obtain
for the assumed
a third
estimate
of functions
way a series
can be generated,
results

function

method

the
numerically
from -an initial
starting
This
of the structure.

on integrating

of

equilibrium

is

an approximate
The
equations.

problems
some cases
cannot

to

approxiwhen the
be solved
in
give

15.

The method is based

is applicable
equationpwhich
independent

variable
one for

equations,

each
function

of

an

algebric

the
within
a number of points
the differential
operators
of
point,
by finite
f(x)
are represented

difference

approximations
of f(x)
of the values
of
the
polynomial
shape for
of the differential
The solution
of the
desired
unknowns of

of

range

of

each

At

range of
x.
the dependent

over a certain
by a finite
number

differential

the

on replacing

can be given

which

as combinations

points,
assuming
some
The boundary
conditions
in the same way.
represented

neighbouring
f(x)
values.

equations
resulting
the problem.

are
homogeneous

gives

equations

the

The application
of the method to buckling
problems
by analysing
buckling
the lateral
of a
can be demonstrated
loaded
I-beam
by a uniform
bending
moment
supported
simply
differential
The governing
M (3).
of the buckled
equation
by,
form of the beam is given
d4E
GJ
EI
w42
dx

The boundary

26

M2
6
Z-I
y

dx

conditions

d E)
2
dx

parts
points

of

width

are

b=

given

eo'el'02'
If

span
X/n the

x=o

of

(2-32)

and x

the

beam into
e at

of

values

the

equal
end and interior

by,

***
interval

(2-33)

e
len-l, n
b

width
function

of the
slope
by the slope
approximated
of
AB or BC (fig.
2.9)
and then,
small,

the

the

are,

at

the

Dividing

(2-31)

is

e at
any

of

chosen
the ith

the

two

-sufficiently
point
straight

may-be
lines

16.

i-l)
(!
dx i(leftf

' and

The differential

(2-34)

d4'

the

at

dx4

ith

(2-35)

b2

44e.
(d

+6e.

1-1

i-r2

4)

-e

461i-i

i-2

(2-36)

the

Substituting
in

of the differential
values
(ea_.
the
2.31),
ealuat-ion
i
is
point
at any
valid

approximate

differential

the

operators
difference*equat-Jon

is

which

by,

given
e

i+2

whe re

6i+l
K,
-

K,

the

EI

EIw

M2 Z4

n2E2,1

conditions

2.37

of

the

wy

e-1 = -e1
is

(2-39)
n4

give,

valid
homogeneous

of linear
of the rotation
moment Mc can

determinant

=0

GJ Z2

Equation

values
buckling

(2-37)

(2-38)

eo = en =0.

a system

+ ei-2

GJ t22

4+

boundary

Oi_l
K,
-

+K 26i

K=6+2
2

and

the

and

-1
b

2 ej + ei_l

(d
2i
dx

and

d2e
2,
dx

operators

26)

dx

i+l

by,

be given

can also

i-e
(! )
dx i(right)=

at

and

n-1

en+l =-

points.
in the
equations

en-1

It

(2-40)

represents

unknown
The approximate
ei*
valu. e of the
by setting
to zer. o
be calculated
equal
of these
equations.
coefficients
n-1

point

17.

The accuracy
of
by increasing
the number
is

the
of

may be improved
either
into
Z
which the span
of the representation
accuracy
However,
two
each of these

solution
intervals

divided

the
or by improving
of the differential
operators.
increases
the labour
modifications

solving
for
unsuitable

and makes the method

differences
computer
program

equations
A finite
obtain

more accurate
d)

Finite

solving

integral

on considering
in the highest
equation
the

the

hand

then

calculations.
be developed
to

the

technique
an approximate
The
method is
equations.

difference

into

beam is

divided
b= Z/n.

differential
derivative

finite

is

method

differential

complex

As for

simultaneous

solution

integral

based
f(x).

can

the

results.

The finite
for

of

integral
an
as
equation
variable
of the dependent
k
of
method the length

parts
a number of equal
is then
integral
equation

width
by
replaced

The
where
one for each
a finite
number of homogeneous
equations
The dependent
lower
derivatives
f(x)
variable
and its
b

replaced
vatives

by a combination
of f(x).

of

the

values

of

the

of

highest

point.
are
deri-

by
of the method can be illustrated
buckling
beam with
of a simply
supported
analysing
loaded
by uniform
cross
a
rectangular
section
narrow
is
The
differential
bending
equation
governing
moment.
by,
given
The application
the lateral

d20

2'/GJ
m
- EI

where

Equation

(2-41)

+0o

2.41

R+K

(2-42)
y

can

be rewritten

f0

in

integral

R dx dx + Ax +B0

form,

(2-43)

18.

E)

where

.R=d

mined

The constants
from the boundary

dx2

of

and B can be deterThese

conditions.
A,

integration

and symmetry

are,

conditions

0=o

at

and

x=o

these

Substituting

018
dx

two

conditions

and

Bo

at

x=

in

(2-44)

Z/2

2.43

equation

gives,
'/ 2

dx

AfR

(2-45)

0
the

As in

in

ax

method,
to
fitted

by a parabola
is given
parabola

may be approximated
this
of R;
values

difference

finite

2+

three

function

adjacent

by,

(2-46)

bx +c

which,
Rj+j

22Ri +R
i-l

Ri+j

,b

Ri

integrals

of

-R i-l
2b

2b

The

the

a xi
6 are

(2-47)

b xi
given

2a
xi
-

by,

xi
R dx

12

5R+8R.
i-i

-R i+l)

(2-48)

and,

19.

x i+l
(4 R

R dx = lb
12

Ri+, )

16 Ri +4

i-i

(2-49)

xi-i

By making the second

2.43
in equation
substituting
-2
TTb (12
- 144

R;

Ri_l

integral
the

the

of

integral

+ 154 Ri

function

equation

+ 60 R

i+i)

:-- ,

and

becomes,

(2.50)

it
is
2.50
at
points.
represents
valid
n-I
Equation
The
homogeneous
linear
critical
of
equations.
a system
by equating
to zero the
buckling
mo=ent M. can be calculated
More
determinant
of these equations.
of the coefficients
details
of the method to the stability
about the applications
in
7.
5,6,
found
be
and
references
problems can
Energy

methods
General

2.3-3.1.
The use

of
is

stability

elastic

the

energy
method to
based on the principle

solve
of

the
the

problems

of

stationary

the equilibrium
characterizes
which
condition
value
of energy
This principle
in an elastic
can be stated
as: "the
system.
does
potential
of an elastic
structure
energy
amount of total
of
configuration
passes from its
not change when the structure
equilibrium
This
can

to

an infinitesimally

be expressed

Uw +V=

near

adjacent

configuration".

as,
(2.51)

Stationary

whe re ,
U is
of

the

the

applied
Equation
6u

w+

total

load,
2.51
6v =6

potential
and V is the

strain

can be rewritten
(uw + V)

=o

Uw is

energy,

the

potential

energy.
in

the

form,
(2-52)

20.

in

6Uw is the increase

in the potential
which,
energy
of the
load and is equal
to minus the work done by the load
acting
during
the virtua-1
displacement,
in
and 6V is the increase
the strain
energy
of the structure.

often
load.

The use of the method to solve

stability
leads
to approximate
of the critical
values
The solution
depends
on using
approximate

shapes that
The accuracy

satisfy
of the

deformation

shapes
Timoshenko

for

the

deformation

boundary

conditions
of the problem.
depends
the assumed
on how close
to the exact
ones.

solution
compared
(2)

problems
buckling

was the

to

first

use

the

energy

method

the

approximate
solution
stability
problems.
of elastic
(3) published
At about the same time,
Ritz
his general
method
in mathematical
for the direct
of minimum problems
solution
Ritzfs
method is quite
physics.
and it has many
general
The method was later
in stability
applications
problems.
(3).

by many investigators
extended
and refined
tion
of Ritz's
method to elastic
stability
illustrated
in the next
section.
The Ritz

2.3.3.2.
Considering
supported
by uniform
given

the

The applicais
problems

method
buckling

lateral

elastic

beam with narrow

rectangular
bending
moment M, the total

cross

of

section

potential

a simply
loaded

energy

U is

by,

U=

GJ
0

The angle

of

dO
(jx)

co-ordinate
boundary
conditions

as the

are

set

of

2
(2-53)

dx

0
by the

finite

chosen functions
functions
must

angle

of

parameters.

rotation

series,

(2-54)

--- +b nen

+b 21P2 + ...

are arbitrarily
These
functions.

a corresponding

Q can be expressed
x

rotation

6x =b l)i
where,, )-terms

21)2m2
d2
dx. - U,
dx
5r

of

called
the same

satisfy
The b-terms
ex

21.

Substituting
(eq. 2.53)

expression

(b
11

U=f

in

which,

bn which
principle

2.54 into
equation
total
U can
energy

from
the

mf
EI * GJ

b
..,

are

energy
be given
by,

(bl,

forms
f 1 and f2 are quadratic
of
of the problem.
are the variables
(eq.
2.52)
of stationary
energy

conditions

the

)
b
..,

(2-55)

the
the

stability

by,

given

3u
Db

Equation

(2.56)

2.56

coefficients

the
the

represent
failure

in

zero

equations.

The-accuracy
increasing

critical

to

by equating
of these

calculated

Ritz's

of

number of terms
dependent
function

is
restriction
of the problem.

functions.
or transcendental
they must satisfy
that
the
However,
in the majority

results
can be obtained
functions
iP form a system of

tory

Numerical

techniques

2.3.4.1.

General

The use
analysing

by ,
method may be improved
taken
to
series
of the finite
ex. However,
success
or

method depends
mainly
on the proper
applyingthe
These functions
q).
functions
the co-ordinate
can

choice
of
be polynomials

2-. 3.4.

homogeneous
n linear
of
buckling
moment Mc can be
the determinant
of the

a system

represents
from which the

equations

of

elastic

the

The

only

boundary

conditions

of

satisfac-

cases

when the co-ordinate

(3).
functions
orthogonal
only

closed

form

on the

displacement

and energy
is
limited

structures,

such

resulting

as

calculations
differential

continuous
needed
equations.

method

for
solutions
to simple

structures
difficulty

of

stability

based

problems
such as single
span beams and beam-columns.
the more
of applying
such methods to solve

amount

bl,,...,

parameters
Applying
the

beams and frameslarises

to solve
the large
sets

The
complex
from
of

the
the

22.

The advent
made possible

to

relationship
involved

in

matrix

the

sets

of

been

five

extended
However,

problems.
to certain

types

to

their

In

the

recognised
load-displace-

routine

calcula-

the applications
years
to the analysis
of
most of these
studies

of

stability
problems.
into
can be divided

(with

method

specific

method.
now be reviewed
and discussed
with
in elastic
stability
problems.

will

applications
Member

2.3-4.2.

the

applications

element

methods

been

the

of

nature

The member stiffness-matrix

boundary
conditions).

Both

long

representing

twenty

The methods used in these

two main types,
namely,

regard

has

programming.

last

The-finite

has

equations

way of
meet the

have

methods

elastic
stability
have been devoted

2.

computer

large
to deal with
new approach
Nevertheless,
the matrix
equations.

computer

During
of

digital

electronic

a completely

sets of simultaneous
formulation
of large
as the most convenient
ment
tions

the

of

stiffness-matrix

conventional

analysis
factors

derivation

method
of elastic
is based

plane
on the

frames,

the

of member stiffness
the member carries
that
bending
assumption
moments and shear
On the other
forces
the differential
hand,
only.
equations
behaviour
the second-order
of equilibrium
of a beamgoverning
the effect
of the direct
axial
strains
caused
column member,, include
by axial

forces.

Thus,

in

comparison

to

the

first-order

matrix
member, the second-order
of an elastic
matrix
but modified
by multithe same stiffness
factors
contains
factors
functions.
Values
plying
of
called
stability
between
functions
depend on the ratio
the acting
stability
stiffness

axial

force

and

Credit
first

to

plane

elastic

use

the

value

probably
the

matrix
buckling

of

Euler

goes

to

stiffness
of

steel

buckling
Livesley

load.
(8)

method to
frameworks.

for

being

the
analyse
He described

the
ina

23.

computer
in-plane
loaded

carry
buckling

program
elastic

tops.

column

at

(9)

Renton
the

the

analyse
frames.

elastic

His

only

to

method
types
certain

only

at

the

study
buckling

of
tops.

column

top

subjected
loads.

given

of

this

based

of

the

the

subject

proposed
to

d4w
EI
+Px
74
y

differential

X)
(ILv
_zd26
2
20
X dx
dx
d2wd26
(7-2 + Yo
2x)
dx
x

matrixpthat

the

question.

of the second-order
the torsional-flexural

columnvis
differential
Eulerian

These

j4v
,p
EI
Z dx4

of

stability

his

thin-walled

axially
form solution
of
by Kappus (14).

method

and asymmetrical
Razzaq and Naim (11),

elastic

Chaudhary's
is

matrix

The derivation
(9) for
by Renton
loaded

Renton's

extended

matrix
on the
stiffness
equilibrium
of the differential
equations
buckling
structures
of thin-walled
however,
have shown
Aly and Sato (13),

discussion

a later

accuracy

(10)

analyse

(12)

form solution
closed
for torsional-flexural
(1).
Vlasov
by
given

sections

space
symmetrical
is applicable
as it
loaded
and to frames

orthogonal
single-bay
single-story
to equal
concentrated
and unequal

unbraced

Chaudhary

in

cross

symmetrical
frames.
Later,

space
to
same method

rigid-jointed
space frames

of

problem
stability
has some limitations

modes of

the

column

of

Chu and Rametsreiter

deflection
the large

to

used

to derive
same procedure
torsional-flexural
the elastic
to
this
He
matrix
used
members.
the

followed

matrix
thin-walled

stiffness
buckling
of

the
and predict
out the analysis
frames
load for two-dimensional

to

based

stiffness
matrix
buckling
of
on the closed
equations

equations

given

are,

(2-57)

=0

(2.58)

24.

d26

x2w

GJ
dx

d40
x-p
x4

EI

(y
x0

which., yO and zo are the

to the centroid.
with
regard
in

The
2.59)

for

v=v

acoo

w=w

aoos

e=ea

from

end

Renton

(9)
to

symmetrical

yx

+ jx

+ ao

(2.61)

sinh

x+eb

6a,

alp
b'
integration

EI

differential

equations
of
displacement
functions

chosen
that
the

joints

py

= -Pxj

sufficiently
acting
at the

zX+

si ''y
Vb

sinu

X)

z X)

2)

two

ends

(2.64)

(2.65)
(2.66)

x0
and

stiff

by,

be given

(w Cos Ix+w
my = -P
xayb

mx

(2.63)

equilibrium
(2.60-2-62).

are

The load

(Vacos

(GJ -Pi
x0

2)

zw

the

(GJ
=
-Pi

ao

and
1
are
,
01
0
can be evaluated

which

2="I
fi

and

assumed
be heglected.

mz =px

10 )19 a

and,

the

then

(2.62)

)
1x+0

Y x+

by,

+ '%Bin yx

conditions

wa rping
the member can

(2-57

(2.60)

wa,

with

equations
is given
section

center

+ ao

solving

(2-57-2-59)

shear

(2-59)

+ alx

)12 =P2=Px 11yz),

yX
EI

In

2
dx

X)=

+ Vbs in yzx

Wb'
constantsof

the

20

the

of

differential

i2

zx

cosh

Vb'

wherejv
a,
independent

dx

co-ordinates

the

of
solution
skew or double

22d20
dw-Zdv+
20
dx

Pz=

-P xa1

(2.67)

for
of

-,

25.

The last
opposite

pairs

stiffness

four

equations
moments or forces

of

matrix

the

of

(2.64-2.67)

define

equal

acting
at the ends.
by,
can then be given

member

and
The

a
a 12

Sjmmjric

a22

Ka
13

a23

a 33

a14

a24

a 34

[all],

where., the submatrices

Appendix
A. 2.1.
2.3-4-3.

(2.68)
a44

[a,

and

2]

Finite

element

[
a 441 are

given

in

method

The finite
whose

active

short

period

solve
of its

structural
theory

of

technique
element
method is a numerical
development
has been pursued
for a relatively
The method was originally
developed
of time.
but the natural
base
engineering
problems,
to problems
in many fields
makes it applicable

engineering.
The basic

method,
when used in
is that
structural
engineering
a whole
structure
problems,
(the
be
by
of subdivisions
can
represented
an assemblage
A set of displacement
functions
is used
finite
elements).
describe

concept

(approximatelyy

of

the

the

deformed

state

of

the

points.
of the displacements
at the nodal
for each typified_unit
soluti
on is formulated
and
to obtain
the solution
for the whole structure.
in

to

terms

to

structure
The
then

combined

linear
of elastic
analysis
conventional
by the finite
structure
element method, the energy concept
the first-order
is often used to derive
matrix
stiffness
The energy concept can also be employed
of the element.
the second order
in elastic
buckling
problems to establish
buckling
In elastic
load displacement
problems,
relationship.
IKE]
is
linear
however, the conventional
stiffness
matrix
[KGI
by another
called
matrix
geometric
supplemented
In the

26.

(stability)
effect

of

For
factor

is

of
less

a value

equilibrium,
stable
than its
critical
by the first
given

element
stiffness
equation
the potential
energy
expression
{P}

in

which,

Eq.

2.69

[K

E]

[A}is

the

describes
In

that

{A}

second

{dP}

{dA}
is
which.,
the displacements
in

(2.69)

+ [K

the

matrix
{dP}
and

G]l

vector.
behaviour

order

prebuckling
is being
conducted
analysis
at
Eq. 2.69 can be modified
to,
[[KE]

of

variation

becomes,

displacement

stability
deformations

load

where the
the
value,

[KG] (A}
+

nodal
the

elastic

elastic
deformations.

represents
matrix
load on the buckling

conditions

of

the

This

matrix.
the applied

it
problems
have taken
a near

of

is

the

usually

place
buckling

and

element.
assumed
the
that

state.

(2-70)

fdAl

of vanishing
is the matrix

small
of

increments

of

corresponding

forces.
At

load,

more than one equilibrium

is possible
and the deformations
of the structure
state
load factor
to a given
infinite
corresponding
can reach
(infinitesimal)
load increments.
for arbitrarily
small
at

the

the

buckling

critical

stage

eq.

where. p{dZ}
instability

load

through
used

to

(2.71)

the buckling
(eigenvalue).

represents
parameter

The analysis
from which the

becomes,

2.70

[KG]l
(d3}
Xe

begins
individual

a pre buckling
formulate
the

analysis.
geometric

values
Thus

deformations

and

Xc is

an

of the applied
value
are calculated
end forces
element
The end forces
can then be
[K
The critical
matrix
Gj*

with

a chosen

27.

Xc times
the
parameter
The instability
of the load factor.
chosen value
problem
then becomes an eigenvalue
the instability
of finding
problem
(eigenvalue)
XC from the nontravial
parameter
solution
of
load

is

eq.

2.71.

to

equal

Such

the

solution

IRE 1+

I RG.1

'c
in
to

instability

IR

when,

exists

(2-72)

=0

IK

the two determinants

whi-ch,
are
and
corresponding
El
GI
IKE]
[KGI
the stiffness
matrices
and
respectively.
The

of
and broad application
potential
the finite
element
method to structural
stability'problems
in a study
was made clear
of the beam-column
reported
problem
by Rodden,
(15),
in 1963.
et al
In the same year Gallagher
(16) published
and Padlog
in which they
a similar
study
simplicity

suggested

to

cubic
polynomials
cements of the beam-column
method to elastic
stability
however,
up to 1969 these
in-plane

applications
(17,18,19,20).

The extension
of
torsional-flexural
elastic

with

were

displa-

in-plane

represent
Many applications
member.
been
have since
problems

buckling

flexural

the

devoted

of

the

presented;
to the

deal
to
method
element
been
has
buckling
problems

the

finite

the growing
coupled
members.
with
use of light
gauge steel
Light
low torsional
their
rigidity,
members, with
gauge steel
have a high tendency
to buckle
in torsional
or combined
torsional-flexural

modes.
has been

flexural

buckling
(1,2,3).
many authors
theory
have been limited
systems.

The finite

effective

tool

such

as

space

of

the

first

to

element

to

(21)

order
basing
his

elements
the differential

well
However,

cope with
frameworks.

Krahula

The basic

equilibrium

of

torsional-

and explained
established
the applications
of the

by

and regular
structural
simple
technique
a very
provides
large
structures
scale
and complex

presented
torsional-flexural
derivation

theory

a finite
on the

equations.

element
behaviour

formulation

of thin-walled
form solution
closed
of
(22),
Krajcinovic

28.

however,
element
thin-walled

was the
technique

the scope of the finite

extend
buckling
torsional-flexural
elastic
He used the energy
method together

first

to

to

of

members.
to derive
the
displacement
trigonometric
function's
with
[K
(instability)
the
matrix
geometric
elastic
stiffness
El and
[K
to the exact
He concluded
that
in comparison
matrix
G] *
for the
the method gives
an upper bound estimate
solution
(22,23).
buckling
load
elastic
(24)
Gallagher
a finite
element
presented
and
formulation
instability
for the torsional
of
and lateral
analysis
beam-column
members based on an approximate
of
representation
the flexural
displacement
and torsional
of the member. They used
[Kj
the energy
the elastic
to derive
concept
and the
matrix
[KGI
The method showed an excellent
geometric
matrix
.
agreement
with
exact
solutions
of beam, column,
and beam(24,25).
The same procedure
by
problems
column
was followed
Barsoum

to
many investigators
buckling
of continuous
at

and one bay

(28).
tops

column
The finite

lack

the elastic
torsional-flexural
(26,28),
unbraced
and braced*

(27),

frames

portal
loaded

analyse
beams

symmetrical

formulations

element

presented
formulationsl

frames

so far

These
and consistency.
are
only to members with doubly
'cross
symmetrical
Furthermore,
bimoment,
the effect
of external
importance
in light
great
members,
gauge steel

generality

applicable

space

sections.
may be of
not been considered.
In
formulation
Vlasov's
members.

the

(chapter

chapter

is

presented.
(1)
concept
of
T4e technique

shape,
and it
for
influence
2.4.

next

This
the
is

new terms

sections

with

behaviour
any

of

of

element

on
thin-wal

sectional
the bimoment

cross

representing

no axes

REVIEW OF

a new finite
is based

formulation

general
for
valid

includes

LITERATURE

3),

which
has

led
,

symmetry.

PREVIOUS STUDIES OF TORSIONAL-

FLEXURAL BUCKLING PROBLEMS

Single

of

span

elements

The elastic
torsional
thin-walled
columrsloaded'by

and

torsional-flexural

either

axial

or

buckling
eccentric

29.

has been extensively

studied
(32)
(1,2,3,29,30).
Renton

thrust
authors

for

solution

cross section.
form solution-of
beam-columns
study
loaded

loaded
thin-walled
axially
(31)
by Culver
Studies
torsional-flexural
with

carried

out

and explained
the
presented
bars
were
buckling

open cross
section.
(33),
by Pekoz,
et al

with
devoted
of

by many
direct
open
to the

simply

supported

A comprehensive
on eccentrically

open
cold-formed
single
symmetrical
columns with
but sufficiently
has led to a simple
sections,
accurate
for
procedure
such columns.
The general
theory
of torsional-flexural
doubly'symmetrical
span beams with
cross

single

closed

cross
design

buckling

of

sectionsphaving
been
end conditionsphas
(1,2,3,34).
A comprehensive

simply
supported
or fixed
by many investigators
presented
literature
survey
of the work done in this
(35)..
by
(36) has
Lee
Nethercot
presented
either

another
buckling

survey

of

the

investigations

single
span beams up to
(37)
R.
ockey
and
presented

of

Nethercot

has

subject
also

concerning
1970.
In
a simple
buckling

been

presented
the lateral

addition,
design
procedure

loads
of the lateral
of simply
This procedure
is based on introducing
supported
a
lateral
buckling
in the critical
coefficient
moment expression,
buckling
to the plate
is
The procedure
similar
coefficient.
for a wide variety
valid
of load cases and supporting
conditions.
for

rapid

estimation
I-beams.

The

superior

accuracy
techniquepin
to the
comparison
differential
solving
equations

of the
finite

finite
difference

integral
methodpfor
the important

in
was made clear
(5).
by Brown and Trahair
This technique
paper presented
was
linear
to provide
used by Trahair
a simple
and Kitipornchai(38)
load of simply
for the elastic
buckling
lateral
approximation
I-beams
loaded
concentrated
central
stepped
supported
with
Trahair
technique,
the
same
another
paper,
using
the
Kitipornchai(39)
study
of
comprehensive
a
and
reported
I-beams.
tapered
Another
lateral
buckling
of simply
supported
integral-technique
is
that
finite
the
important
application
of
(40)
the
Trahair
by
Anderson
concerning
elastic
and
presented
loads.

In

30.

buckling

lateral
They
the

concluded

load

critical

the

smaller
of
Kitipornchai

two

flanges".

Later

in

another
publication
the procedure
and

(41)
Trahair
and
refined
design
for the elastic
simple
expressions
I-beams
monosymmetric
and cantilevers.

provided
loads
of

During
element
buckling

I-beams
and cantilevers.
monosYmmetric
"the
is such that
of monosymmetry
effect
is the
is larger
flange
when the tension

of
that

has

method

confirmed
solutions
(42) used

15 years

last

the

been

the

scope of the finite

torsional-flexural
cover

to

extended

The accuracy
analysis.
of the
for
stability
problems
simple
(22,24,25,26).
are available
the

critical

method has been

for which exact
Nethercot

and Rockey
by

finite

formulation
element
presented
Barsoum and Gallagher
(24) to analyse
the torsional-flexural
buckling
having
different
of single
span I-beams
support
They developed
conditions
at each end.
simple
expressions
for the lateral
buckling
15y either
moment of beams loaded
equal

or

unequal

end

moments.

More
approach

to

recently
deal with

complete
terms.

expressions
The validity

checked

by analysing
the
using

problems
been reported
complex

2.4.2.

that

problems

(43)

Roberts
elastic
for the

a number

of

by means

of

a suitable

of the
failure.

mode of

restraintspeither
or due to the
These

restraints
beam and may, in

The effect
elastic

stability

to

numerical

order
been
has

more
solve
technique(43).

beams

considerable
elastic
the end connections
stability

stability
conventional
(43,44).
However,
it

energy
method
the method can be extended

working
conditions,
beamsdesigned
as being
simply

buckling.

a new
based on

problems
including
strains
second
has
some of these
expressions

of

Continuous

prevent

presented

stability

Under

span

has

of
of

individual

single

span

the

majority

of

the

single

to
are subjected
supported
because
of the size
of
bracing
to
system provided
increase
some cases,

the

elastic'

even

end restraints
beams has been

change

on the
the subject

the

of

31.

many theoretical

beams.

studies.
and experimental
importance
of considering

the

out
pointed
the elastic

He conducted

design

the

in

end restraints

the

and

study
of
buckling

of

the

In

1955,

of
I-

span

of single
theoretical

of
study
buckling

elastic

using
out his
analysis
lacks
the generality
the analysis
However,
due to the neglect
of the warping
error,

method.
energy
in
is slightly

rigidity

influence

the

on the
theoretical

of elastic
effect
(45) carried
load.
Flint
the

and

an experimental
lateral
bracing

(45)

Flint

I-beam.
Austin

the

of

effect
behaviour

et al
flexural

(46)

a theoretical
on the elastic

reported

end restraints
span I-beams
under

in-plane
cases
of single
fully
' The beam was considered
of loading.
against
restrained
twisting
They applied
at both ends.
approximation
a successive
to solve
the differentialequations
procedure
of equilibrium
the critical
for evaluating
and presented
a number of charts
load for
loading,
in-plane
span I-bearoloaded
single
with
that
the flexural
provided
end
stiffnesses
of the elastic
are known.
restraints
Trahair

(47)

of individual
symmetrical
lateral
load of
buckling
restraints
rotation
either

to

warp

end

or

the

of

influence

on the elastic
The elastic
span' I-beams.

restraints

single

were:

considered
and torsional
free

an investigation

reported

major

axis

restraints,
while
fully
restrained

end

axis

minor

rotation,

sections

with

respect

each

span

were
to

warping.
In
restrained
warping
elastic
Salvadori

beam structures,
continuous
bending
against
major
about

by the
stability
(48),

adjacent
of

spans
continuous

(or

and

minor

span) attached
beams was first

is

elastically

axes and
The
to it.
analysed

by

for
bound
lower
approximation
a
load of narrow
the elastic
rectangular
beam
beams.
the
He
treated
of single
as a series
continuous
to
the
beams
minor axis
respect
simply
supported
with
span
the
to
bending
moments
at
major
axis
external
and subjected
supports.

who presented
lateral
buckling

32.

(49)

extended the method used in reference

(47) to the elastic
lateral
buckling
of I-beam elements with
that
provided
of symmetrical
end restraints
any combination
for the
He presented
tables
the end twisting
is prevented.
technique,
buckling
load and proposed an approximate
elastic
buckling
based on these tables,
for evaluating
the lateral
Trahair

by cross beams and

beams, beams supported
of continuous
frames.
In another paper Trahair
one bay symmetrical
portal
(6) showed how to calculate
load for
the elastic
buckling
single
using
span I-beams with unsymmetrical
end restraint
(49).
the tables
in
given
reference

load

The interaction
symmetrical

three-span

buckling

behaviour

two-span

of
beams loaded

continuous
loads
or uniformly

and

either
loads

with

distributed
central
concentrated
twisting
and fully
restrained
against
movement
and lateral
has been investigated
by Trahair
at the interior
supports,
(50,51).
(50) proposed
Trahair
to
an approximate
procedure
the elastic
torsional-flexural
load of
buckling
calculate
action

the

span

the

beams by considering
rectangular
continuous
between
the spans during
buckling.
effect

narrow

starts

procedure

by reducing
the adjacent

the

minor

For
axis

inter-

a given
bending

to
spans according
Such reduced
moment values.
rigidities
the end restraining
can be used to evaluate
on the span
effect
under
load can be estimated
consideration
and then the buckling
(50)
from previously
(47,49).
Trahair
tables
presented
also
and warping
rigidities
their
in-plane
bending

suggested
buckling
graph.

of

the lateral
technique
to evaluate
a much easier
loads
interaction
beams from a simple
of continuous
This
two-span
technique
the
for
be
can
expressed

beam shown in
2.10.
shown in fig.

continuous

fig.

graph

b,

in

lateral

2 represents

lateral

point
loading
critical

span

interaction

which:

1 represents
the
point
the left
span when the
point
the right

iii)

2.1-O. at by the

the

when the

right

left

buckling
span

is

buckling
span

is

load
unloaded;
load

the

for

unloaded,

interaction,
of
zero
point
to
both
is
such
as
make
condition
time.
the
same
at

3 is

for

and

when the
spans

33.

In

addition,

tal

study

(51)

Trahair

the
concerning
I-beams
loaded

continuous
to
order
47,49,50.

by central
concentrated
theoretical
procedure
given

the

verify

He tested

I-section

continuous

a series
of high
beams under different

The

studies
reported
(53)
Hartmann
Munse
and
were
lateral
bracing
on the lateral

effects
of
lateral
bracing
loaded

by central
lower
approximate

(48)
by Salvadori
given
if the loading
condition
simultaneously
without
In a later
discussion
showed that
lower
bound

(52,54,55)

and

the-effect
of

procedure
the differential

of

elastic

rigid-jointed
(52,53)
has

been

equations

the continuity
with
equations
at the
(54) used the method to investigate
Hartmann
continuity
and of the individual
of
stiffness
buckling
behaviour
on the lateral
of continuous

the

the

to

was good.

together

of equilibrium,
interior
joints.

I-beams

references

aluminium

results

instability

Hartmann's
structures.
analytical
based on, numerically
integrating

in

in

combinations
of
level
of agreement

by Hartmann
devoted

loads

strength

loads.
The general
concentrated
the experimental
and analytical

central
between

that

a comprehensive
experimenbuckling
behaviour
of

presented
interaction

both

loads.
concentrated
bound estimate
of

He concluded
load
the critical

the actual
load only
gives
critical
is such as to make all
spans buckle
in the buckling
any interaction
plane.
(54),
(56)
Trahair
of Hartmannts
paper

Hartmann's

procedure

(54)

and Salvadorits
of the buckling

load
method lead to correct
values
if all
individual
only
spans are critical
at the same time.
(56) also
Trahair
the accuracy
that
concluded
of
while
Salvadorits
for I-section
method is not always
good, especially
beams, the Hartmann
solution
overestimated
slightly
gives
values

of

procedure,
narrow
different
laterally
In

the

buckling

Hartmann

rectangular
combinations

to

verify

an experimental
conducted
two-span
beams loaded
continuous

study

load.

in

order

(55)

of

central

by elastic
supported
the experimental

general,
the analytical
with

However,

procedure.

concentrated
bracing
at the

results

were

in

loads
interior
a good

his
on

with
and
support.
agreement

34.

With

the

the

cover

elastic
the method

applied
finite

method to
(26)
Powell
Klingner
and
problems,
beams.
They checked
their
continuous

extension
buckling
to

of

the

finite

element

formulation
by analysing
a two-span
continuous
element
I-beam
the same dimensions,
and loading
with
properties,
(51).
The comparison
that
tested
by
Trahair
conditions
of
below
in general,,
the theoretical
showed that
results
are,

the

corresponding

experimental

Nethercot
for

hand

moment
loading
elastic

of the
beams.

continuous

factor

modification

at
The

consideration.

(Tr

mcm

critical

EI

Nethercot

and Trahair
the moment factor
m
length
factor
K.
Later
for

numerical
load.
They
adjacent

stiffness
the effective
of

the

showed

two

to

allow
factor

method

a simple
loads
of

is

based

for

the

on using
type of

for

the

ends of the segment under

bending
moment Mc can then

be

to

allow

from,

calculated

two

the

suggested
buckling

elastic
The method

m
length

and an effective
restraints

(57)

Trahair

and

calculations

laterally

results.

2
7T EI

GJ,

formula
an empirical
for
and developed
a chart
gave

the

segmentspas
for
it.
matrix
length

(2.73)

Dux and Kitipornchai

calculations
of the

treated

/GJ

2)
K2 91

(58)
elastic

for

evaluating

the

effective

the

method
buckling

extended
lateral

the

with
and prov. '-Ied

critical

segment)together

a sub-assemblage
They also presented

new charts

the restraining
effect
which reflects
technique
The results
adjacent
of their
segments.
finite
the more accurate
a good agreement
element
with
K

solution.
2.4-3.

Three-dimensional
space
In

load,

or

for

the

buckling

analysis

of

plane

and

frames

a three-dimensional
deflections,
resulting

frame
or

analysis
both,, lie

the
in

acting
three-

35.

is
The exact
of such behaviour
space.
analysis
the
that
On the other
hand, it has been proved
complicated.
I
the structure
treating
of
as a series
simplified
procedure
in their
to move only
frames
of two-dimensional
restricted

dimensional

own plane
behaviour

is

study

symmetrical

the

top

description

at

restricted
I-section

study,
of

Three

a joint.
one

of

true

the

by
reported
of lateral
bracing

stiffnesses
frames.
buckling
behaviour
of plane
to single-story
doubly
single-bay
frames
with not more than two members

continuity
on the torsional-flexural

meeting
in the

correct

The comprehensive
theoretical
study
(53),
devoted
to'the
effects
was
lateral
elastic
and of, individual

Hartmann

The

the

may not provide

of the structure.

namely:
column,

a)
b)

cases of
a lateral
two

loading
(sway)

were
force

considered'

atapplied
transverse
forces
applied
equal
the two third
of the
points

at the shear center

of each of
beam, and c) two axial
loads
applied
at the column tops.
(53) showed that
Hartmann
in comparison
to his proposed
method,
(48) can give reasonable
Salvadorits
bound procedure
lower
for the buckling
load of the frame provided
the
that
values
beam is
ends.

considered
In another

fully
paper,
the effect

restrained
Hartmann

against

warping

at

both

an experimental
of lateral'
stiffness

reported

study
concerning
of elastic
bracing
buckling
rectanon the torsional-flexural
of narrow
(sway)
force
loaded
frames
lateral
by
in-plane
gular
plane
an
between. the experiThe comparison
at the'top
of one column.
mental
showed
predict

resulted
buckling

results
and the corresponding
that
the analytical
technique
the critical
load with
good

theoretical
values
(53) can be used to
accuracy.

The study reported

by Renton (9) (Sec. 2.3-4.2)
in a matrix
for the torsional-flexural
formulation
loaded thin-walled-columns
analysis
of axially.

with

doubly symmetric
or monosymmetric or skew cross
the
the
In
to
method
of
validity
order
sections.
check
Renton (9) carried
symmetrical
on rigid-jointed
out some. tests
lateral
loaded
frames
different
by
of
and
combinations
space'
The
tops.
forces' applied
transverse
analytical
column
at
either

results

showed a satisfactory

agreement

with

the

test

results.

36.

analyse
on the

load
constant
is maintained

factor
that

(10)

Chu and Rampetsreiter

the large
deflection

the

provides

buckling

technique.

a method to
based
space frames
technique
the load

suggested
of
this

In

level
at a certain
and the
balancing
internal
forces
has

deflection
to

be

through
The load can then
calculated
an iteration
process.
be increased
to new value
the procedure
and again
can be
to draw the load-deflection
from which
repeated
relationship
the buckling
load can be evaluated.
Chu and Rampetsreiter
used the
deflection

stiffness
buckling

matrix

developed

by Renton

(9)

for

small

to modify-the
and described
a procedure
stiffness
load level
to
and transformation
matrices
at every
include
the effect
of the shortening
caused by the compressive
forces.
axial
Razzaq
the

and Naim
torsional-flexural

elastic

symmetrical
different

space

frames

(11)

with

combinations
of
tops.
The analysis

axial

reported
buckling

a numerical
of

of

study

rigid-jointed

I-cross

section
members under
transverse
loads
at
applied

was based

on the stiffness
matrix
(9)
by
Renton
by Chu and Rampetsreiter
presented
and extended
(10).
The results
the effective
length
showed that
approach
the critical
load of plane
of calculating
or space frames
column

may lead to underestimated

values
of
does not consider
the three-dimensional
frame members.
Trahair

the

load

buckling
interaction

of

as it
the

(59)

buckling
the elastic
that
suggested
behaviour
frame
of a three-dimensional
can be predicted
steel
by considering
the biaxial
bending
of the individual
and torsion
together-with
the interaction
members of the frame,
effect
between
these
integral
He presented
members.
a finite
solution
for

the

differential

flexural
elastic

buckling
end

of

of

,
and experimental
one-bay

symmetrical

the torsionalgoverning
beam-column
loaded
members

with

restraints.

Vacharajittiphan

tical

equations
biaxially

symmetrical
I-section.

(60)
Trahair
reported
and
torsional-flexural
the
of
study
portal
Loads

frame
were

with

members

concentrated

a theorebuckling
having

at

the

doubly
two

37.

tops

The analytical
and at the center
of the beam.
integral
the finite
technique
to
procedure
was based on using
buckling
the elastic
load from the solution
predict
of the
differential
The interaction
behaviour
governing
equations.
column

C3

two

of

types

predicted

fixed

of
the

and

base

comparison
interaction
curves

experimental
In another

(62),

types

failure

of

three
considered
b)
nism failure,

was experimentally
theoretical
and

the

showed a satisfactory
agreement.
(61) extended
the analytical
doubly
I-portal
symmetric

paper the two authors

to solve
procedure
any shape of
frames
loading.
under in-plane
Morino

frame

portal
between

a comprehensive
study
of the
to which a space frame may be prone.
He
modes of failure,
namely,
mechaa), plastic
buckling
failure,
elastic
and c) elastoreported

failure.

The elastic
buckling
analysis
was carried
the matrix
displacement
out using
load
method and the critical
by the determinental
was predicted
for two modes of
approach
buckling,
namely,
a) twisting
modes, and b), sway modes.
plastic

(63)

Citipitioglu
the

buckling

elastic

of

by vertical
loads
displacement
matrix

at

presented

one-sory

the

column

method and
He concluded

an analytical
one-bay

space
the
using
into
account

tops
taking

buckling

moments.

buckling

the calculated
moments may reduce
load by not more than 10%.

buckling

The warping
at different

members
Such behaviour
lengthst
the
In

joint
dealing

however,

depends

the cross
and the

to
warping
Vacharajittiphan
comprehensive
distortional
members.

They presented

joint

jointed
that

the

considering

rigid
forms

the

of

value

prethe

between

joints

pre-

frame

a complex problem.
the individual
angle,

of

the

members meeting
at
itself.
at the joint
arrangement
frames
most of the researchers,
shapes

the joints
are sufficiently
(9,10,11,53,60,61,62).

be prevented
and Trahair

behaviour

at

on the

stiffness

finite

that

inclinations

sectional

with
rigid
have assumed

for

behaviour

of
study
loaded
frames

(64),

however,

stiff

element

reported
a
the warping

of

I-section

concerning
study
between
Joints
rigid

approximate

expressions

to

evaluate

and

38.

the

warping
the type

to

joint.

In

stiffness

and number

of stiffeners
to the finite

comparison
have proved

expressions
the warping

at

restraint

to

The different

effects

nonlinearity,
geometric
in a general
stability
to include
In order

by Renton

(9)

functions

to

frames
They

the

modify

finite

for

material
analysis
the effect

applications
systems

considered
Gallagher
this

a variety

The

contribution

The

finite

of

the

forces

axial

given

procedure

proper

stiffness

stability
of

matrix

stability,
solved
the symmetrical
(62) and the
by Morino
the

two

made by this

element
method
technique
for
regular
and
the majority
were

has

space
comparison

procedures.
study
proved

analysing
irregular

to
the

be the

most

torsional-

beam and

column
discussed

of the previously
to two dimensional
limited

(M
bending
The effect
the
was
of
out of plane
z)
Barsoum
by
in the geometric
and
given
matrix
(24).
direction
taken
for
However,
the positive
is

in

inconsistent

with
Furthermore,

the

matrix
given
the conventional
mation

and
nonlinearity
of three-dimensional

of previously
they analysed

of the method
(22,23,25,26,42,57).

bending

adopted

the

include

validity
of their
and accuracy
of the lateral
analysis
stability
(28) gave a lternative
and Tall

Tebedge

applicable
buckling
of
However.,

problems.

to

the

For example,
problems.
frame which was investigated
between
showed a good agreement

generally
flexural

of
stability
by Birnstiel

a procedure

conventional

formulation

element
beam-columns,

2.4.4.

of

values

member.

To demonstrate

solutions

suggested

they followed.
stiffness,
the use of
which dic. tates

three-dimensional

of

conservative

the
effecting
have been reviewed

building
(65).

structures.
on the bending

these

analysis

factors

structural
and Iffland
axial

element

stiffness.

restraint

of
load

according
angle
of the

the

and
very

give

joint

a given

matrix

study.
by the authors
needed

sign-conventions
the elastic
stiffness
(in signs)
identical

not

Thus

matrix.

stiffness
is

is

the

in

order

to

to

transfora proper
perform
a three-dimensional

39.

frame

Applications
analysis.
Barsoum and Gallagher
were
(24,42,57).
problems
only

of
limited

the

formulation
beam and

to

by

given
column

The geometric
identical
to

by Tebedge and Tall

matrix
presented
(28),
(24),
is
that
by
Barsoum
Gallagher
and
given
the elastic
in
and although
stiffness
matrix
was not given
the paper,
leads
the procedure
to the same stiffness
matrix.
(28)
The transformation
by
Tebedge
Tall
is
used
matrix
and
inconsistent
The only
validity

both the stiffness

with
space frame problem
solved
is that
of the formulation

which the columnsof

tops and no bending

frame

the

moments

and geometric
matrices.
in the paper to check the
(62)"in
by
Morino
given

were loaded
were involved.

on their

equally

The

in the first
thesis
study
reported
of this
part
in order
to pre'sen't
was undertaken
finite
element
a general
formulation
that
can be used for the three-dimensional
buckling
is not limited
analysis
of framed
structures,
and that
by the

deficiencies'of

the

formulations
previous
mentioned
is based on the general
theory
of thin-walled
structures

The new formulation

above.
torsional-flexural
buckling
(1).
by
Vlasov
The effect
presented
of
bimoments
in the general
case of loading
effect

appears

axis
of
teristic
is

the

A finite

order,

4.
and

the

cross

For

symmetry.

called
introduced.

chapter

if

only

such
"sectorial

element

sections

acting
externally
This
is included.
of

section

the

monosymmetric"

of

member has

no

charac(
constant
W)

a new sectorial

is
computer
program
can be used for first

The program
buckling
analysis

of

described
order,

second
The

structures.

elastic

in

accuracy
of the elastic
when used to evaluate
matrix,
stiffness
is discussed
in
the bimoments
torsion
caused by nonuniform
the
finite
5.
The validity
of
new
chapter
and accuracy
buckling
to
formulation
analyse
problems
element
when employed
6.
in. chapter
To. check the validity
of the new
are discussed
terms

which

experimental
behaviour
of
7.
chapter

the

reflect
and

effect

theoretical

simply

supported

of. the
of
study
Z-beams

bimoments,
external
the torsional-flexural
is

presented

in

an

Fig-2-1Warping Of Doubly Symmetrical

I- Beam

-Hkfl

Fig.2.2 Warping Of Cantilever Beam

Zo

Fig.Z3 The Prismatic Member

(Z.

-..

-Z)

ox

0)1

m%
MM m
%
le;

.0

91.
casK

Al

"

Wm

z
Fig.2.4Torsional- Flexural
DisplacementsOf Point (m)

Fig.25 Normal And Tangential

Displacements

eo,

.o

Fig.2.7Distribution Of The Sectorial

CoordincLtes For Z. Sectt*ori

-eF

i-Z

Fig.2-6Sectorial Coordinates.

PX
eY-

El
..-0"
-L--

Fig.2.8 BiaxiallY Loaded Beam-Column

Px
ley

bb
Fig . 2.9 Finite Difference

Approximation For The Angle

Of Twist 0

P2

Pi
12
A)

Pi

(B)

Continuous
A
For
Curve
Fig.210 Approximate Interaction
Beam (Ref. 50

40.

CHAPTER THREE
Finite

Element

Formulation

Buckling

Behaviour

3.1.

GENERAL

of

of the
Thin-walled

Elastic

Three-Dimensional

Systems

A finite

formulation
for the
element
behaviour
buckling
of thin-walled

dimensional

three-

elastic

members is
The formulation
is based on the
by
behaviour
as described

in this
presented
chapter.
theory
of torsional-flexural
(1).
Vlasov

forces
longitudinal
self-balancing
showed that
to points
of the cross
section
of a thin-walled
applied
The
the cross
beam-column
warping
member can distort
section.
by either
longitudinal
section
or transverse
of the cross
Vlasov

to the
of the shear center
can give rise
force
in
the
The
cross
section.
stresses
generalized
normal
bimoment.
to these normal
is called
stresses
corresponding
load-applied

out

It

is

at the moment
torsional-flexural
flexural

equilibrium
are included

stresses

three

conventional

elastic
buckling

assumed,
of buckling,

the

analysis

presented

the

structure

passes

equilibrium
but critical

state

behaviour
has

been

deformations

to the
comparison
can be neglected.
3.2.

to

state.
fourth
term

as the
terms
of the

equation

has
The energy'concept
and the geometric
matrix

matrices
buckling

Vlasov

in

from

a
torsional-

another
The b-imoment
to

be added

of

normal

used to derive
describing
matrix

as very
considered
deformations,
and

to

the

stresses.
the

been

The derivation
of the element.
deformation
based on the small
are
buckling

that

herein,

the

of these
theory.
Pre-

in
small,
effect
so their

BIMOMENT
The basic
(1) for the

a cantilever

concept
case

beam with

of

shown
doubly

bimoment
in

fig.

symmetric

was explained
3.1.
3-l.
Fig.
I-cross

section

by
a.

shows

41.

According
to the
Px.
force
an eccentric
axial
Px is
force
theory
the eccentric
of bending,
Px and pure
thrust
to a combined
system of axial
However,
if
the two principal
planes
yx and zx.
(theory
is
the
of
of
considered
section
cross
forces
system of longitudinal
structures)
another

to

subjected
elementary
equivalent
bending
in
the

warping
thin-walled
(fig.
3.1. e)
the elementary
3.1. e.

fig.

the

and opposite
flanges.
This

additional
bending
moments
is

system

The bimoment,
dinal

(fig.

force

caused

in

the. planes
a bimoment.

acting

called

3-1. e) or
(fig.
2.2)

loads

transverse

by
to the three
given
components
(fig.
As shown in
3-l. b-3-l.
d).
two equal
system of forces
gives

must be added
beam theory

of

the

by either

longituan eccentric
by a nonuniform
torsion
of
is given
by the expression.

(3.1)

MFxh

where,

B is

the

in-plane

bending

centroids
terms of the normal
a_ in
stress
B can also
be given
bimoment
by,

of the
the cross

the

bimoment,

moment, and h is
two flanges.
In
section,

two

the

is

,F
distance

the

flange

between

the

fc,
B
A
in

which,

j is

the

sectorial

different
finite
results
thesis.

methods
element

of

bimoment

of

calculating
full
method with

of such methods
Also presented

of the
study
of Z-beams.

co-ordinate.

in
distribution
and bimoment
by many
beams has been presented
and continuous
(79-84).
A review
of these
studies
and the

The theory
single
span
investigators

(3.2)

j dA

warping

the

bimoment,

including

the

between
the
comparison
5
this
in
of
chapter
presented
are
is an experimental
in the same chapter

stresses

caused

by nonuniform

torsion

42.

If

bimoment

the

(x

BX acting
at
known', the longitudinal

) is

= const.
caused by this

bimoment

section

cross

a given

normal
from
can be evaluated

CB
stresses
the expression,

xw

(3-3)

w
in

IW is

which,

3.3.

the

warping

constant

of

the

cross

section.

STRAIN ENERGY
The

subject

prismatic
is shown

section
ch osen cross
I
the longitudinal
gives

with
2.3.

an arbitrarily
2.7
Equation

displacement
2.7

Equation

m.

point
form

element
in fig.

um at an arbitrary
in a more general
can be rewritten

by,

given

um=u-wz-V,

where,

u is

strain

can

y-

(3-4)

We'X

the

longitudinal
displacement
of the
average
(the
longitudinal
displacement
section
caused by
cross
)
thrust))and
is the sectorial
with
co-ordinate
central
The longitudinal
to the sectorial
normal
origin.
respect
then

be expressed

by the

main

stresses,

of

parts,

strain

energy
1)
namely,

and 2)

The
is

given

the

The strain

Strain

(3-5)

exw

E:
m=u-wz-vy-

two

equation,

strain
energy

strain

energy
due

energy

to

element

can be divided
due to normal

energy
due to shear

normal

U1 caused

stresses.

stresses
by the

normal

stresses

by,
fe

i2
2

m
v

dV

(3.6)

into

43.

in

which,
V is the

E is

the

volume

of

the

of

modulus
of elasticity
the element.

and

material

knowing
that
3.5
from
x
and
em
eq.
to the principal
axes ox and
and y are measured with
respect
Z
is evaluated
the
with
respect
oy and
sectorial
co-ordinate
Ul due to normal
to the sectorial
the strain
energy
origin,
becomes,
stresses
Substituting

for

91
*2

%2

(A

%%2
)

%%2

ex

1uyw+IZv+Iw

Strain

3.3.2.

The

shear

be neglected
by,
shear is given
can

due

energy

22x":

to

shear

due to
strain
(3).
The strain

(3-7)

dx

stresses
is

forces

shear

U2 due

energy

and
small
to torsional

62
dx
GJ -

(3.8)

0
in

which,

the

GJ is

torsional

The total
sum of

strain
U2 and is

U1 and

of

stiffness
Us for

energy
given

the

the

element.

element

is

the

by,

91
u=1
s2jyZwxx
0

3.4.

acting
state

22222
(EAu
+ EI

THE POTENTIAL
The general
on the cross

EI

w+

v%' + EI

OF THE APPLIED

expression
section

x=

6'

+ GJO%) dx

(3.9)

LOAD

the

ax
stresses
normal
in the precritical
constant

of

is,
PX
a=
xAIyIz+

MY

mzB

Y-W w

(3-10)

44.

in

Px is the axial
which,
bending
moments about the
and B is the bimoment.
The shear

thrustyM

axes

principal

T is

stress

and Mz are

given

the

oy and

two

oz respectively,

by,

mm
TYS+S

where,

is

ti

the

tiz

yy
thickness

where T is considered,
of the bending
moments
Sz are

the

section

about
of the

moment

static

tiw

of the cross
section
at
MyI Mz and B are the first
My and Mz and the bimoment

moments

the

of

oy and

oz axes
The
same part.

considered
and SW is the
static

the

point
derivatives
B,
of

part

static

sectorial

moments

of

Sy and
the cross

area

are

by,

given

sss
S3, =Ity

ds,

sz

ds,

=ftz

ftw

Sw

and

ds

000

(3-12)

The transition

the

from the stable

equilielement
brium
to the critical
is associated
the
state
state
with
deformations.
At the critical
appearance
of critical
state
the effect
normal
stresses
ax acting
of the initial
on the
deformed
in the form of three
cross
section
can be presented
loads,

fictitious

a)
b)
C)

qy
z
x

fictitious

distributed

lateral

load

in

y direction;

fictitious

distributed

lateral

load

in

z direction;

fictitious

distributed

torque

about

centers

the

case
discussed
loads

of

are

longitudinal

The evaluation
of
when the bimoments
by Vlasov
given

(1).

by the

the

the

shear

axis.
lateral

B is

equal
The intensities

expressions,

loads
to

jy

zero
of

4z
for
and
has been

these

lateral

45.

iy =-

(3-13)

(myox)
zoex) -

Px(v

(3-14)

qz =- Px(w + yoex) + (Mzex)

The potential

of

loads

these

becomes,

tl

I\

)v
dx
PX(V zoe
x2yx

(M e )v dx

(3-15)

(Mzex)w

(3-16)

00

(1)

Vlasov
precritical
fictitious

)w dx +

(w +ye

vP
221x0x

showed that
B is equal
to

stage
distributed

px

torque

if

dx

the

zero,
is given

in
bimoment
acting
the intensity
of the

the

by,

%%

\%

(Y. w-Z0v+y06x+Z06x+iy0x
m
(3-17)

-6x

in
i

which,
and iz

Y'
and ay are
determined

yo and zo are
are the radii
geometrical
by the two

the
of

gyration

about
of

characteristics

the

shear

center,
z
z axes,

y and
the cross

section

equations,

fIz,
z3 dA +

yAA

fy
z
z

of

co-ordinates

AA

3 dA
+f2 yz

Y2 dA)

2-z
0

dA) - 2-yo

(3-18)

(3 19)
.

46.

In
the

initial

cross
Aix.

the

case

general
B,
bimoment

an additional
this
torque
can

of

torque

be determined

by the

aw

the

The potential
of the
potentials
the

from

(3.22)

of the
fictitious

load VT is the sum of

applied
lateral
loads
qy and qz and
Thus this
potential
can be

ix

+exw*)
V-yo(w,
W-V.
X(-W.

X)

(v 0x +e v) -106x

V\

-my

mx is,

equation,
[f

V +V +V
T1232

Z0

torque

uniform

torque

distributed

obtained

cross

(3.21)

Am

mm

the

the

of

value

of

(3r2 + Z2 ) dA

wI

The total

(3.20)

characteristic

a new geometrical
by,
given

section

I\

Is

(exv +v ex +y, ex.ex) -m

exv
+
ex)m
ex.
v
Y(2
Y*ex.
Y,
m (e

xw+w6x-

z-ex

B w 6x 6x +Bw0x0

to

effect
of
deformed

X)

$w is

where,

in

the

(1),

Arnx = w (B 6x+B6

of

o)

on the slightly
distributed

when acting

in

section,
results
The intensity

equation

the

of

(B-

loading

i0
which,
the shear

is

the

center.

polar

ex

X]
radius

)+m

(2

Oxw -

z ex ex)+

(3.23)

dx

of

Mz OX*w

gyration

with

respect

47.

3.5.

POTENTIAL

ENERGY IN TERMS OF THE EXTERNAL JOINT

As shown in fig.
3.2.
loads:
of the following

action
a)

Bending

the

is

element

LOADS
to

subjected

the

My with

values

M
Yl

and M
at
Y2

end

moment Mz with

values

Mzl

and Mz2 at

end 1

Qy with

values

Q
Yl

at
and Q
Y2

end 1

Q. with

values

Qzl and Qz2 at end 1

moment

and 2 respectively,
b)

Bending

and 2 respectively,
c)

Shearing

force

and 2 respectively,

d)

Shearing

force

and 2 respectively,
e)

moment MX about

Twisting

f)

Bimoment

g)

Axial

forc eP
to

respect

can

and M
at
x2
B wit h values

MX

values

The average
by,
be given

m=i
y2
The average

M1
z2z

with

(M
1-

shear

center

B1 and B2 at

eccentricities

bending

moment My in

yj

the

(-x)

(fig.
Mz
moment

z22

with

end 1 and 2
e

2z12z2xZ

M)+1Qx+1Q

axis

end 1 and 2

x
the centroid.

(M -m)+1QQ
y2
Yl
bending

the

3-4)

yz

and e

with

element

(fig.

+p

(3.24)

3-3)

is,

Y2(k-x)+P

xey

(3.25)

One assumption
is made in order
to simplify
the
that
is to consider
0v=v0
by
analysis,
and 0w=w0
and
xxxx
for the end forces
from eq. 3.24 and 3.25 and
subsituting
(eq. 3.23)
their
derivatives
the potential
load
of the applied
becomes,

48.

v=1
T

%%

(-v.

Px

x\

2CZv.

v W. W

x+2C

y*

c0.6

w6x-

' n\ ) dx
x6x

i
(M

(9,

m+QZx+QZ
y2
12

yl

0
+1(mm+Qx+Q
41Z
i0

Z2

Yl

(9,

y2

) (2 v' 6x + y Ox 6x)
-x)

dx

(2 w\\ 066)
xZxx

-x

dx

91
(Qzl-

-'j

(Q

yl0
1a

QZ2)

(2

Q
y2

(2

Jv+

(B e+
xxxx

0w

C ex)

ex) dx

Zx

B' 0e)

dx

(3.26)

dx

0
C3r = (e

where,

Cz = (e

YO) ,

y-

Co =i2+e+e

of

been
the

z-

(3.27)

yyZZ

It

not

can be noted
by its
replaced

that

in

values

of

3.6.

DERIVATION

of the
function

element.
is given

bimoment

be given

in

3.7.

section

of

the

element

representation
The general
form

of
of

matrices
requires
a
the displaced
behaviour
each

displacement

by,

6=
in

OF THE ELEMENT MATRICES

The derivation
functional

the

(3.28)

6 is

which,
functions

which

co-ordinate

x,

the

displacement

often
and

take
Ai

is

the

di are
component,
form of polynomials

a set

has

The treatment
end 1 and 2.
solution
of the differential

at

by an approximate
torsion
nonuniform
will

equation

3.26

eq.

bimoment

suitable

and,

ZO)

of

nodal

shape
of the

displacements.

49.

The
by the

shape

the

element
by,
given

[du]

{wJ

(OXI: --

in

ment

used

herein

those

are

suggested
behaviour
of

and flexural
axial
The displacement
components

(3.2 9.

dju

[dv]

{vi}

d3 vi +d 4V2 +d 5$1 + d02

[dw]

{wi}

w=d3w,

+d 4W2 +d5 (, +d 02

[de]

{ei}

Ox =d36

vl. 9 wlP
at

components

and

X1
end 1 and

1+d2u2

xl

+d46

uvw0
2'
2'

x2

2'

+ d5X1

are

x2

2 respectively

(!-V) 1
22
dx

The

di

functions

shape

d1xdZ

=-

are

c)

+d 6X2 (3.29.

d)

the

and,

X,

=-(

dx

91

3+3

(x)

x3 + X2
7291

d6

3.9

the

strain

{C:}T

Uf
v

energy

[D]{e}

dV

(3-30)

(3-31. b)

x2
T- -x

x3 +2
d5 =-77

X)i

(3.31. a)

(Z)
91

d= -2 (2)
4ZZ

eq.

and

displace-

by,

given

T2

d23-3
3

From

(iw-) 12.
dx

(3.29. b)
(3.29.

de
IP, =2

can be

{ui}

ul,

which,

of

solutions
(24,26,28).

exact

{vJ

functions

Us can

be written

(3.31.

e)

(3-31.

d)

(3-31.

e)

as,
(3-32)

50.

in

{c)

which,

the

is

the

vector
Hookean

strain

generalized

T
{E: }

strain
vectort
[D]
is the matrix
and

transpose

of

the

representing

for

the

c from

strain

eq.

3.28

the

Us becomes,

energy

T [KE] {A

UE; = {Ai}

[KE]'s

where,
from the

the

constant.

Substituting
strain

is

the

(3-33)

i}

matrix

stiffness

element

which

can

be evaluated

integration,
[KEJ

[D]

{d*' jT

=1f2

(3-34)

{d". } dV

the potential
same procedure,
by the expression,
V can be given

Following
load

the

[K
GI
in

[K
which,

written

G]

the

is

{A

of

the

applied

(3-35)

i}

element

matrix

geometric

which

can

be

as,

[KG]

[P]

{di)T

=f

{di}

(3-36)

dV

V
[P]

is
where,
total
potential
the expression,

the

energy

{A

up

Applying

Castigliano's

equation

becomes,
{P}

matrix

[[KE ]+

of

the

Up of

applied

the

[[KE
+

element

[KG

theorem

first

[K
G]]

external
is then

{A

i}

loads.

The
by

given

(3-37)

the

element

stiffness

(3-38)

51.

The condition
of elastic
that
fact
at the buckling

by the
the

instability
load

the

of the system
potential
energy
This
leads
to an expression
for
condition
by,
which is given
total

1iE1+

1
''

is

characterised

second variation
of
is equal
to zero.
the buckling
criterion

(3-39)

"20

Gl

IR

the
in which
is the determinant
of
El
IR
the geometric
determinant
is
the
of
GI
(eigenvalue).
the instability
parameter

matrix

matrix
[K
G]

distribution
The actual
by,
is given
C3

element

BB
(x)

sinh
sinh

of

k(9, -x)
k9,2

+B

the

sinh
sinh

bimoment

bimoment

B, is

the

of the
the bending-twisting

parameter

given

B along

kx
+ Bt(x)
U

at end 1 of the
is the bimoment
at
at end 2, Bt(x)
the
torsional
to
due
applied
element

which,
bimoment

element,

Vlasov'(1)
thrust

a horizontal

where

the

BPw
h(x)

The

becomes,

(3-40)

B2 is

the

by,

(3-41)

B
the bimoment
by
showed that
caused
h(x)
Px when acting
at point
e on the cross
is w
be given
by,
sectorial
co-ordinate
can
ef

xe

total

the

any cross
section
loading
k
is
and

j
GJ
Ea
Jw

section,

X is

, and

EVALUATION OF THE BIMOMENT TERMS IN THE GEOMETRIC, MATRIX

3.7.

in

stiffness

PEI

bimoment

cosh
cosh

in

kx

the

(3-42)

general

case

of

loading

52.

k(Z-x)
k9.2

sinh
1 sinh

BB(x): --

kx
kt

sinh
sinh

+B

+B

sh
Pxweccoosh

t(x)-

kx
k9,

(3-43)

2
For

a given
to,

be simplified
B(x)

= B, Fl(x)

forg(x)

parameter

+ Bt(x)

+B2F 2(x)

By substituting
of the bimoment

energy

the

of

value

in

eq.

3.43

(3-44)

Pxwe
F
3(x)

3.26

eq.

the

can

potential

becomes,

[f
aw

B22

(BjFj

(x)

+BF+

Bt (x)-Pxwe F
3(x

2(x)

e,,ex

0
(B F, (x)
1

+B2F2 (x)

+ Bt (x)

q
)ex
(x)
Pxwe
F3
-

dx (3-45)

in eq. 3.45 two simplifimaking the integrations

have been made in order
to keep the orthogonality
cations
of
the approximate
dis, used to define
the angle
shape functions
6x, and to keep the symmetry
of the resulting
of rotation
In

geometric

These

matrix.
The term

two

\\
00
has
xxx

are,

simplifications

been

by

replaced

-0

\2

ex6 has been

includes
second term which
x
the symmetry
to disturb
of the resulting
as it proves
The potential
in the geometric
coefficients
matrix.
by,
VB can then be given
of the bimoment
2.

The

[(B,
B2

WS
0
The

numerically.

Fl (x)

+B2 F2 (x)

integrations
For

example

of
to

eq.

energy

(X))O'
I
dx
Pxwe
F3
-

+ Bt (x)

3.46

neglected

have

calculate
0

been

(3-46)

carried
out
B, Fl(x)O %2
dx, for
.

53.

(eq.

kk

a given

value

applied
(end 2)

at
is

(F

was calculated

1
(first

(x))

was then
subroutine

a unit

the

using

3.34.
to

elastic
This

B1=1

bimoment
that

assuming
element
The distribution

by eq.

order)
given
bimoment has

of the
the actual
in chapter

3.41),

of
end 1 of the
fully
fixed.

of

the
the

was

other
end
bimoment

stiffness

matrix
distribution

approximate

be in

with
very good agreement
be discussed
by eq. 3.40 as will
given
The bimoment
B(x) due to B1=1
thesis.
%2
"2
functions
dI Of ex using
a suitable
with
The procedure
A-3.1.
in Appendix
was
proved

distribution
5 of this
integrated
presented
for B2=1

VB
the potential
evaluate
B
the end twisting
Mx
and
and
of the end
2'
The
0
0
Xl'
X2*
coefficients--are
and
resulting
warping
xl'
x2'
from kZ =0
A-3.2
for kk values
tabulated
in Appendix
starting
(x)
been
treated
F3
has
As
kZ
10.0.
to
an
approximation,
up
=
W
F2
PxWe
is
that
being
to
and
an external
end
as
equal
repeated
in terms

bimoment

to

3.8.
.

STIFFNESS

functions
the
can

and Mx =1
Blv
forces

be added

to

to

B1 and B
2*

MATRIX

By substituting
d. in eq. 3.34

volume of the element

by, be given

for

the

derivatives

and integrating
V the elastic

of
with

stiffness

the

shape
to
respect
matrix[KE]

54.

oil-

EKE]
47)

where,
EA

a22 "':

12EIZ

b11 =-a11

ell

a22

C22 = a22

a 33

033

a 33

ca
44

44

=22

all

k3
12EI
a 33 =
7
-3
Z
a 44

1-2GJ
+3
T-

b 33
12EI
ba
44

44

55.

4EI
a 55 :--L

b 55 =

2GJ,, Y. +b
15

77
a 53

a62 =-6EIZ

74

C66

a66

GJ- 2,
+
30

2EIW
vCa77

77

53

a 53

b 62

a62

c62

a62

b74

a74

C74 =-

'74

(3-48)

6EIw
Z2

from
the

-b
47

62

26

can be noted
from
KE resulting

on an approximate
identical
to the

55

c 53

It

matrix

ca55

a 53

bbb
53

35

77

-6EI y
Z2

GJ
10

2EI
b66 =
z
Z

-4EIz
2,-

a-66

2EI

eq.
finite

representation

3.48

of

the

that

74

stiffness

and
analysis
element
behaviour
the element
given

based
is

by a

matrix
stiffness
b
for
the
a
of
except
signs
62' C62'
work analysis
virtual
62'
bending
This
is
due
the
b26
to
of'the
sign
a 26'
and C26 *
functions
M
repreM
follows
the
shape
and
which
moments
z2
Z1
3.10
in
a
the
displacement
However,
section
senting
v.
conventional

suitable

transformation

matrix,

analysis

in

with

3.9.

GEOMETRIC MATRIX

functions
integrating
geometric

conjunction

to

be used

stiffness

in

matrix

a three
KE' is

dimensional
presented.

By substituting
of the shape
for the derivatives
3.36
in
forces
and
d and the external
eq.
end
V
the
the
element
of
to
volume
with
respect
by,
be
given
can
stiffness
matrix

the

56.

Ill

"'i
I'vi

1.,

foxi.

'i'
W2
'u:2,2 Ve
x2,1122

Oilxi

-X2

Id
221

d 33
d

42

d
d62'
d
[KG]

72

d 44

43

Symmetric
d 55

53 Id 54
d64'

d661
d

d
73 74

(3-49)
l e24

e 22

e26 e27

e 33 '934

e37

e 35

e42 e43 e44

33
f

e45 e46 e47

6 53' e
54'

8 55

57
f

e64
866 e67je
1
1
1
e721 e 73 74 e 75 e76 je 77

which,
d

1.2, P

22

9.

42

f22

f43lf44
f

e62

in

22

f 551
53 54
If661
f
64

72 1'73 1`74 1'75 1'76 r77

2 d 22

e 22

22

1.2. P
d

33
1.2

33 =d 33

G0Px

2(Mvl-

0.3

44
2(Mzl
'

+ 0.3
b1

(B

44

d44

d 55

2Px
-15

15

33

Qz2

mz

l -B 2)

+ Qy2]

+ Qyl

+K

ti

f 44
55

2Pxx
d 66

33

e66

m
x]

d44
px 91
30
p
x
30

f 55 = d
55
f

66 =d 66

57.

2
(m-vl - my2 )-Z+ Qzl 91
60
15

2 PX CO Z
15

d77

(Mzl

Mz2)x,
-.
15

[Kb2

(Bi

)Z+
B2
-

r
91 (M
vl

Dc

Lx0
- 30

77

(Mzl

(B

Kt2 mx

(Q
+

(Mzl
+
+rbl

k+
Mz2)
15
(Bl

15

0.6

Px Cz
d--1.2
42

-1.2

0.6

Px Cz

43

zl

Kt,

20

71

(MV,

(Mzl

=e
43

(Mzl

e42

43

60

Mz2)

24

34

1z

+ 0.05

Qzj

+ 0.55

Qz2

0.55
-

Qzj

0.05

Qz2

0.05

Qyj

0.55
-

Qy2

+ 0.55 Qyj

+ 0.05

Q
y2

Mz2)

+-

43

42

0.6

'+

my2)

-y2'

Qz2 k2

Qz2 Z2
60

Mx Z2

(M

2
91

1.2

Qzl

20

+Q
z2
120

2j
) 9,

-+

e 42 -z

2P0.6
-1.1

Mx z2

Qzl z2+

B2)
Z+
-

y2
-Q
120

-m y2)t

yl

t4

15

77

(Q

Z+

-(M

Co k+

2]

yl

1-B2)Z+K

20

-m y2)
60

-M Z-4:
2)
60

Kb4

f=2Px

22
Qyi 91
60

2--20 2
7,2
2

=d

42
d
43

58.

d 62

0.10

pddded e62
x

0.10

pxe

53

e72 'z - 0.10

f
d

53

Qz2 '

PX. Cz - 0.05

(Myl

0.05
My2)
-

Qzj

Px Cy, + 0.05

72

(Mzl

Mz2)
-

+ 0.05

Qyj

+ 0.10

Qyl 9. + 0.45

Qy2 91

+ 0.10

Qy2 91

(Mzl - Mz2) - 0.45

Qyl g. - 0.10

Qy2 91

(Mzl - Mz2 )-0.10

Qyl 9- + 0.05

Qy2 Z

Qzl

0.45

Qz2 9'

Px Cy -+ 0.55

(Mz',

px- cy+0.05

(Mzl

f54

:'- - 1.110 Px Cy - 0.55

-m z2

mz2')
-

0.05

Qylz

'945=-0.10
d 64 =-1.10

PX CZ-0.55

(Myl - My2)

0*lo

e64 =-0.10

PX CZ - 0.05

(Myl - My2)

0.05 Qzl

[Kb3

0- 10 Qz2 91

+ 0,-55

(Myl

0-10
9'
Qz2
0.45
Qzl9'
+,
My2)
+
-

Px Cz + 0.05

(My,

)+0.10
My2
-

Px 'z

d74 :'- - 0.10

0.05

'5

73

Px cy-0.

e46 ' 0.10

QY2 k

e54

53

+ 0.05

e37 =-d

1.10

Mz2)
-

= 0.10

f 64

62

(Mzl

73

1-l'
=
54

53

e 35

0.05
My2)
-

e 27

0.10
=
73

f 73 =-e

53

26

(Myl

C
72

72

62

0.05
-

e 73 =01 . 10 pxcy+0.05

53

Px*cz

72 =-0.10

62

62

pxc0(mzl
(Bi

10.05

'
Mz2)
-

(Myl
0.05

-'m y2)
]
Qyl 9, z

Mx
Z]
w
B2)
+-K
t3

Qzlt

+ 0.05

OoO5
z
Qz2
Qzl

X]

59.

10.05
(Myl

pxc0-

e 74 =-0.10
10.05

(Mzl

(B,
K
-L
b3

d7

vGx
xy

e75

(Mzl

m
zl

30

15
(MZI
30

57
,

(M

2 Px CZ 91
d

(M

f76

Mz2)Z
15

0.05

)Z
Mz2
Q1
+
60

zl
+Q 60

y2

Qyl Z

2
9'

Q2

Z2

)9,

y-m

(3

2
91
QZ2
z2

+A-05

A'
Qz2 92
,
y-m
_v2)
-o
-7
60

15

m
Yl

My2)t
15

Myl

-pxcZ
30

'367

74

Qvj 92
Qy2
0.05
60

Mz2)'
15

15

30

2 Px C

which,

pxcZ

76 --

in

15

76

2
Wv2 X
-m z2)y
+
60

(Mzl

2PXCyZ

f7 5

Mx ']
+K
t3
21)

91
--

Qy2 Z]z

e47

2Pxcy
15

Yl

-1

e74

74

-m y2)

+ 0.05

MZ2)
-

+ 0.05

1
Qz2 91

0.05

Qzl 9,

Qz2 92
,
-o

Qzl 92-

My2)t

60

Ktl,

K
and
t4
t3'
integration

bl ,K b2 ,K b3'
t2,
b4'
from the numerical
resulting
coefficients
different
for
3.46.
Values
these
of
coefficients
eq.
/G-J\
A-3.2.
Appendix
in
values
(k =V -f-I ) are presented

are

the

of
kk

W
by
matrix
presented
geometric
comparison
with
(28),
by
Tebedge
(24),
derived
that
Barsoum and Gallagher
or
(22,23).
(26),
it
by
Powell
those
or
presented
-Krajcinovic
3.49
by
that
be
the
matrix
eq.
given
can
noted
new geometric
be
buckling
to
for
includes
terms
more
cases
more
which allow
In

the

-50)

60. -

The advantages
as follows:

analysed.
be stated

The matrix

can

symmetric

of

be used

the

to

c.-oss section.

new geometric

matrix

can

members
analyse
It includes
the

having

mono-

geometric
the effect

of
characteristics
and
which reflect
yz
behaviour
on the buckling
of the member.
monosymme try
,
The validity
has been
of the new matrix
and accuracy
by analysing
a number of beam and cantilever
examined
for which experimental,
integral
problems
or finite
solutions

are already
in chapter

explained
2.

The analysis
to buckling
are
the

has

available
6 of this

(ref.

40)

be

as will

thesis.

been

that
up
carried
out considering
the bending
deformations
of the structure
by torsion
accompanied
and at the moment of buckling
to another
flexural-torsional
structure
passes

equilibrium
the external

The effect
shape.
load is included

This

is

of the bimoment
in the geometric

caused

by

matrix.

any
only for
without
sections
cross
For these
of symmetry.
axis
sections
a new geometric
W reflecting
the effect
monoconstant
of sectorial
is evaluated.
7 of this
thesis
In chapter
an
symmetry
(based
and theoretical
on the new
experimental
study
effect

formulation)
loaded
son

of

validity
3.

valid

on cold-formed

simply
is
loads

supported

Z-beams

The compariwith
concentrated
presented.
the results
is also presented
as a check of the
terms. in the geometric
of the bimoment
matrix.

The

Qyl and
include
M. 11 M
signs
of the terms which
z2'
Qy2 have been changed according
to the sign conventions
the
for the shape functions
used to represent
considered
the transformaTogether
displacements
with
v
and
i.
(section
3.10)
tion
following
in
the
section
matrix
given
dimensional
three
for
be
the new formulation
used
a
can
buckling
formulation
portal
or

analysis.
have

and

space

experimental
is
comparison

The validity
been examined
frame
results

presented

problems

and

accuracy

by analysing
for which

are already
in chapter

of

the

a number

an analytical
available
and the
6. '

of

61.

3.10.

TRANSFORMATION OF AXES

element
prismatic
sys tem of axes for the subject
the
the
beginning
has been defined
and
chapter
of
previous
at
The x2.3.
is shown in fig.
as coinciding
axis was defined
line
the
the
and
z axis
y
of
while
element
centroidal
with
The
the
the
section.
cross
axes of
with
principal
coincide
The

three

form

axes
In

to

order

frame

analysis

terms

of

mation

all

out

the

involving

operation

use

of

transfor-

a suitable

matrix.
Jennings

for

dimensional
three
a general
in
and deflections
must be stated
from
The transformation
system of axes.
to the global
system can be done through

carry
forces

one global
a member local
axes
a matrix

system.

a right-handed

thefirst
in

frames

order
elastic
which they took

presented

procedure
a general
jointed
space
rigidly

analysis
of
into
the
account

effect
secondary
two displacement

of the members using

The procedure
can be described

of the misalignment
transformations.
following

(67)

and Majid

by the

steps:
[6]
in

Fora

given

local

co-ordinate

is

if

frame

member
a space
structure
the column 'vector
of the displacements
at the two end
to the global
joints
system
of the member with
respect
[6]
is the column vector
corresponof co-ordinates
and
in
the
ding to the components
displacements
these
of
conditions

system

of

the

member. the

compatibility

give,

r6]

(3-51)

aj
in

which,

ai

is

the

member transformation

matrix

given'by,

00
00
ri
0ri

(3-52)

62.

and,
mn

(3-53)

m2n2
m3n3

in

of

the

cosines
2.

3-5),
klP ml, n,
are the direction
shown in fig.
X, Y,
to the global
x axis with
respect
of the local
k2' M2, n2
cosines
axes respectively,
are the direction
local
Z
the
direction
m3,
and
axis
n3
y
are
3'
of the local
z axis.

which

cosines
and Z

(as'

The

Ki

contribution
matrix
of the

frame

[Ki]

ki]

a i]T

in
3.

which,

ki

is

To include

i
member
can be given

the

to

of

the

stiffness

overall

by,

(3-54)

-ai]
local

stiffness

matrix

of the

member.

secondary
effects
caused by the misalignment of the members in the analysis
of space frames,
(67)
Majid
Jennings
the use of a misalignand
suggested
transformation
of
ment matrix
and performing
a double
the

the

displacements.

Fig.

3.6

shows
AB is the

a given
member 1-2 where
joints
A and B of the frame,
PC,
A in X-direction,
end 1 from joint

is

the
line
the
Pc2

of

misalignment
connecting
departure
is

the

at
departure

B in X-direction,
at end 2 from joint
qc and rc are the
departures
from the line
of
of the member netural
axis
Jennings
AB in Y and Z-directions
joints
respectively.
LEE]
in
terms
Majid
matrix
and
presented
a misalignment
The
two displacedepartures
the
Pc2
P
rc.
and
qC
of
,
cl,
transformation
lead to the general
ment transformations
ra. ]
the
from
following
be
evaluated
matrix
which
can
equation,

rai]

[ai]

[F]

(3-55)

63.

3.4

Fig.

the

shows
and Mz2 in the
in the analysis

Mzl,

moments
been adopted

Positive
and )2*
by the use of the

positive

of

positive
for the

directions

which

directions

angles

end
have

i1
rotation
been dictated

of
have

)1 and ip2
d5 and d6 for the
d
d
3'
4'
3.3l.
'v (eq.
the'displacement

of
functions

shape

the

directions

a-e).
of
based
frame
dimensional
three
analysis
out a
carry
in
the
formulation
the
finite
sections
previous
given
on
element
Ia
be
to
has
transformation
the
this
matrixof
chapter,
il
Mzl
the
to
first
and
of
sign
conventions
modified,
according
the
transformation.
include
to
M
war
ping
secondly
and
z2,
,
directions
to the positive
According
of ip, and 2' the
It
[ail
to the matrix
transformation
can be modified
matrix
il
representation

approximate
to
In order

where,
00
00

Itil=

(3-56)
'0 '

ri

and,
ki

pil

mi

k2m2n2

(3-57)
-n 3

3m3

of

The analysis
a framedstructure

comprehensive
double
jointed
Trahair
the

angle

used at
tigation

n1

(64)
of
the
are

of

the

warping

presents

study

concerning
I-plane
symmetric

reported
the joint
joint.
far

that

problem
a complex
behaviour
the warping
frames

the

being

applicable

joints
In
of

Vacharajittiphan
behaviour

warping
and on the number and
the results
Neverthelesst

from

at the
(9).

behaviour

in

type

of
a large

depends

a
rigidand
on

of stiffeners
their
investhree

64.

frames
for

analysis.
the
assumed that

have

membergmeeting
is such
joint

the
The

that

by the
been

has

as the

warping,

ril

submatrix

stiff
For

the
at
condition
continuity
the warping
of one member is elastically
by
The
transformation
given
matrix
other.
transformation
the
include
to
of
modified
the

degree

seventh

'
by eq.

given

of
3.57

freedom
can

at

the

joint.

be replaced

by the

where,

submatrix

ki

m1

n,

'2

M2

n2

-Z 3m3

-n 30

00

jointed
rigidly
joints
are such

frames

For
the.

jointed

on rigid

studies

joints
are sufficiently
(7,9,10,11,12,28,52,53).

be neglected
0
angle,
at 180

to

warping

restrained
eq. 3.56

Previous

frame

dimensional

corner

that

the

(3-58)

the

boundary

warping

of

conditions
is
the joint

prevented.
Ptil

transformation
The general
matrix
the stiffness
and geometric
with
used together
dimensional
for a three
in the previous
sections
and/or

buckling

secondary
by,
given

effect

analysis
of the

frame

and which
of the
misalignment

rIr

of

oi

which

can

be

given

matrices

second
the
include

order

member axes,

is

00

(3-59)
00
000

ri

roi
1%
ri

at

65.

in
r

ri

which,

oi

is

and r" oi

given

are

is

by eq. 3.53,
r"i
by the-following

given,

b1

b4

b7

b2

b5

b8

given

3.57,

by eq.

equations,

(3.60)

01

b3

b6

b9

b4

b7

'2

15

1
8

and,

(3.61)

\6
b

L0

9
0

where,
b1=-rc9,2
b3=

+qcz3

qc k,

b5=-rc

+p

cl

ml - Pcl m3

b2,

and,
Pc2

in

b2,

qc n,

the

=-rc

nl

-P

m3

ci

n3

(3.62.

e)

(3.62.

d)

(3.62.

n3

by switching

obtained

cl

to

bg respectively.

3'

choice

cl

--ob 9 are

3'

The buckling
with

b8
+P

qc m, +P cl

b6

(3.62.

(3. b2. b)

b4-rcm2+qcm3

z3

b7-rcn2+qcn3
bg

k3

PA

b2rcz1-

of

analysis
an arbitrary

of

a given

value

of

structure
the load

starts
factor
and

66.

by perfo'rming
element
end
the geometric
of

this

elastic
found.

the

matrix
by eq.

illustrated
root

a linear
forces
are

analysis
By using

instability
The

3.39.

the

stability,

mre. a number
These
equation.

chapter.

Attention

is

of

methods

form
be

then
lowest

methods are reviewed

focused
on Southwell
plot

description

of

this

the

fo, r, solving

finite.
in,
is
a
computer
method
employed
-This
load from a second order
the buckling
predict
Chapter
torsional
of the structure.
analysis
a detailed

can
the

conditionload is

critical

to

forces

these

equation'.

There

includes

individual

in

the

next

method.
to
program
flexuralfour

computer

also
program.

kI

" :

(A)
ECCENTRICLOAD
Px

Px
AXIAL FORCE

Px14
-PX14
*'-

Px14
14
BENDINGMOMENT
(MY)
Px14

(C)

Px14
p I'L
x

BENDING MOMENT
(Mz)
---Px /4
/4
-P,

----1

14

214
x

BIMOMENT
( B)
-Px 14
Px14
---P x14
P 14
x

Fig-31 Analysis Of The Eccentric Axial

Load Px

Y-l
.

Cly
c 2)V
.0,

'Ile

B2

02

MX2

Z2,,, **j

ey

MZI 01

Fig.3.2 Element End Forces.

dx

+ve My

zclz

Fig. 3.3 Bending My

y
oc
+ve Mz

Fig. 3.4 Bending Mz

UZ2

Fig.3.5 Direction Cosines Between Local And Global Axes

I
2
-p-CE
B

From
Joint
Specified
Ends
Member
Fig3.6Misalignment Of
Positions

67.

CHAPTER

FOUR

Prediction

of

Buckling

the

Load

THE BUCKLING CRITERION

The
having

behaviour
second-order
joints
can be expressed

[[K
Fd=e

(F

el

(4'. 1)

Ad

of the
column vector
nI
[Keel
is the
joints
of the frame,
LKgg]
the
frame,
is
of
stiffness
matrix
{A
is the joint
matrix,
and
geometric
n}
TKee]
The term
+ [K, g]J which represents
of the structuie
can be obtained
matrix

where,
at the

tion

is

[Kg

structure
of a framed
by the following
equation,

the

load

external

acting

elastic
overall
the overall
elastic
displacement

vector.

the

second-order
from the transforma-

operation,

[[K
e e]

rt

[Kg

tK+[K
i]T

gTr

E]

G]]

j]

(4.2)

n
in

[K

[KE]

the
are
G]
co-ordinate

and geometric
stiffness
rtil T
rtil
in
are
and
system,
matrices
,
transpose.
the member transformation
matrix
and its
1
K
The member stiffness
matrices
and geometric
E]
[KG]
by
and
3.47 and 3.48 respectively,
equations
given
are
3.59.
by
is
the transformation
eq.
matrix
given
and
the local

which,

member

the

In

applied

load

an elastic
analysis,
stability
loading
fixed
is
the
pattern
a
structure
regarded
as
on
be
load
F
The
by
factor
X.
can
critical
some
multiplied
cr
X
the
by
load
the
F
defined
of
value
smallest
as
multiplied
become
indeterthe
the
displacements
structure
at which
of
minate

(bifurcation
This

by the

following

of

condition

equilibrium).
can be mathematically

relationship,

represented

68.

[K

e e]
{A d

in

which,
tions.

For

the

column

(4-3)

* 0

of

vector

the

of

solution

non-trivial
is given

condition

nantal

gg]]

is

the

1 n 1

deforma-

buckling

4.3

eq.

the

by,

jigg I

in

IT"I

1-eej
whichj,

geometric
be defined
then

as

the

value.

systematically
first
changes

lowest

the

determinants

root

4.4 is'known
(lowest)
critical
When the load

first

negative.

the

respectively.

matrices

Equation

positive
becomes

are

and

and

Below

(4-4)

+'

eel

EiRenvalue
The
represented

of
general
by the equation,

[A],

critical

eq.

4.4.

stiffness
load

can

has

the determinant
exceeded,
by
determined
load
be
can

the

load

at

which

the

determinant

EQUATION

solution
form

A]

The

the

is

The

critical
for
searching
its
sign.

of

of

as the buckling'criterion.
load the determinant

SOLUTION OF THE STABILITY

4.2.

determi-

X[B]
[B]

{R}

the

eigenvalue

problems

can be

(4-5)

=0

Xis

a scalar
symmetrical
the
{2}
is
called
the
vector
column
a
and
eigenvalue,
called
directly
4.5
4.4
Equations
analogous.
are
and
eigenvector.
be
treated
(eq.
4-4)
as an
Thus the stability
can
equation
'
by
X
The
lowest
multiplied
eigenvalue
problem.
eigenvalue
(lowest)
load
the
first
of
load
factor
the
critical
the
gives
in

which

structure.

and

are

matrices,

69.

Equation
the

determinant

can

be represented
xi

other

solutions

11

a 12 -

21

a 22

The

only

zero
This

if

condition

equation,
b 12

Xi

-xib

n2

nl

than

vanishes.

coefficients
by the following

-xib

aXbaX.
L nl

has

the

of

a 11 a 21

4.5

22

baXb
n2

equation
4.6.
The

--a
--a

ln
2n

nn

the

Xi

bln

Xi

2n

=0

(4.6)

nnj

be
can
system
from
resulting

characteristic
of
derived
by expand ing eq.
n roots
,
the solution
of the characteristic
n eigenequation
are the
{X
4.5.
The
of
eq.
values(X
eigenvector
values)
i) corresponding
Xi can be evaluated
for
Xi
to any eigenvalue
by substituting

in eq.
{xi}
-

4.5

for

and solving

the

ratios

of the

elements

in

The
methods,

by any of the computer-based

of eq. 4.5,
solution
based on converting
into
is often
the problem
a
(69),
by
be
described
problem,
can
eigenvalue
which

standard
the equation,

(4-7)

where,

the

matrix
[B]

and

[I]

is

the

The

is

-1

identity

given

by,

[A]

(4.8)

matrix.

be carried
out using
one
the eigenvalue
solving
of the computer-based
methods for
These methods
into
two main groups,
problems.
can be divided
(69).
iterative-methods
transformation
namely,
methods and
The transformation
the eigencan be used when all
methods
values

and

solution

eigenvectors

of

eq.

are

4.7

can

required.

The iterative

methods

70.

can be applied

eigenvalues

only are
found in

about
71.

-4.2.2.

when one or few

More details
required.

reference.

69,70

Prediction

of

displacement
This

analysis
from the

and

the

these

load

buckling

can be

methods

from

the

load-

curve

method is
the structure

based

on performing

and predicting

of
load-displacement

a second-order
load
the critical
The procedure
starts
of the load
value

relationship.

by analysing
the structure
factor
in order
to identify
A.,
displacements
lateral

under a small
the largest
of the
component
The load factor
is then increased

increment
and
positive
A. is to
displacement

by a small
the lateral

and eigenvectors

the

be calculated
in
repeated

of

value

corresponding
from

a secondto draw

The procedure
is
order
order
analysis.
the curve becomes
the load-displacement
relationship
until
The maximum value
the
flat.
of the loadpfrom
relatively
The
the critical
load of the structure.
curve, orepresents
load

arriving

of the
at higher

4.2-3.

Southwell

method

Southwell

(72)

increments

calculating
load-deflection

the

be kept

must
buckling

Euler
plot

small

in

order

to

avoid

modes.

proposed
an analytical
buckling
load of a real
below the
made for loads

for

technique

a
column using
load
buckling

itself.
by
many
and
applied
refined
for different
loads
buckling
(73),
beam-column
plane
problems
such as
(75).
frames
trusses
unbraced
plane
plane
pand

The method was later

investigators
to predict
the
types

of stability
(74),
frameworks
flexible

with

(76).

joints

Southwell's

can briefly

method

be illustrated

as

f ollows:
If
and

if

the

mid-length
equation

strut
an elastic
initial
imperfection
of

of

the

strut

equilibrium

is
is,

is

not

of
to
equal

quite
straight
the central
line
the

differential

initially
at

the

71.

d2p

+xU+
--1
2
EI
dx
Z

in

which,

is

the

force,

compressive

(4-9)

v)

additional
is
and EI
z

deflection,

Px is

flexural

the

the

rigidity

axial
of the

strut.
Both
series

with

and

coefficients
in eq. 4.9

substituting
becomes,

by Fourier
can be represented
(75).
Tr and v
respectively
n
n,
the load-deflection
relationship

sine
By

(4-10)
F-

in

which,

Pn is

the

The first
the

load

critical

nth

load

critical

for

the

strut.

perfect

from

can be calculated

cr

equation,

(4.11)
er

in

which., 6 is

the

4.11

Equation
whose
P=P
form

total

are
asymptotes
(77).
Equation
cr
by,
given

deflection

of

at

strut

in

be rewritten

can

shown

in

calculated

(4.12)

4.12

describes

fig.

4.1.

from
Fig.

eq.

4.12

known

center.

another

cr

Equation

its

hyperbola
a rectangular
line
and the horizontal

represents
the axis
P
4.11

the

the

4.2

the

The
slope

Southwell

standard

buckling

critical
of

the

plot
load P
cr

is

which
can

be

plot.

shows an alternative
Southwell
as the modified

representation
of
The plot
plot.

72.

between P/6 and P.

a linear
relationship
load P
the
by the intercept
is given
with
cr
the inverse
vl*
slope
gives

represents

The

critical

P axis,

while

The modified
element

computer

Southwell

used

the

in

finite

load

the buckling
predict
in
be illustrated
as will
to

program

second-order
analysis
of this
chapter.

the

from
next

part

THE COMPUTER PROGRAM

4.3.

The main routines

used in the study

program

the

of

The
of
a)

computer

finite
in

reported

developed
by Daviesi.
already
to include
the geometric
matrix
in chapter
which are presented

types

is

plot

program

The program
has
and the transformation

modified
matrix

be used

can

following

the

for

analysis:
Conventional

elastic

structures
of framed
freedom
at each joint

analysis
degrees
of

torsional-flexural
(mode of analysis

Second-order
structures

c)

were
been

three.

with
maximum seven
(mode of analysis
= o).
b)

computer

element
this
thesis

Torsional-flexural
(mode
structures
The

buckling
of

input

data

for

Each

joint

of

analysis
= 12)..

analysis

the

of

consists

problem

a given

framed

of

analysis
= 2).

framed

following:
1.

Joints

supports,

has

ordinates

with

to

numbered
to the
respect

co-ordinates.
according
2.

the

sectional

The

into

of

restraining

Members

divided

members
according

groups

Each

properties.

integer

numbers,

member,

the

Professor
Structural
of

of

be

Degre_es
to

the

the

second

first
and

third

Engineering,
.

including

structure,

and

by

identified
global

adopted

freedom

must

to

refers

is
to

identify

University

and

cross

by

group

of

two

end

of

In

are

identified

the

of

joint.

joints

ela'stic
the

co-

give

the

the

their

member

be

at

connecting

its

system

also

conditions,

the

Salford.

four

the
joints

73.

the fourth
a third
of the member, while
number specifies
joint
the principal
of the
chosen to define
plane
member.
in the data
loads
: The applied
are identified
the direction
joint,
at
sheet by the number of the loaded
(an
from 1 to 7), and
is
the
integer
load
applied
which
by the value
of the load.
3.

Loads

The flow
fig.

of

chart

The program

4.4.

the

is shown in
program
subof the following

computer
mainly

consists

routines:
a)

Main

c)

subroutine,
Subroutine
SPACE,

e)

Subroutine
The main
iteration

and the
load using

buckling

the

SOLVE,

and

subroutine
process

modified

b)

Subroutine

MAPP,

d)

Subroutine

BARS,

f)

Subroutine

STORE.

the

contains
to

the

calculate
Southwell
plot.

The procedure'of
load for a given

basic

critical

elastic

torsional-flexural

the

calculating

organization

by applying
structure
a small
starts
for the displacements
and by solving

of the load factor

value
the largest
of the deflection
component
can be identified.
infinitesimal
of the load can then be applied
at the
value
to start
in the critical
in order
joint
direction
critical
buckling

An
the

displacements.
The

instability

is

by carrying
out
the singularity

linearized

problem
iterative
At each load level
a doubly
process.
[R
Rgg]
At each load
determinant
the
is checked.
+ X
of
ee.
to find
level,
iteration
is performed
out the
also,
an inner
is carried
This operation
correct
of the displacement.
value
for the
the second-order
equation
out by solving
repeatedly
displacements

until
values

consecutive
the adopted
fig.

value

the
of
for

percentage
the critical

the

two

between

difference
displacement

is

less

This

step

is

shown

the

inner

tolerance.

than
in

4.3.
Fig.

technique

to

4.5.

the

shows
the
calculate

flow

correct

chart
value

of
of

the

critical

iteration

74.

displacement

Acr

prediction
Southwell

of

the

plot

is

The procedure
two consecutive

at

The
of the load factor.
value
Xcr using
the modified
load factor
in fig.
4.6.
by the flow
chart

a given

critical
illustrated

continues

X
cr
listing

of
predictions
(0-005).
A complete

tolerance

which includes
Appendix
A-4-1.

double

the

difference
between
percentage
than the
becomes less

the

until

the

of

iteration

main subroutine,
is given
in

process,

to

the linear
solve
matrix
equations
is based on making use of the sparse
nature
of the stiffness
The
matrices
on the non-zero
only.
and operating
elements
the storage
method has the advantage
of preplanning
so that
the exact
the solution
during
size of each submatrix
generated
The method

used

is

evaluated
and appropriate
before
the actual
solution
method will
now be explained

and addressed
reserved
The basic
theory
starts.
of the
in more detail.

The load-displacement
having

F=K6
equation,
can be rewritten

as,

structure

FK

11

F2K

21

KKK
nl

relationship
can be described
If this
equation

joints

is

storage

12

22-

n2.

with

the

2n

nn

the

n
For

fig.

4.7,

-i

individual

the

joints
the

K
of

part

submatrix

terms

the
of

an elastic
by the stiffness
it
is expanded

ln

Lwhere,

for

are

(4-13)

L -j
submatrices

associated

structure.

the

structure

equations

of

is shown in
which
this
part
are given

by,

75.

Fb
F

KIK
bb
K

cc

Fd

Kdd
K
KKKK
tb

Ft

tc

K submatrices
the
where,
depending
on the number of

td

ee
te

of the
degrees

ct

6c

K
dt

6d

6e

et

lK
I tt

(4-14)

above equation
of freedom
of

can be rewritten

4.14

Equation

bt

in

of a size
joints.

are
the

form.

a partitioned

as follows,

FaK

aa

=-1

--.
Ft

the

IK

6a in

calculating
eq. 4.15.

4.17,

the

following

t
are

a,
By repeating
in
of joints
displacement

analysis

(fig.

4.8),

(4-16)

ta

tt

can

the

obtained,

tt

at

6t

from

(4-17)

by substituting
be evaluated
the application
of eq. 4.16
the

reduces
analysis
can be calculated

for
and
for

until
from the

equation,

The
aa

joint

at

number
the
joint

6=F*
n

6t

I---

F*=F-KKF
aa

last

(4-15)

K-ta lKtt

K*
K-KKK
=-aa
aa

the

at,

-I -

By eliminating
following
relationships

kfter

6a

As the

(4-18)
nn
sparse
stiffness

n
matrix

Kaa is

matrix

is

replaced
symmetrical,

by the
K

at

dense
is

matrix
the

76.

transpose

of Kta so that
The elimination

of them.
that
it is only necessary
(or their
K
product).
ta

establish
elimination
imaginary

The solution
a list
of
together
connections
The order

solution.
do not include

each

the

it

to

is

sufficient
(eq.
equations
to

two

the

store

4.16

only
4.17)

store
and

matrices

one
show
1
and

tt-

to
operation
a simple
starts
with
the joints
of
at a near optimum order
including
the joint
with
connections
be created
during
the
which will
of elimination
and the connection
joints
of freedom.
with no d egrees

The elimination
to eliminate
stage,

order

is

next
degrees

the

by selecting,
one of the

performed
joint,
or

the

list

at
joints,

to
joints
freedom
the
for
of
with
i. e. the joint
the smallest
size
which it is connected,
with
K
in eq. 4.16.
The connection
for its
matrix,
connection
aa ,
in a two-dimensional
is contained
list
integer
array, MAP.
The number of degrees
joint
for a given
of freedom
m is
the

lowest

sum of

JS, while
in a one-dimensional
arrayt
another
specified
array
the sum of the number of the degrees
NM, is used to specify
to joint
of freedom
of the joints
connected
m, i. e. the size
The
stage
process.
of Kaa at the current
of the elimination
the
form,
JDF is used to specify,
in a binary
array
the
During
degrees
freedom
joint.
prelimiof
at
each
active
the
the
updating
solution
process
continues
nary mapping
A
to
last
the
the
MAP
NM
joint
of
structure.
up
and
arrays
the
includes
MAPP
list
the
of
subroutine
which
complete
integer

procedure

of

preliminary

The

subroutine

mapping

elastic
stiffness
matrix
by Jennings
and Majid
given

The
elastic
and

framed

considered

subroutine

stiffness

stability
the
contains

of

and

analysis
formulation

given
the

SPACE includes

the

order
analysis
of freedom
are

is

the

and
(67).

It

structures
at

each

in

Appendix

formulation

transformation

matrix
for first-

be used
that
provided
can

the

six

formulation

for
the
matrices
geometric
structures.
of thin-walled
the

of

degrees

joint.

BARS includes

of

A-4-2.

transformation

of

the

second-order
It also

matrix,

which

77.

threea
performing
presented
chapter
A
frames.
dimensional
complete
of
stability
analysis
j.
A.
4.
BARS is given
in Appendix
of the subroutine
is

three,

in

linear

in
arrays,
Having

respectively.

complete
stiffness
by member$in
the

list

in
load
stored
are
elements
matrix
and
WADDR
ADDR,
the working
and
namely
store,
the
the
mapping
operation,
completed

The stiffness
two

for

matrix
form of

appropriate
the
nating

addresses.
joints
one at
list
elimination

for

the

built

up. member
at the
which are entered
by elimithen proceeds

submatrices
The solution
a time

is

structure

according
eq. 4.16

to

the

previously
During
4.17.

and
using
arranged
K-1
form
the
terms
the elimination
of
stiffness
process,
tt
-1
load
Fa,
the
K
written
are
submatrices
modified
and
-K
tt
at
for
later
terms
These
to the backing
are required
store.
forces.
displacements
joint
the
and
member
of
evaluation
displacements

The joint
evaluated
performed
subroutine

using

eq.
the

using
is given

and

are
member forces
These calculations

and 4.18.
SOLVE.
subroutine

4.17
in

Appendix

A-4-4.

A complete

list

and
up
the

are
of

this

Fig-4-2Modified Southwell Plot

7-.

Fig.43 The Inner Iteration Procedure

-..

Start

job
inDut

Preliminarv

Input

joint

member data

and

Optimise

joints
of
order
of
elimination
.
load
for
Map
stiffness
storage
and
.
matrices
Input
load data to mapped locations

'0

V2
(1)
C)
0
F-4

Form

stiffness
.
mapped locations

.H
4-

a)

mode =o

b)

mode = 12,

(D

and

first-order

use
analysis
(SPACE)

or

stability

analysis

Eliminate

joint:,

Evaluate

, --Second-order

a first
order

store

second-order
or
(BARS)
use

g in

joint

Evaluate

4-3
P-1
Cd
+3
to

in

submatrices

optimum

order
I

displacements
member forces

Is
or

second
?
analysis
END

First-order
print

Fig.

4-4.

Flow
-

diagram

of

the

computer

program.

I
Set

Enter
up the

stiffness

to

internal

Set

up the

elastic

analysis

calculate

Select

elastic
r.atrix

Check

singularity

the

New load

I The displacements

forces

geometric

Kee

matrix

First-order

The

data

the

displacement

99

critical
A
cr

Is-----

factor

LO

Y=0.50

the

determinant
?

Ipositive

YES
Second-order
to

find

forces

analysis
internal

the

and displacements

NO

Select

the

Is

difference
piu entage
of
jess
than the tolerance

the

critical,

critical

A
the
?I

YES
New load

Fig.

4.5.

factor

The
flow
find
the

chart
value

of
of

the
A
cr

to
operations
computer
load factor.
at a given

I Previous
load

New load

factor

factor
Solve

for

internal

forces

an inner

iteration.

and
Perform
as that

displacements

the

shown

in

fig.

4.5

have

The

calculations
the
at
converged

given

load

factor
Prediction
the

modified

plot
(P/6-P

Xcr

of

from

Southwell

relationship)
Is
tage

this

prediction
?
the first

NO

of

the

percendifference

the

last

less
?

tolerance
Suggested

New load

two

predictions
than the

YES

YES

NO

Xcr

factor

Print

final

prediction

/END\

Fig.

4.6.

The
flow
from the

chart
of
modified

the prediction
Southwell
plot.

of

Xcr

Fig-4-7Joints Connecting Part Of The Structure

Fig-4-8 Imaginary Connections After Eliminating Joint t

78.

CHAPTER FIVE
Bimoment

Distribution

5.1.

INTRODUCTION

Thin-Walled

in

The behaviour

thin-walled

of

uniform
or nonuniform
in
for investigation

either
topic

light

cold-formed
frameworks

has

of

number

to
members subjected
has become an important

torsion

The growing
use of
years.
members in the construction
of steel
behind
the increasing
main reason

gauge
been the

recent

in

reported

studies

Members

this

subject.

'has the
to torsion
member subjected
is restr'ainedat
However,
if
such warping
longitudinal
stresses
and displacements

A thin-walled
tendency
any
will

to

warp.

section,
cross
be developed

corresponding
The
moment.

to

in

the

these

and deformation
the bimoments

stresses

of
member as a result
be a major factor
and could
The fundamental

in

lized
resulting

force

Black

to

single

arising
are of
design.

bicalled
in a thin-walled

great

importance

torsion
has been
nonuniform
(1)
(1,2,3).
Vlasov
by many authors
the term "bimoment"
as the genera-

to
corresponding
from the nonuniform

limited
was
work
cases of loading.

the

theory

and presented
established
introduce
to
the
first
was

force

The generalized
is
stresses

member.
longitudinal

of

the

longitudinal

torsion.

span

stresses
Vlasov's
However,

beams with

relatively

simple

(79)

and
experimental
presented
a comprehensive
theoretical
stresses
on the evaluation
of longitudinal
study
beams subjected
to'combined
in single-span
cold formed
channel
based
theoretical
Black's
torsion.
bending
was
analysis
and
on simplifying
ting
the St.

the
Venant

differential
torsional

equation
rigidity

of
term.

torsion
In

by neglecto
order

theoretical
the validity
procedure
examine
of the approximate
(79) compared
the corresponding
Black
the results
with
closed
the true
that
form solutions.
He proposed
value
of the bithe approximate
by multiplying
moment can be calculated
value

79.

by a correction
factor.
the value
of kZ, where
support

In
comprehensive

another

continuous
They also

k
the

is

paper,

directly

type

of

and Semple
the behaviour
torsion.

and
in his

the

of
of

procedure
Tottenham

refined
was later
(82) and applied

practical

problems.

on

loading

and

(80)
of
They

reported
thin-walled

the

the

Hardy-Cross

moment
the

for
expressions
This method was used by
different
techniques
of calculating

three-span

and made more

by Medwadowshi

The

beams.

continuous
general
(83) to

by Khan

and

a number

of

A completely
to the problem
new approach
of calculabeen
has recently
bimoments
of continuous
structures
(84).
He suggested
that,
by Davies
as the differen-

proposed
tial
equation

governing
to
identical

members is
bending,
order
The method
distribution
previously

the

torsional

the

differential

behaviour

of solution
same technique
to agree very well
has proved
with
method for a number of continuous
(81),
Khan
by
Walker
and
presented

chapter
be used to

presents

torsional

behaviour

hasbeen

can
the

of secondbe used.
bimoment

beam problems
and Tottenham

on the different
a study
distribution
the bimoment

calculate
can
thin-walled
torsion
the
by
of
nonuniform
caused
formulation
the
finite
include
element
methods
in
three
chapter
matrix
given
stiffness
elastic
The validity
element
and accuracy
of the finite
which

thin-walled

of

equation

the

This

the

depends

and provided
factors.
carry-over

study
distribution

bimoment

to

analogous

technique,

distribution
(81)
Walker

the

Black

study
concerning
beams under nonuniform

method which
distribution

ting

factor

correction
=Ej
the
GJ
V1
W
beam.

presented
for the case of a three-span
form solution
beam loaded
torque.
central
with
concentrated
the use of the bimoment-distribution
suggested

continuous
the closed

the

of

conditions

The

checked

for which highly

accurate
of problems
An experimental
study
on the longitudinal
Z-beams unrestrained
simply
supported

These

the

general
thesis.
this

of
formulation

by analysing

solutions

methods

beams.
of

(82).

a number

are

available.
in
arising

stresses
at the supports

of

against

80.

warping

bending
and
combined
longitudinal
stresses
theoretical
values.

to
subjected
The measured

and

presented.
to'the
corresponding

torsion

is

are

compared

also

METHODS USED FOR CALCULATING THE BIMOMENTS

5.2.

SinRle

span

beams

Closed-form
The differential
behaviour

a thin-walled
by,
given

is

d4o
EIW
x
dx4
in

which,

per

unit

GJ
-

m is the
length.
Equation

nonuniform

(5.1)

dx

can

to

2e
2x=m

intensity

5.1

the

describes

equation
which
beam subjected

of

torsion

solutions

of

the

torsional

external

be rewritten

in

terms

of

couple

the

bimoment

B as follows,
d

2B

dx

where,

k=

(5.2)

k2
Bm
2 -

AE 7GJ
JW ,

the

and

B is

bimoment

given

by,

d2e

B=

EIW

(5-3)

2x1
dx

The general
the form,
takes
A1 cosh

solution

kx

+A2

of

sinh

eq.

kx

A1 and A2 are constants

of
which,
conditions,
on the boundary
and loading
in

particular

solution.

5.2

as given,

by Vlasov

(5-4)

+B0

integration
and Bo is

depending
the

81.

For

supported

a simply

to

concentrated
central
Bo in eq. 5.4 is equal

subjected
solution
symmetry,
the span.

E)
xoBo
dO
x
dx

xo
x2

The angle

for

(79)

Black
GJ
dx
than

x
2

that

due
due

to

St.

to

Venant

warping

d46
EI

(5.7)
2EI

for

(.0

k2-

B and

ex can

end

conditions.

and

be

solutions

shear
bending

the

if

that

stresses
EI

d4e
W

torque
is
x,

component

much smaller
eq.

5.1

can

be

dx4

(5.8)

dx4

Equation
bending

by,

given

M'12

solution
of loading

suggested

(5.6)

+T

to,

simplified

plane

of

Simplified

5.2.1.2.

is

ex

by

B as given

Z12

904x4

sinh
3
2k
cosh

cases

other

of

a)

5.5. b)

bimoment

the

Ix

forms

Similar

half

for

dB
x

0,

twist

of

1T
EIw

d26

be found

(5-5.

solution

kx
sinh
B=T
2K- cosh kY.12

obtained

zero.
only

The general
(1) is,

Vlasov

particular
Because
of the

need
solution
general
The boundary
are,
conditions

at

and

to

the

at

Z and

length
beam with
T, the
torque

which

5.8

is

similar
the
relates

in

form

deflection

to

the
v

equation
and the

of

82.

intensity

of
yz

the

(EI

load

applied

d4v
x4

Thus

for

to central
beam subjected
concentrated
a simply
supported
B at the
torque
T, the approximate
of the bimoment
value
i)
(x
by,
is
given
=
middle
cross
section
2
B

z
2

at

app.

The approximate
9)
(x =
then
can
2

section

the

TZ

value
of
be given

T 93

= 78 EI
x app.
wj

The accuracy
of kk where
value

-t

(5-9)

0x at
by,

the

middle

(5-10)

the approximate
solution
/GxJ/EIW*
in
The error

of
k

cross

depends
the

on

approxi-

by considering
can be evaluated
of the bimoment
mate value
the approximate
between
the ratio
of the
values
and accurate
beam with
Thus, for a simple
bimoment.
central
concentrated
is,
T, the error
torque

app
-B-

value

of

correction
sional
and

kZ42
kZ/2

'

at

x=

Z/2

(81)

B
that
the value
can be
suggested
app/B
F to the approximate
factor
as a correction
He provided
for the
bimoment.
a number of graphs
F for different
factor
values
of the non-dimen-

Walker
considered

k9. cosh
2sinh

beam property
kZ.
boundary
conditions,

cross
at the middle
following
equation,
F.. B

section

app

Under
the

any

accurate
can then

case

of

torsional

of the
value
be calculated

loading
bimoment
using

the

(5-12)

83.

(81)

Walker
factor

for

the

graphs
beams and

presented
supported

simply

to

subjected
tributed

concentrated
central
either
Such graphs
torque.
can be used

5.2.2.

Continuous

This

for

design

purposes.

Beams
Bimoment-distribution

5.2.2.1.

of the correction
fixed
end beams
disor uniformly

method
in

method can be applied

known moment distribution

manner

an analogous

for
solving
used
method
well
(80)
the
Black
Semple
bending
presented
and
plane
problems.
distribimoment
factor
the
the
and
carry-over
of
expressions
Khan
beams.
for equal
and
bution
factors
continuous
span
to

the

(82)

Tottenham

the

established
the following

by defining

in

method

wore

general
for the

needed

coefficients

steps

calculations:
a)

Warping

as the
to produce
other

of

stiffness
bimoment

end

p:

which

is

one end of a simple

the
end while
at this

at

required

a unit
warping
fixed
being
fully

defined
beam

warping.

against

b)

between
the-carryratio
due
to
far
the
fixed
developed
bimoment,
end
at
over
the
the
warping
and
end,
near
at
warping
a unit
of the beam.
stiffness

c)

Distribution

Carry-over

shared
at the

factor

factor

by a joining
joint.

C:

is

and

the

is

D:

of the bimoment
ratio
bimoment
the balancing

the

member to

(82)
Tottenham
and
different
C for three

Khan
of

beam

the

the

presented
of

cases

expressions
boundary
conditions

namely,
1.

Beam fully

2.

Beam with
Beam with

3.

The warping
C

for

each

of

these

at

restrained
far end restrained
far

end

end.

against

rotation

ex only.

free.
p

stiffness
three

far

the

cases

and
are

the

given

carry-over
in

Appendix

factor
A. 5.1.

84.

After
the

continuous

a similar

both

calculating
beam the

analysis
the well

as for

manner

C for each
and
can then be carried

span

out
moment distribution

known

of
in

method.

suggested
torsional
bending,

bendinp,, technique
Analoay
5.2.2.2.
second-order
with
,
(84) who
by Davies
This method has been proposed
describing
the
that,
equation
as the differential
behaviour

is

techniques

identical

differential
governing
bending
Mz and axial

d4v

EI
Z dx4

in
y

which,
qy
direction.

is

behaviour

of

that

the

second-order
The
can be used.

of

analysis

beam subjected

dx

is

(5-13)

2=q

intensity

5.13

to

2v

of
equation
the beam can

of

to

equation
of a prismatic
by,
P is given
tension

+p

the

Equation
differential

identical

of

similar,
torsion

the

in
(eq.

be treated

load

actia-uniform

to

form,

Thus

5.1).
as the

the

in

governing
the torsional

second-order
The
load.

bending
solution
under
combined
and axial
in the form of the modified
matrix
stiffness
can be given
(84).
functions
includes
the
stability
use of conventional
which
torsional
By replacing
the bending
terms by their
corresponding
behaviour

terms

load-displacement

the
T

exl-

ei
2
91

wl

B
EI
T

relationship

becomes,

1
9,

w2

e7C

2
ei

X1
1

ex2

B2
J

91

x2

2
LJ

(5-14)

85.

T
T
torque
where,
at end 1
are the values
of the warping
wl'
w2
1-2 of the beam respectively
and 2 of the element
and Bl, B2
are the bimoment
at the two ends respectively.
values
The functions
functions
stability
These functions
are

5.2-3.

Finite

given

3'

and

second-order
in Appendix

examples,

availablp,
in chapter

were
four
Fig.

the

computer

program

then

can

matrix

stiffness

presented
thin-walled

in

three

chapter

be used to analyse
structures
the validity
to examine
In order
and accuracy
a
of torsional
problems,
solutions
element

under torsion.
of the finite
of

the

to
4 are similar
bending
analysis.
A-5.2.

method

element

The elastic
thesis
can

this

number

the

of

21

A second-order
frame
plane
to carry
out the analysis.

be used

of

19

for

correction
beams
supported

shows
factorsof

the
the

subjected
shows the

The figure
also
(81) using
Walker
(GJ
solution
= o)

to

are

accuratesolutions

the

using
thesis.

analysed
of this

5.1

highly

which

computer

program

presented

finite

calculations
element
of simply
maximum bimoment
central

correction

torque
concentrated
factors
calculated

of
T.
by

the approximate
between
as a ratio
of the bimoment
and the formal
solution
to 4.0,
the
It
of kk equal
up to a value
can be seen that,
form
two-element
with the closed
agrees
solution
very well
in
the
the
4.0,
ki
than
For
bigger
error
of
values
solution.
The
increase
kt.
the
increases
two-element
of
solution
with
four-element
closed

torque

solution,

form

analysis.
the case

eq.

5.6

however,

the
with
very well
agrees
in the
kZ values
considered

for all
the
solution
The same notations
can be drawn from
with
central
of a fixed
end beam loaded

fig.

5.2

for

concentrated

T.

The details
of the three-span
(81) is shown in fig.
5.3.
by Walker
1,,r this
namely,
solutions
problem,

beam studied
continuous
Walker
three
presented

86.

Approximate

a)

torque

solution

component

GJ
dx

b)

The bimoment

c)

The formal
These

solution,

are

distribution

method

(closed-form)

solution.

together

solutions,
shown in

fig.

Venant

St.

the

by neglecting
d2ex

5.4.

finite

the

with

can be seen

It

element
the
that

values
overestimated
and unacceptable
method gives
approximate
The bimoment distribution
method
of the bimoment.
-agree
form solution
at the
with the closed
well
values
of the bimoment
finite
The
beam.
the
interior
mid-span
of
at
supports
and
element
error

solution
in the finite

span is
interior

much closer
element

1.8

almost
support

5.5

Fig.

is

to this
solution
bending
technique

order
been

solved

load

is

formal

for
for

the
the

per

solution.
bimoment

bimoment

The
at midat the

only.

cent

beam with
continuous
(82)v
by Khan and Tottenham
(84)
Davies
presented
method.

two-span

end which was solved

distribution
bimoment

overhanging
the
using
another

while
1.7

almost

shows. the

the

solution

cent,

per
it is

to

by the analogy
secondwith
problem
(eq.
has also
This problem
5.14).

the finite
presented
program
computer
using
element
the
three
that
be
this
thesis.
It
four
in chapter
seen
can
of
the
The
bimoment
identical.
along
values
are almost
solutions
in the finite
The error
in table
5.1.
element
beam are given
loaded
is
uniformly
to
the
is
with
confined
span which
values
the
finite
torque,
in
the
solution
distributed
element
where
replaced
It
tion

that

the

matrix
stiffness
the behaviour
of

at

the

joints.

finite

element
can be used to
beams subjected

formulapredict
to

loads.
EXPERIMENTAL
TO COMBINED

5.3-1.

torques

concentrated

can be concluded

of the elastic
high accuracy

with
torsional
5.3.

by

STUDY OF COLD FORMED Z-BEAMS

BENDING

SUBJECTED

AND TORSION

Object
The purpose

of

the

tests

described

in

this

section

87.

was to

the behaviour
study
of
to combined
bending
subjected
Z-beam was tested
under three
loading.

Longitudinal

sections
of the
the corresponding
5.3.2.

fig.

Test

theoretical

cases

of

were

measured

at

results

to

of

The

span

1.10

the

Z-beam
the

of

m. as

The test
as

shown

were
from

in

section
are shown in
beam was taken
to 3.0 m.
equal
loaded
two equal
with
vertical

fig.

in

shown

between

fig.

5.3.3.

Test

to

equal

loads

tests
of three
a series
loads
test,
static
to 2.0 cm
at a distance
equal
the second test
the vertical

2.0

from

the

to 3.0 cm.
equal
During
5.7. b).
to

the bottom
(fig.
5.7-c).
web

rig

The test
end, conditions
Translational

Rotational

the
to simulate
rig
was designed
for
the beam ab shown in f ig.
5.7:
is

:ua

fixed,

ub

is

and

vb

are

fixed

wa

and

wb

are

fixed

exa

exb

and
dy)
AaA*Ab

X1

and

are

X2

are

fixed

w
w)
(dLx

and
free

following

free

dv)

, and

cLw)
(a - dx ab
:

and

Warping

was

of
first

to the web at a distance

were applied
(fig.
flange
above the level
of the bottom
(Test
the last
3) the load was applied
test
a distance

two

the

5.6.

program
consisted
5.7.
During
the

to the top flange

applied
the web (fig.
5.7-a).
In

at

with

cross

loads

flange

compared

were

stresses.

The beam was symmetrically

loads.
The distance
concentrated
equal,

combined
two cross

proaram

Details
5.6.

different

stresses
The test

beam.

Z-beams
steel
A simply
supported

cold-formed
and torsion.

are

free

are

free

Plate

5.1

Restraining

arrangements

at

the

4r
Mft
4V

St rai ii

iu L,
,s

iL

t Ion

support

88.

The

test

shown in fig.
in
x direction

5.8.

to

In

order
turntable

or..

of

consisted

rig

end

achieve

frames
support
translational
on needle

was mounted

as
freedom
roller

bearings.
In

order
light

conditions,
attached

between

and

flange

each
shown in

fig.

to

simulate

gauge steel
the upright

twisting

ILI

brackets

members of
theoretical
point

the

at
5.9.

the

Reinforcement

and warping
were firmly
the

end

frames

support

zero warping
to
were bolted
of

plates

as
each

the
in order
to prevent
web at each support
The dimensions
of the web.
are
of these plates
5.10.
The restraining
at the
shown in fig.
arrangements
5.1.
am shown in plate
support
the

side of
buckling

In
loads

were

each

the

of

to

applied

first

and third
loaded
flange

the

through
holes
in the
passing
test,
through
each hanger
was passing
to the bottom
connected
of the web.

hangers

5.3-4.

the

static

by means of steel
In the second
flange.
bracket
in
hole
a
a

Instrumentation
In

by the

tests,

order

combined

to

measure the longitudinal

bending
and torsionptwo
cross

strains
sections

caused
of the

beam were provided

strain
with electrica1
gauges.
resistance
The locations
are shown in
sections
of the two gauged cross
the
5.1l. a.
The locations
fig.
of the strain
gauges along
two

cross-sections

are

shown

in

fig.

5.1l.

b and

5.11-c.

Each

13
five
were
strain
gauges;
with
was provided
cross-section
lip.
flange,
for
to
the
three
one
per
and
each
web,
attached
Orion
Delta
to a Solartron
The strain
gauges were connected
(plate
the active
Each strain
5-4).
data logger
gauge formed
arm of a wheatstone
the logger.
into
built
load

each

5.3-5.

while

the

The measured

other
strains

three
were

arms

were

printed

at

increment.

Test results
Static

4.0

bridge

kg/hanger

increments
in
ten
beam
to
the
applied
were
40.0
kg
load
to
increment
of
per
up
a
at each
loads

plate

5.3

Loading

of

test

00%lftftlv

.4

tii:

_4.

89.

hanger.

The load
high

avoid

was kept

geometric

Fig.
compressive
the applied

5.12
strain
load.

a low

such

at

nonlinearity
test
third
where
the

strain-load
torque
caused

the

to

order

nonlinearity.
in

shows the increase

in the top flange,

the

maximum
the increase

with
that
the

can be seen
is the first
when the
case of loading
(Test
1).
The load-otrain
relationship

the

in

value

It

starts
at a very low
the load was applied
is almost
relationship
by the

the

natural
works against
the load component
in the

the

direction

of

2 shows that
In the
level.
test

of
load

the

bottom

linear.

In

to

the

of

eccentricity
twist
of

case
critical
is loaded

most
flange

top

of

flange,
this
load

applied

due

section,

cross
the major

case
to

principal

axis.
The value
Poisson's

and
of E, the modulus of elasticity,
2
beam were 190 kN/mm and
found for this
The theoretical
of the longitudinal
values

ratio,

respectively.
stresses
for the

were

calculated
bending
combined
the

moment and
element
thesis.
a)

bimoment

the

equation
(eq.
bimoment
and

using
value

were

computer
presented
program
Two types
hav
of analysis

linear

analysis,

and

b)

given
3.10).

calculated
in chapter
been

'e
second-order

0.30

by Vlasov

The bending

the finite
using
four
of this
namely,

considered,
analysis.

the
of
strains
shows
dead
1-1
the
beam
equal
weight
acting
under
cross
section
of
It can be seen that
to 40.0 kg per each of the two hangers.
the. theoretical
between
level
the general
values
of agreement
Fig.

of

the

analysis

strain

values

2-2
the

of

5.14
of the
theoretical

calculations)

the

shows
beam.

and

from

calculated

and the corresponding

the linear
Furthermore,

Fig.

order

longitudinal

the

longitudinal

good.
inaccurate

section
between

5.13

experimental
even

strain
the

a second-order
is very
values,

analysis(first

longitudinal

Excellent

at

such

to
order)leads
low load level.

strains

of

crossis found

again
(from
longitudinal
secondstrains
'
the corresponding
values.
measured
agreement

(1)

90.

It

can be concluded

that

the

finite

method
element
bimoments
for this

highly
gives
of the
accurate
predictions
type of problems.
Nevertheless,
the accuracy
of a secondbending
to combined
and
order
analysis
of Z-beams subjected
torsion

is

displacement

still
questionable.
theory
which has

The validity
been considered

of the smallin the deriva-

in
discussed
be
matrices
element
will
The effectlof
high geometric
thesis.
chapter
seven of this
has been avoided
the maximum
here by keeping
nonlinearity
load very low (40.0
kg/hanger).
of the applied
value
tion

of

the

finite

1.0-

uo-

ct
Cl

ELEMENT(2-ELEMENTS)

----FINITE

THE FINITE

-WALKER(SI)AND
ELEMENT(4-ELEMENTS
I

Q60-

cr

C)

0.40LLj
C3
E

Q20-1

0.00
IIIIIIIIII
to

zo

3.0

40

50

7D

6.0

9.0

IQO

BEAM PARAMETERA

Fig.5.1Correction Factors F For Simply Supported Beams

Subjectedto Central ConcentratedTorque
1.0_

LL.
-

0.80-

0.60'zi
Ix

cr
CD
.j

$..
-

0.40-

uJ
E
0

cia
0.2(L

Ip

zo

3.0

4.0

5.0

6.0

70

IQO
9.0
&0
BEAM PARAMETERkI

Fig-5-2Correction Factors F For Fixed End BeamsSubjected To

Central Concentrated Torque

DIMS IN MM

lmm
50.8

66
1. = 3.14x 10 mm
J= 35.46 MM4
<=0.0021 mr
.

6
mm
Jz 273
MM4
0.0046mmA

140c5.10

2mm

x 10

7-

125.4
I
sic. i-1,

S,EC.2.7
2
E=200.0 KNimm

2
G=BOA KNI mm

4-

I=
,

1.0
m

/=
1.0
,

1.0
mi

Fig.5.3 The Three-Span Continuous Beam Solved

By Walker (81)

Approximate

4
"1

Solution (JG=0.0)

-0.20
Walker

-0.15

Error At B=1.7 %
Error At C=1.8 %

Bimoment Distrip-Method
Solution

__Formal
Finite

Element Sotution

10 Elements I Span

-0.10

/ /4'
-0.05
A

f/

/":
i/

_k

-..

-.

-QO5

Fig.5-4The Distri6ution Of'The Bimoment As Calculated By

Walker (81)And By The Finite Element Method

J-G
il(o

m 0.74271m

IOOKN.mlm
.
2.0

2.0

2.0

320 KN.m

B=100KN. M2

Ox=0

gx=O

1.S

1.5

1.5

II
bcdef9

(I

ii

Fig-55 Bimoment Diagram Calculated By Three MethodsOf Analysis,Namely,

(a) The Bimoment Distribution (82),The Analogy With Second-Order Bending (84),
And The Finite Element

LIC 1124
r)yozo"'?
L

CL

d e

Khan/Tottenham 376.0 - 483 -129.2 -7 0.0 279.4 3 2.0 -17 2.2-2 5.6 BS.4
( 82)

71.7 100.0

375.9 -50.1 -129.2 -70.8 279.5 32.0 -1716 -25.5 BS.5

131.7 1717 -25S SS.6
2783
-131.4 -72.6

72.0 100.0

Davies (84)
Finite

Element

729

100.0

5.5
Fig.
Shown
Beam
The
Of
Values
51
Bimoment
Table

y
45.0
150

DIMS IN MM

t= 1.60
pp

095M

0.95m

15.0
45.0

Fig-5.6The Tested Beam And Its Cross Section

y
lle

11

Fi(AYTest
1
ey=6 0.0

;- 60.0

ez zl. 20

( C) Test(3)

(B) Test (2)

Fig.5.7 Positions Of The Applied Loads

In The Three Tests

(1)
1

12 mm t Shaft In Roller

Bearings

To

Allow For Rotation, About z-Axis

le To
About

Fig.5.8 The End Support's Frame

7%, Cor4ti C21

Point

Of Zero

W2rpinq

at
V Cross Section
Ptan,
Fig.5-9 The Light Gauge Steel (L), Brackets Used To Simulate
The Torsional And Warping Conditions

50.0

'TLn
Plate
Reinforcement
The
Of
Fig.5.10The Dimensions
Dims IN MM

( CL)

;i

t1
-

v
,

j:! Ocm

cm:

,110.
_j2.
Ocm

9%0

i5i 17

515 4-7

f1l _71 r6- -lTq

5
23

18

Z.

m-

9S.0 cm

:iII

iSi 1 i 17'_4_.

14

19

30

DIMSAN mm

. #-20
30
.--.

L 21

23

26
t2Ln2
4

io

i5i 175 175

1
(C) STRAINGAUGESAT
SEC. 2-2

+1
GAUGtS AT

(b) STRAIN
SEC. 1-1

F19.5.11The Positions Of The Strain Gagges

For Sec.1
And 2-2
-1

Kg/Hanger

(1)

0.00 v
40
W.
Vu

so

10101110101
120
160

200

240

280

320
MICROSTRA

Fig.5.12MaximumCompressiveStrain In The Top Flange

Scale

Fig. 5.13 Longitudinal Strains At Cross Section I-1

Under Static Load =40 Kg/Hanger

EXPERIMENTAL
THEORETICALLINEAR

_-THEORETICAL

NONLINEAR

MICROSTRAIN
400

-300
-200
100

0.0
Scale Of Longitudinal strain

2
Section
Cross
AT
Strains
Fig.514 Longitudinal
-2
Under Static Load= 40 Kg / kanger

91.

CHAPTER SIX

Application

Element

Finite

the

of

Method

to

Buckling

Problems

GENERAL
The derivation
of

of the
torsional-flexural

the

elastic
has
structures

of the elastic
discussed.
It
the

validity

been

in

given

new finite
buckling

chapter'3

formulation
element
of thin-walled

where

and geometric
stiffness
is the purpose
of this
of the finite
and accuracy

the

matrices
chapter

to

for a variety
by presenting
of problems
solutions
by alternative
solutions
accurate
or highly
exact
herein
is'-limited
The study
reported
available.
in

described

element
to predict

displacements
no torsional
The finite
bimoments).
external
in

chapter

The chapter
number
validity
employed
buckling
of

the

as will
6.2.
.

of

4 was used
begins

the

prebuckling

the

computer
it
then

by presenting

program

for

for

means are
to problems
(no
stage

to
solutions
to illustrate

evaluating

CONVENTIONAL STABILITY

a
the

is
which
the critical
the accuracy

to establish
proceeds
formulation
more complex
when used to analyse
in the following
be discussed
sections.
loads.

which

program
computer
loads.
the buckling

conventional
stability
problems
I'Southwell
technique"
of the modified
in

demonstrate
formulation

element

with

main fea tures

have been

problems

PROBLEMS

in
this
discussed
section.
are
separate
examples
is presented
In each case the solutiong;
vehP4 finite
element
(2).
Timoshenko
by
the
of
exact
as a ratio
solution
given
Four

Pure

torsional

The pure

torsional

buckling
buckling

of

an axially

loaded

',

Under
is considered
here.
section
with a cru'Ciform cross
the strut
displacements
such loading
only.
exhibits
angular
(2)
The exact
is given
by,
Timoshenko
by
solution
presented
strut

92.

.p=L(+
cr

(6.1)

GJ)

12

6.1

Fig.

of
shows the distribution
of the angle
twist
ex just
to buckling
by the two-element
prior
given
that
the twoIt can be seen from the table
solution.
given
the exact
has an excellent
element
solution
with
agreement
6.2 shows the modified
Fig.
from
Southwell
solution.
plot
the

which

buckling
In

predicted.
6.2.2.

400

solution

for

are

0.17

modified
shown
6.2-3.

of

a simple

bending

beam by uniform

length
I-beam
with
supported
shows a simply
The exact
by uniform
bending
cm loaded
moment MZ.
(2)
Timoshenko
is,
this
by
of
case as given

It

while

buckling

been

6.3

/ll
EI

m
cr

solution

of the eight-element
case has
line.
is a straight
case the plot

this

Lateral
Fig.

Z=

load

v'

(GJ + EI
yw

2
l-)
2
91

can be seen from fig.

0.66 per
differSby
only
four
and eight-element,

and 0.13
Southwell

6.3
cent
the

that
from

the

two-element

the

finite

respectively.
cent in error,
for the eight-element
plot

solution

exact

element

per

solutions
The

solution

is

6.4.

by fig.
Lateral

bucklinp

concentrated
6.5

Fig.

400 cm loaded
=
-Z
the shear center.
is,

er

(6.2)

Yl

of

a simple

beam by central

load
length
I-beam with
supported
shows a simply
load Py acting
by central
concentrated
at
(2)
by
Timoshenko
The exact
solution
given

VIEIV y

GJ
(6-3)

93.

in

yj is
which,
2
(Z
ratio
. GJ/EI

a dimensionless
Values
of
W

factor
yj

are

which
tabulated

depends
in

upon

the
2.

reference

The results
of the finite
element
calculations
are given
in fig,. 6-5- It can be seen that
there
is an excellent
agreement
between
the finite
of
solutions
value
element
and the exact
M
The lateral
displacement
of the middle
w
cr'
to buckling
is also
of the beam just
prior
shown
6.6
Southwell
fig.
while
shows the modified
plot
element

solutions.

6.2.4.

Lateral

buckling

y2 is
where,
2
(Y' . GJ/EI
W).
Fig.

the

VIEI v2
12
91

by

was presented

case

GJ

(6-4)

factor
on the
a dimensionless
which depends
2.
in reference
Values
of Y2 are tabulated
6.7

shows

eight-

beam by
a cantilever
end
at the free

The exact
solution
of this
(2),
by,
Timoshenko
and is given

Mcr

for

of

load

concentrated

section
cross
6.5,
in fig.

the

finite

element

solutions,

for

ratio

this

The finite
case.
for the buckling

overestimated
values
solutions
give
element
load.
The error
changes from 2.59 per cent
to 1.2 per cent for the eightfor the two-element
solution
The lateral
displacement
prior,
w just
solution.
element
6.8
6.7,
iliustrates
fig.
to buckling
is shown in fig.
while
the

three

predict

operations
out by the
carried
the eight-element
solution.

computer

program

to

that
the
examples
can be seen from the previous
solutions,
element
when compared, to
of the finite
accuracy
The procedure
followed
to
is excellent.
the exact
solution,
load,
is known as "the
the buckling
modified
predict
which
It

Southwell
forward

has
plot"
technique.

to. be a very efficient

proved
and straight
One major advantage
of the computer
the displacement
that
it
so that
calculates

program
used is
the general
shape

of

the

buckling

mode can be observed.

94.

6.3.

ELASTIC
This

out

the

using

LATERAL BUCKLING OF CONTINUOUS BEAMS

section
finite

the

presents

of

results

a study

carried

program
computer
given
in
4 in order
to examine the validity
chapter
and accuracy
the method when used to analyse
the torsional-flexural
buckling
beams.
of continuous
The effect
interior

of the elastic-lateral
on the elastic-lateral

supports
beams has

continuous

been

investigated

Solutions
method.
(54).
by Hartmann

element
studied

part

of

The

second
buckling
of

lateral

the

solutions
by Trahair

buckling

account

the

is

the

stiffness,

and

stiffness,
in the

study.

shear

center

and of lateral
Three types
of

continuity
load.

c)

axial

weak-axis
All
loads

of

the

stiffness,
flexural

were
beam.

tested
section.

beams

continuous

a)

in

the elastic
deals
section
with
loaded
beams with
spans
all
buckling.
Finite
element

study
presented
a theoretical
beams taking
behaviour
of continuous

namely,

problems
finite

presented

(54)

of

considered
through
the

the

been

beams previously
continuous
in the second part
of this

restrained

effects
on the critical

flexural

this

of

spans
for a series
of
(51) are given

elastic

the

using

of

at the
load of
finite

buckling

procedure

continuous
during
interact

Hartmann

bracing

for
obtained
The comparison
between

part

Elasticallv

stiffness
bracing

have

and Hartmann's
this
section.

solutions

element
the first

where

element

of the
into

bracing
lateral
b)

strong-axis

stiffness,
were
assumed to be acting

(54)
Hartmann
by
The analytical
procedure
given
was
the beam into
based on dividing
a number of segments
n, where
the ith
is a straight
i
of the beam between
any segment
part
loads,
joints.
Applied
+1)th
and elastic
and the(i
reactions
the
joints
restraints
occur
at the. specified
connecting
(54) presented
Hartmann
the differential
segments.
equations
which

describe

continuity

the

equations

buckling
at

joints

behaviour
i

of segment
+ 1) in
and (i

and the
terms
of

95.

the

elastic

The

solution

equations
ting
the
is

representing
bracing,
Hartmann
y suggested

coefficient

Ku. Z3
48 EI

.1--

and
Z is

in

which,
the length

inertia

to

the

of

Ku is

joints.

was based
equations
of the differential

on

and construcThis deter-

conditions
equations.
criterion.

(6-5)
y

the

axial
beam,

of the
beam cross

the

the

at

Ku of the
the axial
stiffness
(54) used the nondimensional
(45),
by Flint
where,

The strong
axis
torsional
stiffness

stiffness
and Iy is

of
the

the

bracing,

lateral

least

moment

of

section.
flexural
of

the

Sx was also
related
stiffness
beam using
the nondimensional

6 where,

coefficient

X-9,
GJ

6.9

Fig.
problems
the finite

bracing

buckling

as the

considered
For

lateral

lateral

integrations
numerical
for a given
set of boundary
determinant
of the resulting

performing

minant

of the
differential

stiffness
of the

shows the three

investigated
by Hartmann
for
solutions
element
the interior
supports

types
(54).
these

beam
of continuous
to present
In order
cases the lateral

by beam members
For
I
having
moment of inertia
an area A br and strong
axis
br*
y the cross
a given
of the nondimensional
coefficient
value
by,
is
the
Abr
bracing
given
of
sectional
member
area
bracing

at

in

48 Y'Iy'p'br
br

k br is the

which,
t

was simulated

(6.7)

Z3

length

of

the

lateral

bracing

member.

96.

For

assumed

The

br

critical

-GJ
"2 4EZ

buckling

load

cr

in

which,

of 6 the
becomes,

a given
value
bracing
member I br

a is

the

Hartmann

moment of

inertia

(6.8)

br

cr

is

by,

given

(6.9)

critical
(54)

the

of

load

parameter.
different

considered

values

the

of

beam property
kZ
k=V
GJ/EIw
where
-He
kt,
for all
the values
of y
using
a value
of
lateral
to 11.0ithe
displacement
at the interior
equal
to about 2 per cent of the maximum lateral
support
was reduced
types
displacement
Thus in each of the three
of
of the beam.
nondimensional
that
reported

problems
considered
to
was taken
equal

in

the

(fig.

study

6.9)

the

coefficient

11.0.

6.1
load
Table
the
the
of
critical
presents
values
,
(4
by
the
finite
method
elements/
a calculated
parameter
element
6.9. a, where one of the two
span) for the case shown in fig.
beam is loaded
with
concentraspans of the continuous
central
by
The table
the
ted load P.
a
of
given
also
values
shows
Hartmann
ted
is

using

(54)

and the
Salvadorits

an excellent
(54)
Hartmann

lower-bound

corresponding
(48).,
method,

between
agreement
and the corresponding

It

values
can be seen that

calculathere

by
of a given
values
finite
values.
element
finite
solutions
element

The percentage
difference
of the
compared to Hartmann's
calculations
is shown in fig.
and Sx* 9'IGJ = 370

the

for
6.10.

the

case when kk =2
For the four-element

the difference
is about
0.19 per cent while
solution
the number of elements
to which each span is divided
in 0.09 per cent difference
only.
elements
results

increasing
to

eight

97.

the
in

The effect
lateral
bracing
6.11

fig.

the

strong

on the
different

critical

of

flexural
of
stiffness
axis
load parameter
a isshown
U.

can be seen
that
the general
level
between Hartmann's
method
of agreement
low
However,
for
is
and the finite
element
solutions
good.
Sx*'/GJ
values
of the nondimensional
some differences
property
between
The difference
the two solutions
can be noticed.
for

depends

value

of

equal
(narrow

also on the
Sx* k/GJ is taken
kX=

a beam having

for kk=
almost
zero,
while
increases
to 2.6 per cent.
(54) about the approximate
tial

equation
representing
Nethercot
beam.
However,
Hartmann's

is

method

values

of

It

For example
of U.
to 50 the percentage

of
solution
the torsional
(36)

dependent

showed
on the

the

value

difference
section)
difference

rectangular
cross
2.0 the percentage
No details

if

for
is

by Hartmann
were given
differenthe governing
behaviour
of the
the

that
value

accuracy
of U.

of

6.12

between
the finite
shows the comparison
load parameter
a and the
element
solution
of the critical
(54)
Hartmann
for the case shown iIn
by
of a given
values
6.9. b. when each of the two spans of the continuous
beam
fig.
Fig.

is

loaded

load P.
For beams with
concentrated
(kZ
'
Hartmannt
s method
= -) ,
section
For
finite
the
method.
element
with
Hartmann's
to 4.0,
value
equal

a central

with

narrow
cross
rectangular
is in excellent
agreement
kk
beams
continuous
with
method

overestimated

gives

values

of

the

load

critical

a.

parameter

6.2

load
of the critical
6.9.
for
the
c.,
parameter
where the
a
case shown
beam is loaded
at
central
span of the three-span
continuous
load
P.
It
the middle
transverse
can be
concentrated
with
a
for
to
the
finite
solutions,
seen that,
element
compared
Table

high

values

presents

the

values
in fig.

of the nondimensional
Hartmann's
method is

accuracy
of
high values
with

of

the

warping

coefficient
very
rigidity.

good

Sx* 2/GJ-* the

e,en for

beams

98.

6.3.2.

Interaction

buckling

of

continuous

Trahair

a series
presented
the effect
of
concerning
buckling
torsional-flexural

(6,49,50)
elastic
elements.
beam with

He showed
only

one span

the

the

major

the

can be

the

unloaded

3EI
(dy)
dx B*

m
B

of

loaded

continuous
MB is given

theoretical

studies
on the

end restraints

buckling

two-span

bending

axis

the

of

effect

restraining
loaded
span.
For

that

of

beams

beam
symmetrical
load of a continuous
by considering
3alculated

beam shown
by,

on the

spans

adjacent

in

z2

6.13

fig.

(6.10)

'2

The effect
of the major
MB can be given
by,

bending

restraint

on the

bending

moment

3EIZ,
(dv)
dx B"

mB

in

is

which,

The parameter

z1

the

major

is

given

axis
by,

bending

end restrain

1
1s

where,

Following
restraint
4 can

zi

Z1
Izl
=k1

/S

(6.12)
z2

and
,

z2

and

z2
k2

flexural
the
minor
axis
procedure,
same
2 and the end warping
restraint
parameter
by the two equations,
be given
the

21+s

41+

yl

/S

hloSyl

parameter.

(6.13)

end
parameter

(6.14
y2
/h

(6.15)
2 -S y2

99.

s=IIsI
yl

where,

and h1,
flanges
If

the

Y'l

h2 are
of the

y2

the

=Z2

distances
sections
is
section

in

which,

cr":

12
v/i7;
91

y is

the

Trahair
different

load

buckling

critical

load
,,,

parameters

of the
centroids
of span AB and BC respectively.
used for both spans then,
the

1
91
212,1

4 =1+

The

(6.16)

between

cross

same cross

y2

Pcr

(6.17)

is

by,

given

-uj-

(6.18)

load

critical

parameter.

(6)

for
tables
presented
values
cases and different
2' and
4*

the
of

for
y values
the end restraint

(50)

the case when

discussed
another
paper
the spans of the continuous
beam are loaded
at the same
all
buckling
the
in
time.
In this
span
participates
case each
behaviour
the adjacent
spans.
of the beam by interacting
with
Trahair

In

The

evaluated
vidual

load

buckling

critical

the
by analysing
6.14
Fig.
spans.
shows

rectangular

continuous

two-span
narrow
a uniform
beam where the two spans are loaded
with
The beam
loads
P1 and P2 respectively.

concentrated
central
is prevented
from twisting
The major
spans

beam can then be

of the continuous
behaviour
interaction
of the indi-

flexural
axis
by,
can be given

and deflecting
end restraint

2/p
1+pk
2211
T-T Y,
2/Zl
1+pk2
12

1+k

/P
1122
1/Z2

at

the

parameters

supports.
of the two

z2
(6.19)

12

(6.20)

100.

support

To satisfy
the
B in the lateral

parameters

21'

1-

where, y2l

lcr

(6.21)

loads

and P 2cr

lcr

are

V1EIYj

and 122 are

the

GJ

and

critical

2cr

load

by,

given

Y22

k2
12

interior

axis
end restraint
as follows:

Y21
p

the

z2 /xi

21(l

buckling

critical

at

condition
the minor

be related

must

22'

21
-

22

The

and

continuity
direction

-I

2-VEI
P,

y2-

GJ

(6.22)

parameters.

P
the
and
given
2/pl
21l
be calculated

parameters
and
can
major axis
restraint
11
12
By selecting
the proper
from eq. 6.19 and 6.20.
of
values
0

the values
the minor axis
parameters
and
restraint
22'
21
by Trahair
from graphs
given
of '22 and Y21 can be evaluated
For

of

values

6.15

the

ratios

load
combinations
of the critical
Al
by Trahair
Z2
for
presented
span ratio
= 5.0,
parameters
there
that
It
significant
combinations
are three
can 'be noted
as
Y22 and Y2 1 and these
can be stated
of the parameters
Fig.

shows

follows:
(i)

and P2 ' 0

P1>o,
in

this

case,
- 01-,' 11

Y22/y2l

The minor

211

(50).

-+Z

axis

1
79,
21

1+Y,

restraint

2/Yll

and

parameters

22 -: - Co

' '*
12 '2

(6.23)

are,

(6.24)

(50).

101.

P2 >0,
in

and P0

this

case,

y2l/Y22

The minor

0=21

(iii)

axis

00 9

Zero

and 12 =1+1z

11

parameters

restraint

122 =1+Z1

and

It

(6.25)

/Z
12

are,

(6.26)
2

interaction

parameters
case the minor axis
restraint
betwe. en
interaction
is no minor axis
a2l ' 22 'm o and there
a
The major axis end restraint
the two spans.
parameters
ill
6.20.
6.19
The
from

be
critical
and
eq.
and
calculated
can
12
from the
load parameterp
121'
and Y22 can then be evaluated
(50).
Trahair
by
graphs
given
In

this

diagram

The interaction
combinations
6.15.
in fig.

of

the

critical
(50)
Trahair

between

load

has

by linear

be approximated

analytical
experiments
beams under
loads.
laterally
evaluated
modified

significant

relations
by the dotted

as shown
shown indicates

that

he

correct.

was essentially
In

three

a curved shape as shown

that
the interaction

suggested

curve
can sufficiently
between the three
significant
points
6.15.
The comparison
lines
in fig.

these

order

to

examine

procedurepTrahair
on high strength
different

Each
at
from

the
(51)

carried

aluminium

combinations

of

from
beam was prevented
Experimental
the supports.
the

Southwell

load
plot

the calculated
general,
loads
than
were higher
4.0 per cent.

and accuracy

validity

of

of
out a series
I-section
continuous

central
twisting

concentrated
and deflecting

critical

loads

were
the
using

measurements
and deflection
(51)
Trahair
showed that,
method.
critical
of the experimental
values
the

measured

the

failure

loads

by almost

in

102.

6.16

Fig.
(A-20.20.20.2)

three-span
(51) which

the

shows

beam

continuous
have been

by
analysed
of reference
from
beam
The
prevented
the finite
was
method.
element
Central
the
laterally
deflecting
supports.
twisting
at
and
level
the
beam
to
the
of
loads
at
transverse
were applied
(right
left)
two
The
were
flange.
and
the top
spans
outer
the
P1
load
the
span was
middle
loaded
while
same
almost
with
loaded

In
additional
above the
section,
Ph/2

load

with

2*
to

order
effect

shear
has to

section

being
h

where
center,
be considered.
to

corresponding

cross

perform
the load

of

of

each

finite

the

at
applied
is the height

An additional
degree
of freedom

the
span

included

been

has

the
h
2
the cross
to
equal

analysis,
a distance

element

of
term

ex at the loaded
in the geometric

matrix.
Table

6.3

between

the

comparison
shows
(4
the
and
elements/span)
solutions
Trahair
by
load
the
given
critical
of

element

values
three-span
that

in

the

finite

continuous
the
general

loads

critical
failure

loads
Fig.

element
accuracy
elements.
per

cent

6.17

the

can

corresponding
Trahair
before,
experimental
his predicted

of. the finite

error
6.3.
High
in table
each

experimental
solution

eight-element

be seen
by
calculated

It

cent.

by dividing
the

the

span

into

value
differs

four
of the
by
-0.9

only.
Fig.

6.18

span continuous
techniques,
which
(a)

extrapolated
than
higher

shows the percentage

test
the seventh
for
to

the

than

as mentioned

generally
were
4.0 per
by almost

solutions
can be achieved
In comparison
load,

critical

less

finite

experimental
(51),
for

6.16.
beam shown in fig.
load
the
buckling
of
values

method are
element
However,
values.
experimental
the
(51) has pointed
out that

the

shows
beam shown

Extrapolating
using

the

interaction
6.16,
in fig.

the

curve

for

calculated

the

three-

by three

are:
from
modified

the

experimental
Southwell
plot

measurements
method.

103.

(b)

The finite

element

method

four

using

elements

per

span.
(c)

(50),
Trahair
The approximate
by
method given
is based on analysing
the interaction
effect
individual
Uhe

by Trahair
method for
(when P1=0,

For

finite

is

more

when P2 = 0,
the rest
of the

P2>0,

interaction).
element

results

spans.

the approximate
method
can be seen that
(50) is in good agreement
the finite
with
the three
of the interaction
major points

It

while

in

the

which
between

close

P1>0,
interaction

agreement

element
curve
zero

curve,

the

the

with

method

approximate

and at

presented

given

experimental
(50)
by Trahai

is

conservative.
6.19

lateral
the
of
shows
for the case when
displacement
stage
w at a near buckling
loaded.
It
beam
the
the three
almost
are
equally
of
spans
the
buckling
in
this
be
that
outer
at
starts
case
can
seen
Fig.

spans

which

a similar

the

are
distribution

most

6.21

shows

Fig.
computer

program

calculated

the

values

critical.

as

shown
the

ELASTIC

four

the
evaluate
of the lateral

to

of one outer
cross
section
of the load factor.
6.4.

distribution

span

in

The angle
6.20.
fig.

steps
buckling

of

carried
load

displacement
and the

ex has

twist

by the

out
from
w

corresponding

the

at

the

middle

values

LATERAL BUCKLING OF MONOSYMMETRIC BEAMS

AND CANTILEVERS
The effect
The differential
describe

the

uniform
as follows,

with

buckling
bending

of

monosymmetry
equations
behaviour
of

of

equilibrium

which
beam loaded

a monosymmetric
by
Timoshenko
Mz
presented
moment
were

d46

d4w
EI -x4 - Mz
Y
dx4x

` 0

(6.27)

(2)

104.

d4E)
xEI
w dx4

in

0z

which,

For
z

is

given

(GJ

is

-X7

-M)d26x-md2w0
Z* Z dx 2Z

the

monosymmetric

constant.

the

in

shown

section
cross
by the equation,

Y3 dA +

(6.28)

Z2 y dA)

6.22

fig.

the

constant

(6.29)

2yo
-

zAA

of

the

from the effect

results
on the buckled
cross
bending
and tensile
compressive

The monosymmetric
property
bending
when acting
stresses
on the slightly
For
torque.

stresses
acting
form a resultant
torque
resultant
(,
zero
= o).

section
may
I-beams'
the

rotated
cross
doubly
symmetric

stresses
caused by the bending
beams, however,
In monosymmetric

bending
stresses
Oz 4 o).
Such

that

and

is

to
equal
there
is

torque

the

between

nobalance

the

buckling

During

section.

caused by the compressive

due to the bending
tensile

causes

effect

a change

in

the

effective
be seen

stresses
tor-

the
in
of
cross
section
as can
sional
stiffness
the critical
is such that
The effect
of monosymmetry
eq. 6.28.
bending
load is larger
stresses
acts
on the
when the tensile
of the cross
section.
smaller
of the two flanges
For
with
uniform
M
is given
cr

Tr VIEIY

er

21
27T

where,

supported
moment Mz,

a simply
bending
by (40),

=V//T

jJ. .z

to

I-beam
monosymmetric
the elastic
buckling

loaded
moment

GJ
7r8

7-

Z2

7r6%2
+. (n6)2

(6-30)

(6-31)

/EI
and

7.

2/

GJ

(6-32)

105.

1972,

In

theoretical

comprehensive
the buckling
under
based

Anderson

flexural
simply
P, the

If

intensity
(40)
v

is

u,

symmetric

load

critical

parameter.

load is uniformly
acting
the critical
of the
value

is

'y 3
' 73

the

Anderson
Y3 values
to verify

the

(6-33)

the

er

carried
behaviour

the
solve
the torsional-

to
method
which describe

'y 2 /EP
I_GJ
2Vy
91

Y2

which,

where,

integral

finite

equations
buckling
For
behaviour
beams.
of monosymmetric
load
beams loaded with
supported
concentrated
central
(40)
load
the
P
is
by,
value
critical
of
given
cr

cr

in

study
concerning
beams and cantilevers
of monosymmetric
The theoretical
analysis
was
cases.

loading

the
on using
differential
governing

presented

and experimental

behaviour

different

(40)

and Trahair

V//iI

distributed
load

with
be given

can

G
J

by

(6-34)

load

critical

parameter.
(40)

and Trahair

presented

tables

for

y2 and

of
In

beams and cantilevers.

order
monosymmetric
(40)
Trahair
Anderson
the theoretical
and
method
investigation
on the buckling
out an experimental
of high
I-cross

transverse
end with
the shear center.

strength
section.

mono-

with
cantilevers
aluminium
Each cantilever
was loaded

concentrated

load

at

at

the
from

a distance

The effect
in a monosymmetric
of moment gradient
beam was not included
in any of the previous
studies
which
for the elastic
formulation
presented
a finite
element
(24,26,28).
torsional-flexural
thin-walled
buckling
of
systems
The new geometric
stiffness
this
includes,
thesis
for

matrix
the first

presented
the
time,

in

chapter

effect

of

3 of

free

106.

This

in the geometric
effect
appears,
torsional
in the effective
stiffness
bending
stresses.

monosymmetry.
as 'a reduction
of

the

acting
This

the

validity

the

elastic

presents

section

of
and accuracy
lateral
buckling
Three

cantilevers.

types

of

a study
the finite
loads

of

problems

matrix,
in terms

to

undertaken

examine

element

evaluation
of
beams and
monosymmetric
have been studied,

namely:
a)

b)

c)

Lateral

buckling

I-beams

under

Lateral

buckling

I-beams

under

Lateral

buckling

loaded

of

of
free

supported

simply

concentrated

central

the

at

monosymmetric
supported
simply
bending
moment.
uniform
of

monosymmetric
load.

I-cantilevers
monosymmetric
transverse
concentrated
end with

load.
Two types
this

in

considered
(i)

A cross

(ii)

section

6 are

K and

where,

section

A cross

The finite
have

been

cross

monosymmetric
and these
study

of

given

have

sections

are:

with

R=0.10

and

6=-0.10

with

R=1.0

and

6=-0.60

by

element
the
with

6.31

eq.

been

and

of

solutions

6.32

respectively.

cases

a,

b and
of the

c
buckling

corresponding
values
by
Anderson
theoretical
the
loads
method
given
using
evaluated
formu(40).
The
the
finite
Trahair
of
element
validity
and
lation

compared

has

lever
results

also

problems
of the

6.4.2.

Simply

the

section
critical
be seen

tested

by analysing
two
checked
by Anderson
and Trahair

finite

simply
having

supported
R=0.10

bending

element

supported

6-4. a.

Table
for

been

study

beams

presents

the

uniform

finite

using

cantiThe

now be discussed.

will

monosymmetric
and 6=-0.10,

moment evaluated
there
is an excellent

that

by

of the
(40).

moment

element
beam with
as ratios
6.30.
eq.

agreement

solutions
a cross
of the

between

It

can
the

107.

finite
eq.
cent
per

cent

beam with
form

closed

6.4.1 3.

Simply

is

in

solution
element

the

finite

0.44

error
the
for

are

simply
in

given

agrees

very

per
of 0.01

well

with

beam.

beams

supported

by

of

an

gives

element

this

of

solution

solution

error

solutions
(ii)
type

section

cross

Again,

b.

form

closed

solution

eight-element
The finite

only.

the

and

two-element

the

while

supported
6.4.
table
the

solutions

element
6.30.
The

under

central

load

concentrated

A closed
form
for
this
solution
case is not available.
The results
of the finite
element
analysis
are given
as ratios
buckling
load
6.33,
of the critical
calculated
using
eq.
where
the load
from
parameter
72 has been taken
the tables
by
given
(40).
Anderson
and Trahair

6-5. a. presents
the finite
element
solutions
for the beam with

(i)
(!
cross
type
section
= 0.10,6
=-0.10)
6-5.
table
b. presents
while
the results
of the cross
section
(K=1.0,6
type (ii)
It can be seen that
for=-0.60).
'
both examples
the agreement
between the two methods is very
the convergence
good, nevertheless,
of the finite
element
Table

solutions

is

sectional

properties

6.4.4.

slightly

Cantilevers
load
Finite

at

the

analytical
level
of

loaded

transverse

free

procedure

for

Comparison

an experimental

type

of

cross-

concentrated

end

given
between

of
with

this

in
given
(i)
type
section

case

a cross
The results

(ii).

type

are

are

load

calculated
by Anderson
and Trahair

the

two

methods

is

given
the
using
(40).

excellent

as
The

as it

tables.

between

and the experimental

Anderson

second

=-0.60).

with

buckling

agreement
can be seen from the

6.4.5.

the

element
solutions
for the cantilever

6.6. a.
6.6.
in
table
b.
and
the
of
critical
ratios
table

for
slower
(R = 1.0,6

and Trahair
study on the

the finite

element

solutions

results
(40)

presented
torsional-flexural

the

results
buckling

of
of

108.

I-cantilevers.

monosymmetric

the

The study
theoretical

was undertaken
procedure

of
obtain
a verification
the elastic
(40) for
the two authors
calculating
beams
loads
and cantilevers.
monosymmetric
of
two

to

given

by

critical

the

cantilevers
solutions
6.7.
in
table
Trahair
tested
are presented
65.0
inches
had a length
Each of the two cantilevers
and
of
1GJ
-R(-R2
EI
to
Z2)
equal
Tr2
=
property,
a cross
sectional
W
VEI
/GJ)/Z)
the
two
7
for
6(6
0.475,
the
of
=
values
while
y
,
0.18
0.18
+
to
respectively.
and
cantilevers
were equal
to each cantilever
along
The concentrated
load was applied
distance
the
free
a
at
end
the
at
section
of cross
y axis
0.027
to
taken
equal
from the shear center
a was
where
a
Finite

of

element
by Anderson
and

and

0.22

inches

of

respectively.

are
element
analysis
of the finite
loads.
The
buckling
given
of the experimental
as ratios
6.7
table
in
correlation
shows an excellent
comparison
given
between
the finite
values
method and the experimental
element
6.7
be
from
table
It
load.
seen
of the buckling
can also
The results

that
(40)

the

method given
is not at the same level
of agreement
their
in
The
experimental
error
results.
the absolute
increase
of the distance
with
It
formulation

can be concluded
in chapter
given

and Trahair
the
with

by Anderson

theoretical

that
three

the
of

method

increases

a.

new finite
thesis
this

element
can be

buckling
the torsional-flexural
of monoanalyse
Compared to closed
form
beams and cantilevers.
symmetric
integral
finite
and test
results,
Solutions,
solutions
formulation
has
the accuracy
finite
the
element
of
new

used

proved

to

to

be excellent.

109.

6.5.

BUCKLING OF PLANE FRAMES

TORSIONAL-FLEXURAL

this
in
The study
was undertaken
section
reported
to examine
element
the validity
of the finite
and accuracy
to
3
in
carry
out a
formulation
used
when
chapter
presented
The
frames.
three-dimensional
of
plane
analysis
stability
to
4
used
evaluate
was
chapter
computer
presented
program
for
frame
the critical
problems
loads
of
plane
of a number
available.
are already
results
which
experimental
in

this

The first
element
tically

of
part
for the lateral

solutions

the

presents
section
loads
buckling

frames

in-plane

loaded

with
both experimentally
Trahair
and

investigated
Vacharajittiphan
finite
results
presented

buckling

interaction

element

and
solutions
by Vacharajittiphan

given
in the

second

part

re_ctanaular

the-elastic

of

plane

(53)

Hartmann

this

of

the
In

procedure,
tion
on the

lateral

order
Hartmann

to

I-plane
was

by
between

the

and theoretical
(60) is

section.
of

restrained

elastically

frames

method

of elastically
The
frames.

plane
single-bay
single-story
integrations
been based on performing
numerical
include
differential
which
equations
governing
effect

obtain
(53).

loads

transverse

a theoretical

trained

to

elasThese

symmetric

experimental
and Trahair

presented
buckling
analysis

lateral

doubly

and theoretically
(60).
A comparison

buckling

Torsional-flexural
narrow

of

concentrated

the

two

of
frames.

plane
rectangular
narrow
restrained,
(55)
in
Hartmann
by
order
tested
two frames
were
to his theoretical
method of analysis
verification
The

finite

of
the

for
res-

method
the

has

restraining

bracing.
the

examine
(55) conducted

accuracy

of

the

an experimental

theoretical
investiga-

buckling
lateral
of narrow
rectangular
elastic
tests
Twenty
frame
frames.
beams
rigid
continuous
and plane
frame
bases
The
of
each
were
column
were carried
out.
in the plane
designed
of the frame and in the
to be hinged
the
base
the
it,
to
about
rotation
while
plane
perpendicular

110.

longitudinal

Each
column was prevented.
to one of the
loading
by sidesway
was loaded
applied
the frame at the mid-depth
of the beam by means of a
for
the
There was a tendency
running
over a pulley.
stick

the

by friction

at

In

to

tested

simulated

high,

t he finite

having

of

inertia

strong
moment
axis
by,
area A br is given

48 yf.. I
br

of

cable
bases

to

k3
c

for

solutions
element
at the knees was

bracing

by beam members

and

knees

loads.

present
the lateral

order
frames,

frame

the

of

axis

area

sectional
a cross
I
The cross
br

br

sectional

(6-35)

br

the
beam
the
length
P.
the
is
member
representing
of
where,
br
knee bracing
used
coefficient
and yf is the nondimensional
Ku of the
(55) to represent
the axial
by Hartmann
stiffness
lateral
bracing
at the knees.
The nondimensional

in

which,

is

the

ZC is
least

The
member

EI

strong

16
br

6 is. the
where,
(55) to represent
lateral
bracing.

(6.36)

y
length

the

of

moment

simulating

by,

given

Z3

Yf =u

is

yf

coefficient

of

inertia

the
of.

axis
moment
the knee bracing

-GJ
4 'kb

column
the

of

given

frame,

br
by;

of

the

The-nondimensional

beam

(6-37)

P.
br

nondimensional
the torsional

and
section.

cross

column's

inertia
is

the

of

coefficient

used
effect

restraining

coefficient

by Hartmann

6 is

given

of, the

by,

Iy

ill.

Sx"'b

(6-38)

GJ

in

which,
flexural
axis

the

Zb is

strong

Yf

the

first

flexural

axis
to zero,

equal
the first,
held

of

stiffness

During

series

stiffness
the axial

while
tests

five

of the beam and Sx is

the knee bracing.

length

in

the

of
at a value
to 760.
equal

constant
was taken

(10

tests

of
of the

knee

stiffness
the
series,

20 while

the

tests)

strong

the

bracing

Sx was set
For
Ku was varied.

coefficient
for
the other

yf
five

was
tests

The effect
stiffness
of the lateral
of the axial
loads
buckling
bracing
KU on the critical
of the frame is
tendency
for
the
6.23.
As
there
in
fig.
column
was
a
shown
(50)
the
Hartmann
friction,
bases to stick
by
presented
load
for
the
theoretical
of
critical
values
corresponding
both

hinged

be seen
between

6.23

fig.

from
the

frames

based
finite
(53),

and for
that
there

based

frames.

It

can

is

excellent
agreement
and the theoretical
method

method
both methods

element
and that

of Hartmann
the experimental

fixed

agree

very

well

with

results.

the nondimensional
of tests,
series
the
the
axial
stiffof
reflects
effect
coefficient
yf,
which
knee
Ku
Joints
in
terms
the
bracing
the
lateral
of
at
ness of
held
the
the shear stiffness
constant
at a
was
of
column,
During

value

equal

to

the
represents
of the lateral

the

second

6, which
The nondimensional
coefficient
flexural
the
stiffness
axis
of
strong
effect
Sx in terms of the torsional
bracing
stiffness
For
the
first
five
the frame,
varied.
was
760.

of the beam of
6f
tests
in the series,
4.0,
the last
for
while
Fig.
stiffness
of
of the frame.
case
case

6.24

to
held
a
at
value
equal
constant
was
6f was taken
to
7.30.
tests
five
equal

flexural
of the strong
axis
effect
buckling
load
bracing
the
on the critical
for the
The theoretical
are presented
results
shows
lateral

the

hinged
for
the
frame
the
bases
the
and
also
are
of
when
in the plane
bases are fixed
of the frame.
when these

112.

The

level

and
first

the

of

between

agreement

the

corresponding
in
five
tests

finite

element
buckling
loads

solutions
for
the

experimental
(6 = 4-0)
However,
is good.
the series
(6
the results
tests
the
five
are higher
other
= 7.30)
of
by the
than the corresponding
given
of the buckling
values
(55)
that
the
finite
Hartmann
reported
method.
element
these

results
of
the friction

at
from

conditions
dicular
to

were
of the

bases

the

of
for

to

than

frame

changed the end

the plane
perpenin fig.
comparison

expected

which
fixed
in

partially
the frame.
values

due

higher

to

hinged

plane
shows that,
theoretical

also
Hartmann's

The
of

6 between

and

zero

8,

overestimated
method seems to give
the corresponload in comparison
the buckling
with
This
what was
confirms
solutions.
element
6.3-1.
the
found in section
when discussing
(fig.
6.11 and 6.12).
beam problems

values
of
ding
finite
previously
continuous

tests

the

6.24

5.2.

five

buckling

Interaction

of

doublv

T-portal

symmetric

frames
1973,

In

for

a theoretical
buckling
of
plane

analysing
method
I-portal
doubly
symmetric

concentrated
the finite

using
differential
of

technique

integral

proposed

lateral
loaded

frames

The method
the
to solve

is

inwith
based on

governing
buckling
behaviour

the

describe

which

(60)

and Trahair
the elastic

frame.
In

procedure,
series

of

strength
was fully
Rigid

loads.

transverse

equations

the

transfer

Vacharajittiphan

to

obtain
order
Vacharajittphan

on two small
I-section.
aluminium
tests

The beam-to

fixed.
the

all
lateral

structural

supports
the knee

movements at
loaded
two
with
vertically
the middle

a verification
and Trahair

were
joints

column

joints

the

theoretical

out a
carried
frames
made of high

end

of

each column
to
were designed
the binoments.

except
the lateral
to prevent
provided
Each frame was
of each frame.
action

at
of

the

with

(60)

portal
scale
The lower

loads
concentrated
equal
the
the column tops,
while
span

of

vertical

load

P1 each,
applied
beam was loaded
P

2*

at

113.

6.25

Fig.
of frames
interaction

tested

first
(60).

of the
shows the dimensions
by Vacharajittphan
and Trahair

group
The

is

also
shown in
between
the finite
6.25.
fig.
element
of agreement
load
is
the
buckling
the
of
values
measured
calculations
and
the theoretical
be seen that
It can also
method
excellent.
(60)
to
Trahair
Vacharajittphan
by
give
seems
and
given
load.
of the buckling
underestimated
values
this

for

curve
The level

group

6.26

frames

of

interaction

buckling
curve
of
presents
beam length
the second group of frames
which had a column-to
the
finite
Again
2.0.
of
to
solutions
element
ratio
equal
test
the
this
results.
corresponding
with
case agree very well
Fig.

6.6.

the

THREE-DIMENSIONAL
The problem

behaviour

BUCKLING ANALYSIS

of analysing
has been
frames

the

elastic

OF SPACE FRAMES
buckling

a topic
many years
The comprehensive

for

for

space
investigation
by many research
workers.
(9)
toward
formed
Renton
by
attempt
a
good
study
presented
deal
to
such
with
procedure
establishing
an analytical
limited
to
is
the
However,
space frames
procedure
problems.
has
been
The
joints
loaded
the
method
previousonly.
corner
at
ly

of

discussed
In
on the

study

unbraced

in

chapter

1980,
elastic

thesis.

this

2 of

(11)
Naim
a theoretical
presented
and
behaviour
instability
of rigid-jointed

Razzaq

single-bay

single-story

to
subjected
space frames
The
loads.
top
study
column

concentrated
or unequal
the modified
stiffness
matrix
elastic
was based on applying
(9)
to carry
by
Renton
elastic
out a aecond-order
proposed
the
By
frames.
the
proper
selecting
space
analysis
of
equal

the
loading
of
value
in
case,
each
mode
became
load-displacement
the
relatively
curve
which
load.
the
was taken
as
critical
buckling

The
the
the
Finite

study

herein

reported

of
and accuracy
validity
buckling
three-dimensional

solutions
element
by Razzaq and Naim (11)

of
are

the

was undertaken
finite
element

analysis
the

the

space

presented

of
frame
in

load
flat,

to

examine

solutions
space frames.

problems
6.8.
table

at

of

considered

114.

loading
the
system
and
shows
Naim
Razzaq
by
frame
and
jointed
studied
space
of the rigid
P4
P3
P
Pl.
loads
(11).
and
Static
concentrated
vertical
2'
defined
be
loads
These
tops.
can
at the column
were applied
Q as,
in terms of a load parameter
6.27

Fig.

pi

where,

Ai

is

dimensions

the

XiSQ

6.8

and

load

a constant

(6-39)

i=1,2,3.4

factor.

multiplication

between
the finite
comparison
loads
the
buckling
the
corresponcalculations
and
of
element
by Razzaq and Naim (11).
For the first
ding values
given
case
(case a), when the four
columns
of the frame are equally
the finite
loaded,
0.03 per cent
method gives
only
element
Table

shows

the

load given
by Razzaq and Naim (11).
more than the buckling
The convergence
for this
case
of the finite
element
solutions
6.9.
the
between
is shown in table
The agreement
of loading
(case b), when the
two methods for the second case of loading
four
is
loaded
but not equally,
columns
of the frame are all
by 0.04 per
The finite
differs
excellent.
element
solution
cent only from the value
Razzaq and Naim (11).

of

the

buckling

load

given

by

is
column
one
case
where
the
the finite
loaded,
of
element method gives a value
load almost 10.0 per cent less than the corresponding
buckling
(11).
information
No
detailed
by
Razzaq
Naim
and
value given
(11) about this
has been given in reference
case -However,
this difference
mode considered
may be due to the buckling
The
by Razzaq and Naim (11) not being the critical
one.
by Razzaq and Naim (11) gives
given
procedure
analytical
long
the
load
buckling
the
as
as
of
predictions
correct
frame.
the
joints
the
is
joint
of
corner
one of
critical
lies
joint
the
loading,
on
However, in this
critical
case of
Thus, the procedure
beam near the loaded joint.
the longest
to
lead
(11)
in
by Razzaq and Naim
some cases,
followed
may#
loads.
buckling
overestimated
In

'c'

of loading,

only

115.

The level
the
in

of
rest
the results

the

difference

6.7.

the

of agreement
loading
cases

in

cases

in

case

Id'

is

good.
If'
is

and
decreases

Iel

two

the

between

While

methods for'
the difference

2.0 per cent,

almost
to 1.5 per cent only.

SUMMARYAND CONCLUSIONS

has
the
finite
The applicability
method
of
element
by analysing
been demonstrated
of torsional-flexural
a variety
in
the
buckling
computer
using
program
presented
problems
this
thesis.
4
In
and
of
all
cases,
good convergence
chapter
have
been
by using
few elements.
accuracy
obtained
excellent
The results
of the
following
points:
Four

cases

types

of

pure

buckling

lateral

buckling

uniform

bending;

lateral

buckling

at

Finite

(b)

loaded

of an axially

were

strut;

of a simply

supported

beam by

of

supported

beam by

a simply

buckling

of a cantilever

the

end.

free

element

central

agreement

with

The following

beam by a concentrated

of these
(2).
solutions

solutions

compared with closed-form

showed a good convergence
excellent

problems

the

load;

concentrated

load

in

be summarised

namely,

torsional

lateral

can

buckling

conventional

investigated,

1.

considered

of the finite
the closed-form

types

problems were
The comparison

element

of continuous

solutions

and

solutions.
beam problems

were

analysed:
1.

elastically

restrained

interaction

buckling
(1)

continuous
of

continuous

beams;
beams.

of
was devoted to the effect
behaviour
lateral
bracing
of
on the buckling
stiffness
the
for
been
have
Solutions
beam.
presented
continuous
(54).
by
Hartmann
theoretically
problems analysed
the
dependent
is
on
warping
The accuracy
Hartmann's
method
of
The, first

part

116.

rigidity

of

rectangular
is very good.
the

beams
For
narrow
with
dross-section.
beam
the
Hartmann's
method
the
of
accuracy
cross-section,
increase
the
of
decreases
with
The accuracy
EIW of

rigidity

warping

beam has

a continuous

element
given

accuracy

been

the

between

spans
adjacent
finite
the
using

investigated

test
previously
The comparison
results,
with
(51),
and excellent
showed good correlation
finite
solutions.
element

method.
by Trahair
of

beam.

behaviour

The interaction
of

the

the

in chapter
The new geometric
matrix
presented
thesis
this
contains
new terms corresponding
to examine
In order
of monosymmetry.
effect

(c)

and accuracy
of the new terms,
have been investigated:
problems

validity
types
of
1.

2.

3.

the

3 of
to the
the
following

lateral

buckling

I-beams

by uniform

lateral

buckling

I-beams

subjected

monosymmetric
of simply
supported
load;
to central
concentrated

lateral

buckling

of

concentrated

of

load

simply

monosymmetric

supported

moment;

cantilevers
monosymmetric
end.
at the free

under

integral
The comparison with closed form, finite
(40)
(40),
test
showed a good converresults
solutions
and
finite
the
element solutions
of
accuracy
and
excellent
gence
in all
cases.
(d)

the
three-dimensional
of
solutions
element
buckling
behaviour
of plane frames, were oUained.
test results
The comparison was made with previous

Finite

for

two types

buckling
plane
2.

test

namely,

restrained
of elastically
frames under sidesway loading
buckling
of doubly
transverse
to
subjected

narrow
(53);

Interaction

symmetric

frames

vertical

the

results,

rectangular
I-portal
loads (60).

been
has
observed
agreement
cases, good
the
corresponding
and
finite
solutions
element
frame
member.
each
four
per
elements
using

In both
between

of problems,

117.

(e)

The validity
analysis
of

and accuracy
of the
the three-dimensional

of
frames

space

solutions
theoretically

were

have

finite

element
behaviour
buckling

been

obtained
by Razzaq

Finite
element
examined.
for the cases investigated
(11),
the
Naim
using
and

second-order
matrix
method proposed
The comparison
showed good agreement
4 elements
using
element
solutions,
the corresponding
of buckling
values
(11).
Naim
and
followed
procedure

Razzaq
the

predict

the

buckling

relationship
loads.
buckling

may,

in

by Renton
between

(9).
the

finite

per member, and

load given'by

showed that
study
by Razzaq tnd Naim (11) to
load from the load-displacement
However,

some cases,

the

lead

to

overestimated

1-Pure Torsional Bucklin2

vmmw

400.0Cm

CRUCIFORMSECTION 20OX20OXIOmm
MODULUSOF ELASTICITY Ex 200.0 KN/mm2

NO OF
ELEMENTS

PREDICTION

W0002

1.00002

ID0083

RAD.

ox

-4.

OX10

Fig.61 Distribution Of The Angle Of Twist Ox

10

28x 10

2
6x 10

4xlo

200

0.00
IIIIIIIIII
0.8Pcr

0.9Pcr

Pcr

Fig.6.2Modified Southwell Plot (8-Elements)

2- Lateral Buckling By Uniform Bending

a
I-CROSSSECTION
2
WOO KNI mm

2
OF SECTION: An 20.0 Cm
PROPERTIES
ly 8150.0CM4
2
JGzl000.0
KN Cm
I"s soo.0 CM6

t00.0 cm

.i

NO OF
ELEMENTS

PREDICTION

1.0065S

1.00165

UO 131

Mm
000

-2.0
-410
z
-6.0

Fig.63 DistributioryOf Lateal Displacementw

-soxio

tOAD FACTOR/A

2
SX10

2
WO -

2_
3XIO

002
72K10

2
io_

OR

0,00
O'sPcr

Pcr

Fig.6.4ModifiedSouthwell Plot (S-Elements)

Py

3-LATERAL BUCKLINGBY CONCENTRATED

,=1
LOAD
I. CROSSSECTION

M.
400.0Cm

NO OF
PREDICTION
ELEMSt-4TS

Er-200-0KNIMM2
PROPERTIES OF SECTIONs
2
Am20.0 Cm
ly a 180.0cm
2
JGr.IOOaOKN H
I.. a 500.0CM6

r-B

1.00480

Q999457

0.9967

OLM

0.0

-2D

-6.0
-8.0
-10.

3
OX107

Fig-65 Distribution Of Lateral DisplacementW

3
2XIO-

10

0.00
IU

Pcr

Pcr

Fig.6.6ModifiedSouthwell Plot (8-Elements)

4-LateralBucklingOf CantileverByConcentrated
Load

400.0 cm

LCROSS SECTION
Em2000KN I MM2

PROPERTIES
OFSECTION
A- 20.0CM2
ly. I aao CM4
Gjr.1000.
OKN CM2
1'. 500.0CM6

NO OF
PREDICTION
ELEMENTS

1.02588

1.01342
1.01228

mm

p-x

0.10
0.20
030
z

Fig.6.7Distribution Of The Lateral Displacement

LOADFACTORIA

Fig.6.8Modified Southwell Plot (S ElementsLoadApplied

Pcd

ta
DETAILSOF CROSSSECTIONa-a
AT INTERIOR SUPPORT
I

Ii
0

f--4

-+-

a -f.

Cc)

-0-a,

Fig.6L9Continuous Beams Studied By Hartmann (54)

ExamplesTaken For Comparison
I=Span
,

=4.0 ms

E=20000.0 KN/Cr
242

Section Properties= A=30.OCm Iy = 400.0Cm GJ= 500Kn. Cm 1,,,ToBe Calcutated

,

el

KU
.
4BEIy

S'X
.
GJ

(k)
CRITICALLOADPARA
METER
LOWER
BOUND HARTMANN FINITEELEMENT
(48)
SALVADORI
(S4)

2.0

11.0

370

41.70

52-90

53.0

11.0

190

28.40

34.40

34.54

6.0

11.0

180

25.20

29.60

19.64,

&0

11.0

110

23.90

27.30

27.25

00

11.0

50

22.0

23.30

23.66

Table6.1 Critical Load Parameter0<)For TwoSpanContinuous

Be6m-One Span Is LoadedAt The Middle(Casea)
.o

NumberOf ElementsTaken=4/Span

% Difference
2.0-

KI z 2.0
S11-370
GJ
Ku.1'
48EJy

No Ofetements/Span

0.011
246

Fig.6.10Convergence Of Finite ElementSolutions

Two Span Continuous Beam 'Kase a)

q#

1.0
.

50.0.

>, 40.0
-4.0

ix

64

30.C

(54)

-Hartmann

r%

so

Mao

----.

5-25
201

Finite Etement

-. -Lower

%i

Bound (4B-

4SEly

-9

so

100

ISO
sx.t
S
Gi

200

250

300

350

Bracing
Lciteral
Stiffness
Of
Flexural
Effect Of Strong-Axis
Fig.6.11.
(Case
Beam
Span
Two
Of
ci)
Load
On The Critical

iAA
Tund(Otl_'_)
Lower E

4.0

10.
1.
"

30P-

'el -( )0

Lower BoundI

Hartmann(5 1,)
-.. --100-

Finite El iment
Lower Bc und(48)

0.00

io

310

40

5.06070

90

80

SJ
SM

GJ

Fig.612Effect Of Strong-Axis FLexural StiffnessOf Lateral Bracing

OnCritical LoadOfTwoSpan Beam
Loaded
Are
Two
Spans
-The
(Case 6)

let

48EIY

Sx
Gi

Critical Load Parameter-IN)

LOWERBOUND HARTMANN
SALVADORW,'
48)

9
FINITE ELEMENT

2.0

11

477

51.6

820

81-53

4-0

11

374

35.10

5140

51-86

6.0

11

112

31.20

41-50

41-56

co

11

47

27.2

29.80

3050

Table6-2 Uitical LoadParameter(k)For Three

Span Continuous Beam-CentralSpan Is Loaded
At The Middle(Casec)
Taken
Elements
Of
Numer
=4/Span
*

P
Ac
2

Ir"

"I.,r- rr,

Fig-613TwoSpan ContinuousBeam With

OnlyOne Span Loaded
I

P2

30

12

4'rFig.6.14Two Span ContinuousBeamWith

TwoSpans Loaded
40.
35.

interaction

30,
25,

EXACTMETMOD
vi

:z

METHOO

---APPROX
is

10

Pi

P2
2

s
0

Y2t VALUES
IL)
I11.
20
5
10
is

25

30

Fig.6.15Critical Load Parameters For TwoSpan Beam

f2it,
50
)
5.0
tref
With
.

0 25

Elyal. 92 KIP. In2

506L-#

!,T1
I

GJ a 0.241KIP. In2
EIw a 0.154K IP. In4
KIP a 4.45 KN

Dims In Inches

0.055

Q62

i INCH z 25.40 mm

ly
CROSSSECTIONOF BEAMA.20.20.20-2
Pi

Pi

20.0in

20.0 in

20.0in

Fig.616.The Continuous Beam A. 20.20.20-2

Presented In Ref. (51 )

TEST

LOADPj(LH)

NUMBERIN POUNDS

LOADP2

LOADPI(RHI FINITE ELEMENT

IN POUNDS IN POUNDS CRITICAL

LOADFACTOR

44.8

45.0

44.8

Q9534

37.4

10.0

37.4

0.9894

39.6

20.0

40.2

09672.

42.0

31.5

41.0

0.9681

10.0

51.8

10.0

Q9974

20.0

57.0

20.0

09885

32.0

60.4

31.8

1.0007

46.8

63.8

47.2

G9369

49.8

9
10

36.0

35.8

1
1

0.9540
0.9786

Table 6.3 ComparisonBetweenThe Experimental

Critical LoadsAnd The Finite Element
/Span)
(40ements
Results
For BeamA- 20.20.20.2

LI

; PAN

! %-.II%.

W--

%0 11

liffiv.

Lkcillcill

For Test 7 (Re( 51)

7
EXPERIMENTAL

FINITEELEMENT
5
RAHAIR'AMETHOD151)
L

10.0

20.0

30.0

400

SO.
0

A20-20-20.2
Beam
Of
Loads
Fig.618Critical
(Ref. 51)

lb
448
INCH

2 O-D

lb
448

200*

lb
448

20-e

-8.0
-6.0
-4.0
-2.0

'I

00

U-

-2.0
3
10
x
-4.0

Fig.619 Distribution Of Lateral Displaeementw

ROO.
-4.0
-2.0
0.0

.T

X 10

3
ex

Fig.620 DistributionOf The AngleOf Twist Ox

LOADFACTOR/&
l
2xl?

FIRST PREDICTION
a 2.59837
PREDICTION
SECOND
al.03342
THIRD PREDICTION
a 0.95212
PREDICTION
xQ9s340

31
lo:

2- 1
-.

Lree

LOAD FACTOR

12P

Plot
Pcr)
(Load
Applied=
Southwell
Modified
The
6.21
Fig.
Span)
(4E[ements/
I
For Test

ly
bi

=Al
2z
lZ

(j ydA +J yd A) -2 yo

YO

AA

-4

Shear

Center' -hw

21

Fig.6.22MonosymmetricI. Cross Section

ly
mz

1
2ELEMENTS4ELEMENTS 6 ELEM.

S=-0.10
And
(ci)i=0-10,

1.0045

1.0004

2ELEMENTS 4 ELEM.

(b)K=1.0,And S=-0.60

1.0007

1.0005

0.9995

BELEM.
1.0001

6 ELEM

8 ELEM.

0.9971

1DOO1

Table6.4.Finite Element Solutions For Monosymmetric

Simply Supported Beams Under Uniform Moment
y
I

py

3c
. 46

W=010,
And S=-0.10

b)1-(=
1.0,And0.60

2 ELEM

4 ELEM

6 ELEM

8 ELEM

1052

1.0033

1D032

1.0032

2ELEM

4 ELEM

6 ELEM

8 ELEM

1.1863,

1.0348

1.01Bi

1.0105

Table 6.5.Finite Element Solutions For Monosymmetric

Py
Load
Central
Under
Beams
Supported
Simply

y
py
(a)l<=O.
lOAnd S=0.10

8=0.60
(b)R=1.0 And

2 ELEM

J
-6
4ELEM
ELEM

1.0810

1.0149

1D078

1.0064

2 ELEM

4ELEM

6 ELEM

BELEM

1.0356

1.0029

1.0028

0.9987_

I BELEM

Table6.6 Finite Element Solutions Of Monosymmetric

Cantilevers Loaded With PyAt The Free End

ly

65.0

6=0.18Anda=0.0 27

2ELEM

4ELEM

6ELEM

8ELEM

tO471

1.0082

1.000S

0.996S

R=1.0052

R=Ratio BetweenTheoretical Result Given

In Ref-40 And The Experimental Value
Of The Critical Load
(b) S=- 018And a z-0.22

2ELEM

4ELEM

1,0457

1.0017

6 ELEM
IDO12__

R=1.0265
Table 6.7.Finite Element Solutions Of The Monosymmetric
CantileversTestedBy Anderson And Trahair" ( 40) -,

8 ELEM
1.0013

tA

p
(IbOj

oj
1

0
6108,0
X, (Ku 13c/Ely)
tif

26

700

'800

f ness Of Knee Bracing On Buckling Load

Fig.6.23 Effect Of Axial
Of Frame(Comparsion Between The Finite Element And Ref.55)
No Of Elements Taken = 12
100

5 Tests
805Tests

e-

60p

ALL
Sx

Ku
-f

(lbs)
40-

Sect a-a
5 Tests

aI
Ic

(Hinged)

20-

-Hartmann
(Fixed)
--Hartmann
Element(Hinged)
-_Finite
Bement(Fixed)
_Finite
It3

i-&=3d
_T

f=Ku CIETy=760

0.00
48

5f = 5x. Lb IGJ

12

Fig.6.24Effect Of Torsional Stiffness Of Knee Bracing OnTheBuck.Lcad

(ComparisonBetweenThe Finite Element And Ref.55
No Of ElementsTaken= 12

2V-

20.0-

11
-3

oExperimental
Finite integral (Ref 60
.
II
Element
----Finite
(16EIemen

ru
10.0E2
rG
(U
cc

pp

1PI

14.Se?7,V. M17

.-- '"P
0.00
40.0

0
SO.
120.0
160D
COLUMNLOAD P1IN 1b,
I

120.0

Fig.6.25Interaction Buckling For Frame 1

In Ref 60
70.0-

609-Fini

Experimental
te Jntegra (Ref 60)

5aO-

16Elemen
C-4 400.
-0

30.0.
E
ru
ai
co

20.0.
,
29. 6
10D
15*
0.00

10D

20.0

30.0

40.0

50

2
Frame
For
Buckling,
Interaction
Fig.626
In Ref 60

Iy
I'

X40L
9-

4e

Y2'4

23
26
250.(

10

17

1.1

18

19-

22

25

/16

.)2

27

e
120.

---

12120

.
-=J.
isd

PLAN VIEW

29
'1

/Z-

FIN E ELEMENT
ARRANGEMENT
(4ELEM.IMEMBER)

%N
150.0

Fig.6.27The Dimensions And Loading System Of The

Frames Studied By Razzaq And Naim (11)

LOADING

LOAD MULTIPLIC. FACTORS FINITE

IN KIPS
CASE

X,

REF-(II)

A3

12

ELEMENTSOLUTIONS

CRITICALCRITICALCriticat Load
JOINT Deg-OfFreE Factor

601.0

1.0 1.005 1.0

1

1.0

16

1.0003

590.0

2.0

2.0

1.0

10

1.0004

520.0

1.0 0005 -

24

Q9012

576.0

1.0 1.0

30

1.0190

590.0

1.0 Q005 V05

1B

1.0149

591.0

1.0

32

09824

4.0

1.0 1.0

Table.6-8 Finite Element Solutions For The Space

Frames Studied By Razzaq And Naim (11)

NU OF ELEMEN
PERMEMBER
FINITEELEMENT

1.0160

2
1.008

1.0036 1.0003

SOLUTIONS

Table.6.9 Convergence Of The Finite Element

SolUtions For The Space Frame Loaded
With Case (ci) Of Loading.

118.

CHAPTER

SEVEN

Second-Order
7.1.

thin-walled

and accuracy

Z-Beams

second-order

members

of the new finite

torsional-flexural
in

presented

by analysing
a number of
examined
in
illustrated
been
has
chapter
as
loaded
to
limited
structures
were

this

study

element
behaviour

has

been

different
6.

stability
However,
the

problems

comparisons
flexural

in-plane
with
be
deformations
neglected.
can
only,
where prebuckling
the
the
the
of
cross-section
considered
problems
of

loading
In

of

INTRODUCTION

The validity
formulation
of the
of

Behaviour

Torsional-Flexural

each

members

structure

had at

least

one axis

of the stiffness
deflection
theory

The derivation

of

symmetry.

matrices
geometric
where the prebuckling

and

was based on the small

transverse
In
Z-beams,
however,
ignored.
deformations
a
are
the
in
lateral
load
have
of
plane
a
component
will
vertical
This load
the maximum rigidity.
causes a compara
component
buckling
before
large
displacement
starts.
tively
component
(stability)
The geometric
matrix,
3 contains
new terms representing
for
cross-sections
monosymmetry

chapter
sectorial

This

symmetry.

results

effect

obtain
stability
torsion.

study

quantitative
behaviour

report -ed in
information
of

Z-beams

of

of
with no axis
from the longitudinal
stresses
(bimoments)
in
and is given

moments
caused by the warping
terms
cross-sectional
of the sectorial
using
can be calculate
coefficient
The

presented
the effect

in

property
3.32.
equation

this

This

was undertaken

chapter
and
on the second-order
bending
and
under combined

to

the
are
results
and
program
Two
types
the
finite
of
solutions.
element
compared with
The
theoretical
of sectorial
effect
were
performed.
analyses
in
the
first
included
solution,
while
was
not
monosymmetry
this

effect

A test

was considered

was undertaken

in

the

second

analysis.

119.

TEST PROGRAM

7.2.

The test
The tested

,
tests.
of five
of a series
consisted
the same type and cross-sectional
Z-beams used for the
supported

program

beams were of
as the simply

dimensions

tests

bimoiaent

reported
shown

are

cross-section

The details

5.
chapter,
5.6.
fig.

in
in

the

of

in the
of 3.0 m, while
beam
2.0
Each
to
the
three
tests,
m.
span
was
equal
other
beam
to
the
loads
two
loaded
at
applied
point
equal
with
was
hangers.
by
three
the quarter
of
oints
means
p
quarter
and
in
tests
the
five
in
are
shown
Loading
of
each
arrangements
fig.

7.1.
The test

shown

rig

the required
simulate
These end conditions
in chapter
reported
tions

is

in

given

the

in

fig.

were
5.
A full

description

of

these

condi-

end

5.3-3.

section

were applied
displacements

horizontal

and 5.9 was used to

beams.
of the tested
to those
of the tests

5.8

conditions
identical

end

loads

Static
while

beams had a span

two

The first

the

of

the

to

incrementally
flanges

of

beam

the

beam

three-quarter
half,
points.
and
quarter,
were
from
the angle
twist
At each cross-section,
calculated
was
of
the
top
between
lateral
displacements
the
and
the difference
of
Vertical
deflections
bottom
were measured
ends of the, web..
dial
26
the
Locations
the
of
cross-section.
mid-span
at
gauges used
7.2.
fig.

the

at

measured

displacement

the

for

measurements

in

TESTING PROCEDURE

7.3.

During
to the
applied
Load increments
measured

the

first

top

flange

were

load

each

after

kept

(B - 1) dead weights
were
two
hangers.
by
beam
the
of
means
of
The support
movements were
small.

test

increment.

to
rig
was designed
free
and fully
warping

The test
conditions
the test,
which

shown

are

were

of
however,

connected

it

was noticed
to the flanges

the

simulate

restrained
that
the ILI
at

the

end
twist.

During

brackets,

theoretical

point

Plate

7.1

Beam B-2

at

the

moment

of

failure

"

"

"
...
-

-a

PlaTe

7.21

T,
'eb

and

flange

failure

of

beajfi

B-2

120.

(fig.

of

zev-o warping
twisting
at the

5-9),

In

supports.

ILI
of the
connections
Z-beam
made sufficiently
were
flange

The

these

nuts.

fully

prevented

connection

check
the

that

from

twisting.

web

imposed

the

warping

the

test

the

of

It

the

same

with

program

the

flanges.

theoretical
a)

namely,

end

rig
of

free,

then

tightening
success-

the

decided

for

to

compare
limiting

two
end

these

restraint

to

and
b)

and

for

nuts

Nevertheless,

was

the

of

arrangements

a noticeable

solutions

warping

two

carefully

these

twisting,

flanges

the
using

and

showed

arrangements

stiffening

with

L-bracket

to

by

slight

this

prevent

with

stiff

for

allowing
to

order

brackets

the

each

loose

were

continue

the

results

conditions,
restrained.

warping

All

beams were tested

In each of the
up to failure.
five
tests,
buckling
first
conducted
overall
started
followed
by a complete
collapse
of the mid-span
cross-section.
Failure
shapes of beams B-2
and B-3
are shown in plates
7.1 - 7.4.
7.4.

DISCUSSION

OF THE RESULTS

After

load

each

increment,

the

at each dial
displacements

deflection

The vertical
gauge was recorded.
and horizontal
into
in the
of the web were resolved
components
the two principal
axes of the Z-cross-section.

directions

of

Horizontal

move-

ments at the top and bottom

ends of the web were used to evaluate
Expethe angle
at the corresponding
of twist
cross-section.
displacements
the corresponding
are compared with
rimental
theoretical
the
As previously
theoretical
values.
mentioned,
was done twice,
The calculations

analysis
bimoment.
the
the

new geometric
longitudinal

matrix
stresses

firstly

were
terms which
caused

The

proved
of the
energy

study presented
the accuracy
that
of
bimoments
is excellent.
of
in

bimoment,

numerically
A-3-1.
appendix

out

carried
given

the

the effect
by neglecting
of
taking
into
account
repeated

the

the
reflect
by the bimoment.

in

chapter
the finite
For
integration

5 of

this

ellement

calculating
of

effect

of

thesis

has

calculation
the potential

eq. 3.46 was

BIMOM which

the subroutine
using
Values
of the coefficients

is

resulting

the

121.

from

integration,

the

kt,

the

ratio
beam.

for

the

of

the

of
to

the
the
the

values
experimental
the
free
of
results

the

span

the

of

the

of

lay

displacements

model and those

the value
However,

end warping

model.
end warping
restrained
displacement
horizontal
of the shear center
are
the
fully
restrained
model.
of
warping
results
the
fully

stresses,
the
using

bimoment

of

effect

did

the results
affect
The bimoment
model.

end warping
restrained
the displacements
increase
however,
free

in

given

effect
change the

and experimental
In
7.6,7.7,
and 7.8.

bimoment

of

of

values

calculated

model.

end warping

fig.

is
stresses
the calculated

B-2
beam
of
results
loading
this
case
and does not
small

very
displacements.

comparison
shows
that
up to a load of
of the angle
results

twist

of

ex,

7.8

Fig.

can be seen
theoretical

of the angle
of twist.
0.75 kN per hanger,
the

the

closer

not

The theoretical
are
the

the

fully

Including
of

and X is

on

of the
show the comparison
to the theoretical
In comparison

and 7.5
beam B-1.

7.3,7.4,

displacements

depend

KW
and
'

-*

b2'
bl,
/-G-J1EIw
k=

where

Fig.

results,
between

It

calculated
to the test

the

using

free

With
close
values.
are
model,
end warping
to change
6 starts
load the angle
increasing
of twist
X-1
between",
theoretical
the
difference
the
and"
direction
and
difference
This
begins
to
increase.
results
experimental
very

be due

to

could
beam at high load
insufficient
theory

applied
beams is

In

the

at

the

the

large

levels
to

displacements

which makes
deal with
this

by the

exhibited
the

small

type

of

(B - 3,4,5)
three
tests
other
The kk
level.
flange
bottom

deflection,
problems.
loads

value
of the

were
these

of
kX value

2/3
is
almost,
which
equal
(B
the
Thus
1,2).
beam
3.0
coefficient
the
span
m
of
in comparison
the
K
bigger
have
with
K
values
b2' . *. *, g t4
beams.
the
3.0
of
m
span
coefficients

in

fig.

to

The results
7.9 - 7-17.

1.04

of
It

tests
can

B-3,
be seen

B-4
that

and B-5
the effect

are
of

Kbll

shown

fk--

Plate

7.3

Beam B-3

at

the

moment of

failure

-id
Plate

7.4

Web

and

flange

failure

ol

beam

B-3

122.

bimoment
son with
is limited

the

to
Fig.

twist
the
of
of
angle
comparison
shows
between
the theoretical
The agreement

for the
values
the theoretical

increase

to

load

of

large

the
levels

Table

applied
displacements

solve-this

7.1

tical

loads.

angle

1.

may

the

end warping
The theoretical

the

theoretical

loads

buckling

of
failure

to the actual
model are close
6.0 per
loads
by almost
are higher
22.0 per cent for beam B-2.
and by almost
beams the theoretical
models give highly
That

in

prebuckling

the

to the

may be due

load.

stage.

CONCLUSIONS
A number of important
conclusions
discussion
in section
7-4:
The effect
of
beam property
of
of
free

The
the

the
kk

kZ considered
the bimoment

bimoment

stresses

stresses
k =V GJ/EI

, where
(kZ
in the study
has been
stresses

warping
at the
fully
restrained

are
bimoment
2.

difference

by the beam at
exhibited
deflection
theory
small

overestimated
of the buckling
values
large
displacements
which took place

the

increases

the theorebetween
comparison
buckling
loads
of overall
and the actual
beams B-1-B-5.
two
For thefirst

cent for beam B-1

For the rest
of the

7-5.

model
The
ex.

problem.

presents

predictions
loads
failure
of
beams, B-1
and B-2,
free

the

makes

hanger.

per

of twist
and test
results
Again,
this
load.

the

which

to

insufficient

the

model.

end warping

to
0.50
M
load
up
a
of
good
results
than 0.50 kN, the theoretical
levels
higher

the

effect

the

7.17

gives underestimated
between
difference

high

free

compari-

this

is

and test
At load

with
be due

in

Nevertheless,

the

of

results

beam B-5.

ex for

greater
two tests.

stresses
the previous

tests

three

these

in

is

is

of
calculations
displacement
small

can be drawn

is

For

W.
= 1.56,1.04)
confined
the ends

supports.

If

against

warping,

very
the

and

small
buckling

theory

is

dependent

the

the

from

on the
two values
the

to

the

of

the

effect

effect

case of
beam
of

the

can be neglected.

load

have

not

always

shown

that

sufficient

to

123.

deal

Z-beams

with
torsion.

Under

displacements
flexural

subjected
such loading

before

it

bending
combined
the beam exhibits

to

to'buckle

tends

in

and
large

a torsional-

mode.

The test

to simulate
the boundary
rig was designed
conditions
of free
end warping
and full
against
end
restraint
twist.
The warping
conditions
were not accurately
The connection
between
the L-bracket
simulated.
and the
flange
the longitudinalof the beam slightly
prevented
movement
supports

of
to

the

flange

semi-restrained

changing

the

warping.

conditions

of

the

pp

0.25

05

0.25

20.0
ECCEN. IN m.m

30

B-1

B-2

B-3

20.0

B-4

B-5

0.0

20.0+

Span i=2.00m

11=
Span
3.00m,

Fig.7.1Loading Cases Of The Tested Beams

LR

44

114

c1

Ll

L2

-9/4

C2 -L

'C3

R2 -L-.

Gauges,
Dial,
The
of
Locations
7.2
--, ' -Fig.

Rl

LOAD IN KN

SPAN 3.0

1.0

ol

lo

EXPERIMENTAL
FRff WARPINGT,, NEGLECTED)
FREEWARPING(Y.,CONSIDERED)
FIXEDWARPING(BOTH CASES)

0.50

ritm

IIIII

5.0

10.0

15.0

20D

25D

Fig.7.3 Test B-1 Vertical Deflection At Mid-Span

30.0 V

LOADIN KN

Span
MidMovement
At
Horizontal
B-1
Test
7.4
Fig.

Load In KN

100

Q50

0.10

Q20

Q30

0.40

Fig.7.5 Test B-1 Angle Of Twist At Mid- Span

ex

LOADIN KN

Fig. 7.6 Test B.2 Vertical Deflection AT Mid-Span

C
to
CL
V)

cu
E
cu
0

M:

76
a
0
0
CN

cu

cn
E
I-

LOADIN K14

Mid
At
Twist
Of
Angle
Test
B-2
7_8
Fig.
-Span

I-,

ru
N

. 4C
0
LI

a)

rn
CD
%A
0'

U-

z
lid
z
M
-CC
C)
-j

Cb
rd

CD
rli

-0

cu
0
1:

7;
0
0

rn

cn
cn
iT-

Se
C3
.:L
C)
-.i

LOADIN KN

C r) AU

')

%.0

Fig.711 Test B-3Angle Of Twist At Mid-Span

>

a
ro

ca.
V)

cu
cz

-;B
cu

>
-4t

cl
C14
W-1.
rl-

an
ir

z
Id

c
m
CL

cu
r--CL)

c:
0
N

CID

-0%A
(U
F-

rn

"i

C4

CL
(/l
12

CD
.ecn

91-1

rZ
.ch
kz

z
bd
z

CD
-i

r4

cs

C$
1

LOADIN KN

Fig. 715 Test B-5 Vertical

Deflection At Mid-Span

LOAD IN KN

Fig.7.16Test B-5 Horizontal Movement At Mid-Span

LOAD IN KN

ED
RED)

S)

o.

Fig. 7.17Test B-5 Angle Of Twist At Mid-Span

Experimental
load
failure
kN/hanger

Test

Theoretical

warping

buckling

free

warping

B-1

1.245

1.318

3.012

B-2

1.920

2.350

3.968

B-3

3.333

6.550'

12-545

B-4

2.963

6.014

11.805

B-5

3.755

7.626

13.212

Table

7.1.

Finite
loads
failure

solutions
element
in comparison
with
loads.

load

of the buckling
experimental

restr.

124.

CHAPTER EIGHT
n nV

TT Cl T MT

rIT

in
the
first
the
The object
part
work
presented
of
finite
(Part
has been to develop
thesis
I) of this
a general
instability
lateral
the
torsional
for
formulation
and
element
is
The
formulation
thin-walled
structures.
new
of
analysis
by
based on the theory
presented
of thin-walled
structures
Vlasov

(1).

The validity
by presenting
examined

and accuracy

problems
derived

which already
by alternative

between

the

of

these

finite

solutions
have exact
means.

element

problems

of
for
or

were
new formulation
a number of instability
highly
solutions
accurate
the

level

The g aneral

solutions

and

the

of

existing

agreement
solutions

was excellent.

to obtain
program
was carried
out in order
to combined
on the behaviour
of Z-beams subjected
torsion
element
and the validity
of the finite
The first
teat
this
program
of
such beams.
part

A test
information
bending

and

of
analysis
was devoted

stresses
evaluation
of the longitudinal
finite
the
the
by
the
bimoment
element
and
of
accuracy
caused
The second part
was
calculations
values.
of the bimoment
finite
the
to
the
in
of
examine
performed
order
accuracy
behaviour
buckling
the
element
second-order
analysis
of
and
of

to

the

Z-beams.
PRINCIPAL

CONCLUSIONS

8.1.1.
The derivation
has

been

buckling
placement
means that

based
the

on the

the

and geometric
at the instant
a torsional-flexural

stiffness
that
assumption
of

from

structure
passes
torsional-flexural
shape to another
included
the bimoment
are
stresses

shape.
in the

matrices
of
disThis
analysis.

125.

8.1.2.
Compared

more

types

undertaken
formulation

this

in

derived

buckling

of
to

examine

presented
previously
22,23,24,26,28),
matrix,

The geometric

the

following

conclusions:
with
behaviour

loaded

cantilevers
The matrix

with

includes

y and z which re fleet

'
behaviour,
monosymmetry
on the buckling
The new terms of the geometric
matrix,
characteristics

effect
of
the beam.

corresponding
been presented
In

to

the

in

effect

any

of

previous

have

monosymmetry,

the
the
of
not

study.
(24,28),

the
previous
studies,
M
Mz21 Qyl , and Q , in the
zl'
y2
have been corrected
to
matrix,
according
conventions
of the shape functions
used to
the displacement
to carry
In order
out
v.

comparison
signs
of the

geometric
the sign

the

with
forces

represent
" three-dimensional

instability

" new transformation

method has successfully

of frames,
analysis
The
been developed.

has
matrix
been applied

to

torsional-

frames.
space
and
the
between

buckling

problems
of plane
Excellent
agreement
was demonstrated
finite
element
solutions
and the available
or

geometric

can be used to predict,

the torsional-flexural

matrix

accuracy,

geometric

flexural

the

for
contains
new terms allowing
to be analysed.
The study
problems
the validity
of the new finite
element

beams and
of monosymmetric
in-plane
transverse
forces.

(ii)

the

of

thesis

in

results

excellent

forms

the

(stability)

geometric
matrix

to

highly

accurate

of

solutions

these

experimental

problems.

8.1.3.
By including

the

bimoment

as the

stresses

fourth

term

and by performing
of normal
equation
stresses,
a
integration
energy
of the bimoment,
of the potential
numerical
Each of these
in the geometric
terms
matrix.
new terms appear
w,
the
is function
property$
called
of a cross-sectional
in

the

coefficient
valid

only

of
for

sectorial
cross

monosymmetry.
sections

with

This

no axis

coefficient
of

symmetry.

is

126.

8.1.4.
two
in
parts.
separate
performed
was
program
longituthe
the
to
devoted
The first
of
evaluation
was
part
The
bimoment.
the
by
was
dinal
second part
caused
stresses
Z-beams.
behaviour
the
to
of
second-order
study
conducted
the experimental
between
drawn from the comparison
Conclusions
The test

and

theoretical

that:

were

results

The elastic
stiffness
thesis,
can
of this
accuracy,

the

subjected

to

The effect
behaviour

matrix,

of

presented

be used to predict,
bimoment
in thin-walled

chapter
excellent

with
structures

torsion.

nonuniform

of

in

bimoment

the

on the second-order
depends
on the beam property

Z-beams

kZ

beams
For
the
beam.
with
on the end conditions
of
the effect
of
ends fully
against
restrained
warping,
bimoment
is very
stresses
and can be neglected.
small
and

The
to

deflection
small
deal with
Z-beams
This

and torsion.
that
often

take

theory

is

subjected
is due to

place

before

sufficient
not always
bending
to combined
displacements
the large
beam tends

the

buckle.

to

8.1.5.
A finite
out

the

element

theoretical
before

available

computer

calculations.
the

according
modified
The method used to

start
to the

of

was used

program
The program

this

project

element
new finite
load
the buckling

predict
is the modified
displacement
results
to be applicable
The method has proved
in this
study.
presented
problems
8.2.

but

Southwell
to

all

to

carry

was already
it had to be

formu'ation.
from

the

load-

technique.
plot
buckling

SUGGESTED FURTHER WORK

8.2.1.
The application
dimensional
out

assuming

of

the

new formulation
and space frames

of plane
problems
the joints
that
are

sufficiently

to

three-

was carried
for
stiff

127.

to

warping
and Trahair
the
to

be neglected.
(64),
showed

one.
correct
in
incorporate

formed
the

warping

sufficiently
of the frame.
the

actual

rigid,
There

warping

that

The results
the analysis.

members
behaviour

steel

The

study reported
this
assumption
study
of their
With

the

to

by Vacharajittiphan,
is
is

use of coldspace frames

growing

and
portal
construct
which are
of the joints,
in
important
role
an
plays

a need to present
behaviour
of the frame
is

always
difficult
rather
not

not
the

always
behaviour

information

about

joints.

8.2.2.
.

A finite
ment

theory

is

element
needed to

formulation
allow

for

based
the

on large

analysis
The Z-beam

displace-

of crosshave
tests

with
no axis
of symmetry.
to
the existing
technique
shown that
which is restricted
deformations
is not sufficient
to overestimated
and leads
load.
values
of the buckling
sections

small

PART
-------------

(II)

128.

CHAPTER NINE

Diaphragm
Plate

Pyramid

Structures

Folded

with

Roofs.

9.1.

INTRODUCTION
Safety

of

Truncated

in

Action

structural

and economy are

Achieving
design.

structure
economical
in
any development

has

always

two

major
lighter

a safe but
been the main

always

structural

the

criterions
and

motivation

more
for

engineering.

The improved

utilisation
of the mechanical
properties
has led to a better
of structural
materials
of, elements
shaping
distributions
It has also
with
economical
of the material.
helped
in creating
more advanced
and new
concepts,
structural
methods

of

analysis

to

It

has

realised,

achieve

an optimal

use

of

material

capabilities.
been

through

investigaexperimental
tions,
that
the classical
the structure
as
way of analysing
from
main and secondary
systems
and going
consecutively
to main is not always
to describe
the right
secondary
approach
the

behaviour
In fact,
proper
of the structure.
each component
has its
to the whole stiffness
of the
part
own contribution
be treated
In many cases the structure
as
structure.
should
its
different
types
own
one system with
with
each
of element,
function.
The
of
in

contribution
building
framed

a steel
the traditionally

Theories
cost

to

explain
in
structures

plays

In
low-cost

1974,

skin
BS449 Appendix

at
building
AD).

can

calculated
the steel

which
in
role

a dominant

the

of

the
the

Mytchett

steel
result
stresses

by the

Metropolitan

Education

(MACE)

to

be used

the

stiffness
decrease
substantial

skin
in-a

and displacements.
have led to new low-

skin action
in-plane
of
strength
behaviour.
structure

near
was tested
The building

developed

to

Camberly,
(1,2)
in

Surrey,

the

sheeting

a full

size

accordance
with
was one of the units
Consortium
for
Architects

as nursery

schools.

The roof

of

129.

the

framing

formed

cold

the

had

building

a truncated

of

shape

and

members

covered
Later

profile.
of trapezoidal
sheeting
Department
in
the
test
out,
was carried
to
Salford,
University
the
study
of
of
bay Pyradome

structure

There
it

was no precise
to
was necessary
of

rements

by corrugated
steel
in 1980 a similar
of
the

Civil

Engineering

behaviour

of
Brothers

by Oldroyd
prefabricated
such structures,
method to analyse
the
to satisfy
test
unit
a typical

a twoLtd. (4).
thus
requi-

BS449(3).

The stability
Pyradome

light

with

pyramid

of

both,

the

on the

structurepdepends
Apart
from the

two-bay

and the

MACE unit
diaphragm

the

of

action

roof
in

given
approximate
calculations
to
1
2
for
the
MACE
building
method,
and
no precise
references
to analyse
has been presented
the knowledge
of the writer,
The study
in this
was underchapter
such structures.
reported
the
taken
to establish
a theoretical
method for predicting
sheeting.

elastic

behaviour

cularly

under

of

truncated

under
the

trapezoidal

a single
in-plan

pyramid

parti-

structures,,

loading.

asymmetrical

Initially,
studied

the

of

cases

panel

loading

using

was
of the roof
three
methods of

full
frame simulation,
and
element,
that
the
The
have
simple
truss
shown
results
simple
modeling.
displacement
the
to
is
truss
explain
sufficient
modeling
the
to
trapezoidal
the
predict
and
panel
of
configurations

analysis,

fastener

forces.

a space
in-plane

frame

finite

has then been simulated

The MACE unit
with
the
to
truss
having
represent
members
model
The
twothe
flexibility
sheeting.
of
roof

shear
bay Pyradome
structure

has

been

also

analysed

the

using

same

procedure.
Because

degree
forming
been
rigid

it

was not possible

of the connections
of fixity
the skeleton
of the structure,
considered,

namely:

a)

hinged

to

simulate
between the
two

limit

connections,

the
precisely
frame members
bounds
and

b)

have
fully

connections.
A full

and those

comparison
during
recorded

between
the

test

the

calculated

is

presented

displacements
at

the

end

of

130.
p

this

method

frame

has

comparison
to predict

be used

can

the

shown that
deflections

of the
11.5.2.

model

section

MACE unit

are

also

STRESSED SKIN DIAPHRAGM ACTION

9.2.1.

General
denotes

diaphragm

proposed

of

and discussed

presented

9.2.

The term

the

such
The resulting
forces
in
from the space
predicted

accuracy.
with
sufficient
to frame member fasteners

structures
the sheet
in

This

chapter.

a planar

system with
a
the overall
is very small
thickness
that
compared with
in
Such a system possesses
dimensions.
substantial
rigidity
flexible
in
the
it
transverse
its
very
remains
while
own plane
This property
direction.
makes the diaphragm
enormously
of

capable

in-plane

resisting

forces.

shear

fastened
to the supporting
properly
sheeting,
-Steel
framework,
acts as a series
of shear
of a steel
elements
in a secondary
Such diaphragms
diaphragms.
may be used either
the structure
against
sway or in a primary
sense to support
sense
9.2.2.

the

in

shell-type

of

case

diaphragm

C. B. Johnson
was tested
In
experimental
to
Cornell
shear
test"

under

the

1960,

Nilson

effect

of

(5),

investigations
the

diaphragms.

factqrs

He then

as an experimental
flexibility
of diaphragms.

panels

They

predict

the

pointed
shear

the

out
where
lateral

presented

first

attempt
California
in

in

the

out in
carried
that
influence

out the
flexibility,

studied

the

corrugated
and
need for theoretical
and proposed

study
1947 by

of some
results
the University
of
the behaviour
of

behaviour

of

to

building

a full-scale
loads.

the so called
suggested
technique
to evaluate
the

(6)
Ammar and Nilson
different
types
using

sheet.

(5),

by Nilson

was carried
action
J.
Converse,
F.
and

study

plates.

background

Historical
As indicated

the

and folded

structures

"Cantilever
shear

of

shear

orthotropic

a finite

methods

to

element

131.

model to describe
diaphragm.
In
20 years,

the

1973,

distribution

of

a research
after
(8) published
the

Bryan

internal

forces

lasting
program
first
book in

in

the

almost
the stressed

This book represents

the first
comprehensive
action.
in the design
action
guide to the use of the shear diaphragm
To make use of such action,
he suggested
frames.
of steel
diaphragm
that
the flexibility
of a complete
can be obtained
flexibilities
by calculating
each of the component
and summing
skin

them.
Davies
given

by Bryan

profile

corrugated
to wrong

estimate
technique

energy
finite

proposing
of

(Cl.

that

flexibility

expression
of the

is

tables
and provided
(11) improved
Davies
for

representation
a more accurate
the corrugated
profile.
Finite

the

lead
not
and
always
valid,
can
1)
They described
of the deflections.
an
C1.1,
to calculate
tests
it with
checked
and

results,
Later,

element

applications.
shape

(9,10)
Lawson
showed
and
(8) for the distortional

element
method to

appropriate
forces
internal

has

modeling
in
study,

detail,

been

for

practical

the

method,
the distortional

adopted
as the
the distribution

most
of

Nevertheless,
the amount of
and deflections.
type computer
data and the special
required
make it
program
(12) suggested
Davies
to be used.
difficult
a simplified
the
be
diaphragm
whereby
can
components
of
analysis
method
frame
different
types
by
of element.
with
plane
a
simulated
The analysis
frame
plane
agreement

can

then

be carried

out using
a conventional
The method showed an excellent

program.
computer
with the finite
element
the

modeling.

the
most comprehensive
study
concerning
is the book published
by
skin diaphragm
action
stressed
(13)
The book deals
Bryan
in 1982.
Davies
with
every
and
'many
detail,
the
in
in
discusses,
techniques
subject
and
aspect
the diaphragm
in the design
to incorporate
action
of steel
So far,

structures.

132.

9.2-3.

Diaphragm

action

9.1

Fig.

shows

a pitched
roof
Under the
sheeting.

corrugated
steel
load, shown by the
This

clad

with

of the vertical
effect
the top of the frames.
the apexes
the eaves tend to move outwards.
by in-plane
distortion
of the

at
arrows
to move downwards
while
movement is accompanied

tend

frame

portal

The sheeting,
its
in-plane
with
sheeting.
enormous
this
to resist
tends
movement by acting
stiffness,
as the web
The two outermost
of a deep plate
girder.
purlins,
at the
roof

form
and apex,
due to
forces

eave
axial

the

of carrying
foundation

flanges

bending.

the
of this
and carry
girder,
The end gables
be capable
should
the deep plate
to the
girder

reactions

of

important

application

(8,13).

Another

the

of

shear diaphragm
flat-roofed

be used to prevent
sway of the
9.2,
In the structure
shown by fig.
side
directly
in the plane
of the
are applied

to

is

action

the

structures.
eaves level,

loads

at

sheeting.
The deep plate
composed of the sheeting
and the two
girder
load back
the lateral
outermost
at the eaves,
purlins,
carries
load
Such gables
the lateral
to the end gables.
may transfer
by diagonal
bracing
to the foundation
may act as
or the gable
Vertical
loads
sheeted.
are
a vertical
by the main system which
taken
as a simple
can be designed
The horizontal
in the
wind bracing,
and column structure.
(7,13).
be
the
omitted
can
of
roof,
plane
diaphragm

if

it

is

beam

folded
building
plate
with
shows a low-cost
Such buildings
rely
on the diaphragm
action
entirely
of
roof.
loads
to
lateral
to
the
carry
and
vertical
roof
sheeting
Uniformly
distributed
load on the roof
is
the end supports.
9.3

Fig.

the
in

fold
the

to the fold
sheeting
into
themselves
resolve

by the

transferred

lines
two

elements

plate

Each

plate

element

deep

plate

girder

the

structure.

the

stiff

with

system
The load

end gables

and

which
its
two

meet
fold

lines.

Line

loads

in-plane

loads

acting
line.

fold
at a given
line
members forms
to the length
equal

a span
with
by the
is carried
then

to

the

deep

foundations

girders
(13).

a
of
to

on

133.

9.2.

_4.

Diaphragm

arrangements

9.2.4.1.

Basic

arrangements

shows

the

9.4.

Fig.

The

diaphragms.

and

two

components

basic

of shear
may be directed
sheeting
(fig.
9-4. a)
the diaphragm
arrangements

of the profiled
to the span of
(fig.
9.4. b).

span

either

perpendicular
to it
or parallel

is defined
unit
of a diaphragm
as the area
.
by two consecutive
of sheeting
enclosed
rafters
and by the
The diaphragm
may be fastened
edge members.
either
on all
(direct
four
to the
shear transfer)
or on two-sides
sides
(indirect
(13).
transfer),
members only
shear
perpendicular
The design

Components

9.2.4.2.

lengths

a)

Individual

b)

Perpendicular

of

profiled

members.

The

fastened

to

c)

Parallel

members.

d)

Seam

these

fasteners:

sheeting

to

perpendicular

f)

Sheet

to

parallel

member

edges

firmly

of

members
directly

fasteners.

fasteners:

member

perpendicular
can

be

must

widths.

Sheet

sheeting

sheeting.

members.

e)

and

panel

connect1longitudinal

sheet

adjacent

a diaphraZm

of

are

be

If
the

at

parallel
level,

same
to

connected

the

the

the

parallel

members.
g)

Shear

the

over
equal

to

the

Seam failure.

failure

the

fastened

no

Failure

The possible
a)

have

members

perpendicular

can

importance

and
four
in

used

to

parallel

sides
the

members:

these

diaphragm

modes
modes of

the

depth

members.

parallel

on

be

pass

with

connectors,

perpendicular

is

sheeting

connections
(13).
action

9.2.4.3.

to

sheeting
between

Connections
If

the

members
parallel
shear
the difference
in level,
the

connect
h)

If

connectors:

diaphragm

are:

134.

b)

Failure

in

C)

Failure

in

d)

perpendic-lar
Failure
due

e)

Failure
axial

first
9.2.5.

(if

used) at the
shear connectors
the fasteners
the'sheet
connecting
members.
to buckling

of the
forces.

outermost

is
The diaphragm
(12).
three
modes
Structural

of

most

behaviour

the

of

to

hipped

ends.

sheeting.

perpendicular

likely

panel
to

fail

roof

members

due

by one of

the

to

structures

A somewhat different
stability
example of a structure
9.5.
is the MACE building
by diaphragm action
shown in fig.
load the lower horizontal
distributed
Under uniformly
vertical
by the corner
frame members act as a tension
ring supported
The upper horizontal
frame
columns and the roof sheeting.
mainly by the
members form a compression,, which is supported
line
The upper horizontal
hip members.
members carry their
At a
by bending action.
back to the apex joints
load can be resolved
into axial
the resulting
apex joint
components in the three framing members meeting at this
determinate
Thus the frame member forces are statically
the stressed
skin
considering
without
can be calculated
case.
which is of a secondary nature in this loading
loads

given
joint.
and
action

by the
of the frame members is restrained
This
in
the
the
restraining
plane
of
roof.
skin action
stressed
bending
the
in
and
significant
of
reduction
a
effect
results
(2,13).
in
hip
deflection
the
members
particular
of
The bending

loading,
however, the stability
Under asymmetric
of
depends mainly on the stressed
the structure
skin action
of
forces
The axial
in the
the trapezoidal
panels of the roof.
frame members meeting at the apex joints
no longer balance
behaviour
of the structure
and a complex three-dimensional
The behaviour
of hipped roof structures
under
results.
by testing
loading
could only be predicted
asymmetrical
(13).
full-scale
unit

analyse

The theoretical
the hipped roof

method proposed
will
structures

in this chapter to
be shown to be in good

135.

agreement
carried
site

the

with

experimental
MACE building

out on the
and, b) the prefabricated

9.3.

of:

results
which
two-bay

the

a)

test

was constructed
on
Pyradome
structure.

LOADING TESTS ON THE MACE UNIT AND THE PYRADOME

STRUCTURE

9.3-1.

The MACE unit

shows a typical
by the Metropolitan

constructed
Education

30 unit
building
Consortium
for

MACE type

9.5

Fig.

Architects

in

and tested

BS449.
The roof
accordance
with
was
Framing
truncated
in
had
and
a
pyramid
shape.
square
plan
the changes in slope,
from
were constructed
members, following
Details
of these
of the cross
sections
cold formed steel.
9.5.
in fig.
members are given
Intermediate

to

the

allow

the

columns.
were used to
lateral
resist

supports

lower

were

square

by cladding
panels
provided
to span 10.80 m between
two bracing
members
corners,

members
four
At each ofthe
the column and they
support
wind loads.

were

members were designed

square
In the test,
this
light
on top of the roof.
the
deck
through
by
timber
a
which
replaced
(1,2).
light
was applied

to

The top

The roof
depth

and a net
frame
supporting

diameter
of

sheeting
thickness
members,

had
of

a trapezoidal
0.67 mm. It
four

on all

designed

support

light
load

to

a roof

panel
was
from the roof

80 mm
the

with
profile
was fastened-to

sides,

with
through

6.1

mm
the troughs

self-tapping
screws
Seam fasteners
were self-drilling
corrugations.
4.1 mm diameter
and placed
at 250 mm
screws with

self-drilling

alternate

self-tapping
centers.
9.3.2.

The

MACE unit

As it
designed
structurescale

unit

to

was not

loading

possible,

make a reliable

tests

at
prediction

the

time
of

under asymmetrical
particularly
the
to satisfy
had to be tested

the
the

structure
was
I/
behaviour,
the

loading,
requirements

a fullof

136.

BS449
a)

(3).

The test
tests,

stiffness

Stiffness

a)

into
program
was divided
tests.
and b) strength

Three
stiffness

Test

Test

different

1.

Load

Load

2.

Load

3.

tests

= Dead load
the whole

area.

= Dead load

+2x

the

third

was kept

test,

on the

= Self

for

(horizontal

x imposed load
(asymmetric
roof
load

the

tests

over half
loading).

was maintained
evaluated.
the load

recoveries
were
the deflections,

the

strength

(dead

wt, +2x

load

5 the

test

was unloaded

tests.

the

roof.

entire

4+2x
at

later

use,

load)

+ imposed

eaves

wind load
level).

for
was maintained
and the recoveries

24 hours
were

was a test

As this
required

out

= load in test
(horizontal

Load

structure

over

tests
Load

5.

In
the

the

level).

recording
carry

load

load

wind

and the

covering

Test

satisfy

out

x imposed

+ 1.5

two

released

after
to
roof

Strength
4.

eaves

first

the

each of
24 hours,
then

Test

carried

were

= Dead load + 1.5

the area of the

In

b)

to

tests

at

In

parts:

requirements:

Test

for

two

on an actual
structure
to failure
no tests
were

then

measured.
was

which

undertaken.

to the roof
directly
were applied
56 lb each.
Plywood
sheets
sand bags weighing
sheeting
using
Lateral
loads,
the load of the bags.
were used to distribute
by a system of
the wind effect,
were applied
simulating
Vertical

pulleys
in two

and wires

loads

at

eaves

level;

wind

test

was carried

'-

out

steps:

a)

Lateral

load

of

2.46

kN/m applied

to

one

side

b)

structure.
Lateral

load

of

1.23

kN/m applied

to

two

opposite

of

the
sides.

137.

In
increments
gauges

supported

along

9.37

took

members

are

two

in

given

connect

longitudinal

9.3-4.

Loading
The test

tests

were:
Test

1:

fig.

Test

2:

3:

together

with

structure
Pyradome

11-37.

used had a trapezoidal

profile
It was fastened
mm thickness.
four
through
on all
sides
every
3.2

self-drilling

used

of

tests

on the

adjacent
two-bay

included
program
tests
side loading

Load

taken

Load

taken

up to

applied

uniformly

Initially

the

load

(4).

increased

over

(0.8

load

acceptance
over the

the

of

area
test

2)

applied

roof.

,
2
kN/m

(1.19

load
area

entire

was applied
2
0.4 kN/m

loading

kN/m

of

the

roof.

over the whole area

The load was then
(fig.
9.39)
up to
loading).

had been recorded

(fig.
IBI
9.39)
area

reading

load

main vertical
The vertical

of the roof
up to
'A'
increased
only
over area
(asymmetrical
full
load
working
deflection

corrugation

Pyradome

three

up to working
over the entire

with
to the

self-tapping
screws.
to
at 200mm centers
(4).
sheet widths

and placed

edges

uniformly
Test

are

together
MACE units
connected
forming
the roof
units
were
together
to form
on site
and bolted
Details
pyramids.
of the framing

truncated

members
supporting
Teks code
trough
using
Similar
fasteners
were

and

deflections

structure
consLtd.,
in the Department
and tested
(4).
University
The
of Salford

industrially

five

dial

using

shape of two
The trapezoidal

The sheeting
60 mm depth and 0.7

tests

recorded
(9-5.2.1)

two-bay

the

one eave.

prefabricated
the shape of

The

five

or

results.

the

of

four

recorded

were

section

shows the
Brothers

Engineering

structure

in

in

applied

readings

Pyradome

by Oldroyd

Civil

was

theoretical

The two-bay

tructed

load

scaffolding.

discussed

and

Fig.
of

on

corresponding

9.3-3.

the

deflection

while

presented
the

test

each

the
up to

After
load
full

was

138.

load.

working
the

at

to'

The

the

of
2 ).
kN/m

entire
(1-48
load

area

using

bags

sand

To carry
out
and IS' (fig.

load

the

increased
to

tests

the

two

columns
bolts
anchor

above the
to loading.
prior

lifted

base
in

Side
level
detail

141.

reference
During

and horizontal
increment
at
displacements

applied

were

of the bases and supported

on skates
loads were applied
loads
as concentrated
at the
hydraulic
jacks.
These tests
using
are explained
in

test

kg.

25

weighing

over

prototype

loads

loading

,
were

then
up

roof

Vertical

each

side

9.39)

was

and

vertical

side

movements were
locations
sixteen

of the working
(test
loading
test

asymmetric
(9.5.3)
together

with

loading

tests

the

recorded

after
every-load
The recorded
dial
using
gauges.
(test
1) and the
load test

in
considered
theoretical
corresponding

the

vertical

3)

are

section
results.

9.4.

THE BEHAVIOUR OF THE PLANE TRAPEZOIDAL PANEL

9.4-1.

General

9.5,
is
The roof
shown by fig.
of the MACE unit,
diaphragms
by
trapezoidal
supported
composed of four
plane
Each diaphragm
the four
acts as a deep plate
corner
columns.
the two
depth.
Frame members represent
with
variable
girder
flanges

while

the

Under

uniformly

roof

sheeting

acts

as the

web of

the

deep

girder.
distributed

of the
stressed
skin action
loading,
Under asymmetrical
structure
trapezoidal

depends

mainly

sheeting
however,

on the

vertical
is of

loading

the

nature.
of the
of the

secondary

the

stability

diaphragm

action

panels.

of the
of the behaviour
analysis
is
trapezoidal
under in-plane
cases of loading,
panel,
three
described
The analysis
herein.
out using
was carried
the shear flexibility
different
to simulate
techniques
of
is
the diaphragm.
A full
of the results
comparison
A comprehensive

presented

later.

139.

The three
1.

Finite

2.

Plane

3.

Simple

used

are:

element
modeling
frame simulation
truss

The four
analysis

techniques

analysis.

in-plane

cases

loading

of

considered

in

the

are:

Case 1:

distributed

Uniformly

load

kN/m acting
the trapezoidal

2.6

of

the bottom
along
chord of
(the
panel
component
of 2.46 kN/m horizontal
one of the eaves).
wind load along
upward

Case 2:

Two vertical
loads
concentrated
of 10.0 kN each
downward at the two top corners
acting
of the
trapezoidal
panel.

Case 3:

One vertical
a top corner

Case 4:

a)

load

10.0

at

kN acting

was carried
the junctions

at
connections
hinged
connections,
Finite

9.4.2.

hinged

proposed
the method

elements
trapezoidal

a top

to
of

corner

at

the

chords)
of
the trapezoidal.

out for two different

of the framing
member,
fully
connections.
rigid

types
namely

modeling
shear diaphragms
regular
(6).
Davies
and others
trapezoidal
panel
of the

to

analyse

action

of

the
uniform

wind

load,

considering

joints.

model described
shows the finite
element
Five types
2 and used in the present
analysis.
the components
have been used to simulate
of the
diaphragm.
These types
are:

Fig.
reference

(parallel

downward

modeling
element
of
by Ammar and Nilson

the

under

corner

and b)

element

The finite
was first
(2) used
MACE unit

of

of 10.0'kN
acting
the trapezoidal.

One horizontal

The analysis
of

load

9.7

1.

Orthotropic

2.

Orthotropic

3.

Beam elements

rectangular
triangular

plate
plate

elements
elements

in
of

140.

4.

Spring

elements

for

sheet

5.

Spring

elements

for

seam fasteners.

these

element

Each

of

to

frame

types

fasteners

now be considered

will

in turn.
Design

of

rectangular

The steel
corrugated
two-dimensional
orthotropic
degrees
of freedom
at each
matrix

of

sheeting

plate
can

elements

be represented

by

two
rectangular
with
elements
To derive
the 8x8
stiffness
node.
(6) developed
Ammar and Nilson
the

the

element
D where,
matrix

elasticity

E
x

1y-

-Y xy

y
yx

E
-y
x
xy

YX

0
0

yy
(1-Y

00

xy

)Geff
Yyx

moduli
of elasticity
and Ex are the effective
y
two axes of orthotropy,
are
and y
y and x, y
xy
yx
is the
the corresponding
of Poisson's
and G
values
ratio
eff
The
be
two
li
of
modu
elasticity
can
modulus.
shear
effective
(6)
Ammar
Nilson
by
from
the
and
expressions
given
calculated
in

whichjE
the
along

as follows:
t

and

E0
3r

E0

(9.2)

E
x

E0

(9-3)

I/ Jh
4-

Figg. 9.6.
in
the

= the

whichto
modulus
Io

and
where,

I=
the

of

length

developed

elasticity

of

the

Profile

one corrugation,
10 and I
material,

dimensions.
EO =

of

are

given

(9-4)

dt3

7-

12

t(b

h2+2L
03o

notations

are

3)

given

(9-5)
in

fig.

9.6.

by,

141.

The two
be calculated

y
y

and
where

of Poisson's
ratio,
following
relations:

values
from the

y0 is

and yxyp

can

= YO

(9.6)

Ex = yyx. Ey

(9.7)

zrx
xy.

yyx

the

value

Poisson's

of

for the
An expression
from the total
can be deriven
(14).
This expression
sheeting

for

ratio

the

material.

shear modulus G
eff
shear deformations
of the
takes
the form:

effective

[d2*5

in

which

eff

=E 0/

b=

the

frame

fasteners

pieces

in

profile
the

and

triangular

have

the

been

inclined

of

the use
avoid
as far as possible

in

the

edges

of

the

node,

Fig.
to

simulate

9.9
the

such
(13).

have
elements,
of the ortho-

Beam elements

degrees
Beam elements
have three
the frame members of
to represent
9.4.2.4.

of

as infil
analysis
present
The elastic
diaphragm.

for rectangular
mentioned
previously
the 6x6
to derive
matrix
stiffness
triangular
element.

9.4.2-3.

to

sheet

to

used

function

elements

properties,
been used
tropic

is

which

arrangement

the

to

parallel

constant

recommended
strongly
the diaphragm
analysis

they
at

diaphragm,

is

It
elements
However,

the

Orthotropic

9.4.2.2.

(9.8)

b0

depth
of
R=
the
and
dimensions
(13).

corrugations,
of the profile

)(l+

2(l+y
tt1 -5

2h

elements

-Sprina
shows
sheet

the
to

for

of
the

sheet

freedom

at

each

diaphragm.
to

element
spring
with
frame members fasteners

frame
zero

fasteners

size
and the

used
seam

142.

fasteners.
two

Each

directions,

orthogonal

fastener
have

been

has

different

seam line

the

been

for

The finite
is

analysis

9.4-3.

Plane
This

to

similar
values
individually
spring
rectan-

openings
element

The

that

frame

computer
by Davies

used

to

frame

taken

member
as 0-151n, 9110

used to
(2).
others

program
and

carry

out

simulation

method

method

is

plane

simplified
each

stiffness
to frame

frame

based

computer

The analysis
of
by eliminating
joint
in

(12).

however,

member fasteners.
the x-component

force

the

of

program.

allowed

be
can always
the displacement

diaphragm

a rectangular
the x-component

of

having
only axial
elements
spring
the sheet
K can be used to simulate
y
diaphragm,
In the trapezoidal
the

produced
strain
included
to
be
has.
member
of

The model used to, simulate

.
diaphragmis
trapezoidal
shown
to

diaphragm

Hence,

Y-direction

inclined

the

on representing

in-plane
forming
of, prismatic
an
members
a convencan then be, carried
out using

by a number
The analysis

assembly
frame.
tional

element

sheet
been

(12)
by
Davies
was first
proposed
who
it to regular
diaphragms
and to diaphragms
with
it agrees
the finite
and showed that
with
well
types
analysis
of these
of diaphragms.

applied

were

have

of

of both the
fasteners
and the seam fasteners'has
in the two orthogonal
directions.

the

tend

fasteners

seam

fasteners

The flexibility

at

the

sheet

they

since

in

elements.

gular

the

to

frame

Ky

displacement.

and

elements

and

modeling
each seam fastener
by a small
replaced
number of equivalent
the seam line
along
at the nodes of the

placed

elements

force

Spring

a given
Instead

Each

y.

individually

of

values

In

x and

Kx

stiffness,

equal

simulated

9.4.2-5.

of force.
they have

has

spring

move in

x-direction.

the
in

shear
fig.

by the

axial..

in

analysis.

the

flexibility
9.8.

An additional

of

The joints
type

of

143.

for
those
beside
used
spring
provided
element
(12).
Such
diaphragm
simulate
elements
analysis
rectangular
in
fasteners,
frame
to
the flexibility
the
x-direction.
sheet
of
Types of elements
now be considered.
used in the model will
been

has

Beam elements

9.4-3.1.
Conventional
freedom

each

at

node

three
degrees
beam elements
with
have been used to represent
the

of
frame

members.

9.4.3.2.
Each
shear
of

diagonal

flexibility
to

sheet

truss

zonal
--Dia,
sheetinq

member has

been

to

used

of a sheet
fasteners.

width
p, where
By equating
the
to that
sheeting
of the

frame

of a width
p of the
diagonal
member, the cross
takes
the value,
b-t-G
A=_,
p-E-h

the

members representing

sectional

simulate
p is the

area

shear

the

pitch
displacement

equivalent
diagonal
of that

eff
2

(9-9)

where:
b=

depth

of

diagonal
E=

diaphragm

by the

represented

member,

the

elasticity
length
of the

Z=

panel

Geff

= effective
by equation
h and t are the

modulus
diagonal

shear
9.8,
depth

of

the

material,

member,
of

modulus

and thickness

sheeting,
of

the

given
sheeting

profile.

9.4-3-3.
These
lity

condition
They
member.
strain

cross
length

to

members
at the
have

one

been

joints

been

area

truss

have

be neglected.

sectional
of

Vertical

to

used

connected

chosen with
The vertical

equal

corrugation.
C

members

to

2tot,

satisfy
to the

the

compatibidiagonal

for
stiffness
a
member has been given
where 9,0 is the developed
sufficient

144.

9.4-3.4.

Prismatic
fasteners

The total

flexibility

by a prismatic

been

seam has
stiffness.
the value,

the

of

simulated
The cross

seam fasteners
member having
A

area

sectional

the

representing

members

of

such

seam
in

a given

axial
member takes

(9-10)

n-E-f

'E
of
member,
which,
elasticity'modulus,
fasteners
the
in
of
given
number
seam, and f. = flexibility
n=
(taken
fastener
0.15 mm/kN).
of a seam
length

in

the

Spring

9.4-3-5.
frame

As mentioned
member fasteners

spring
these

elements
fasteners.

simulated
springs
flexibility.
and

for

elements

before,

it

with
The y-component

Details

of

to

frame

fasteners

to
to model sheet
necessary
In the present
analysis
have been used to represent

is

precisely.
finite
length

by vertical
springs
have been provided
to

sheet

has been
of flexibility
Inclined
1.0 mm length.
with
the x-component
of
simulate

these

springs

are

shown

in

figs.

9.10

9.11.

9.4.4.

Simple

truss

analysis

This

to the approximate
method
method is an extension
(8)
later
Bryan
diaphragms
by
and
of regular
analysis
proposed
(11,14).
9.12,
by fig.
by Davies
the
As illustrated
modified
truss
diaphragm
trapezoidal
is simulated
by a plane
system.
Diagonal
for the overall
shear
members have been designed
flexibility

of

in-plane
carry
for their
axial
The
be evaluated
them
adding

the

diaphragm.

bending.
strain

Vertical
to

flexibility
overall
by considering
together.

to
members are allowed
members have enough stiffness

Edge

be neglected.
of

the

diaphragm

each component
These flexibilities

assembly
flexibilities

are:

can
and

145.

1.1

due

to
to

distortion

the

profile

a.

b.

1.2

: due

C.

C
2.1

: due

to

at

the

sheet

d.

C2.2

: due

to

movement

in

the

seam fasteners

e.

C3:

due

to

axial

strain

in

shear

of

movement
member fasteners

the

in

strain

sheeting
sheeting

the

to

edge

perpendicular

frame

members.

these
flexibilities
The expressions
used to calculate
9.1.
flexibility
A.
The
in
appendix
overall
shear
are presented
takes
the form:
C of the diaphragm
assembly
C=C+c+c+c+c
1.1
The
used
width

cross

1.2

2.1

sectional

to

simulate
overall
can be calculated

which,
inclination

perpendicular

9-4.5'.

diagonal

the

of

member
a diaphragm

the diagonal,
length
0= angle
of
of
=
and
dto the direction
with
of the diagonal
respect
to the corrugations.

The derivation
appendix

Ad of
flexibility

(9.12)

E-C-Cos
Z'

shear
from:

d2

Ad=P,

in

area

2.2

of

equ.

between

the

9.12

is

in

presented

A. 9.1.
Comparison

results

of

the

three

methods

the
three
of
comparison
results
trapezoidal
the
to
the
behaviour
of
used
study
methods
Four different
in figs.
9.13 up to 9.29.
is given
diaphragm
in
the
been
loading
in-plane
have
analysis.
considered
cases of
that
the
full
has
the
deflections
The comparison
shown
of
A full

frame
with

between

and
simulation
the more accurate
The

calculated

sheet
from

to
the

the

the

truss

simple
finite
element

frame
internal

both

agree

forces

have

by the

simple

model

well

method.

fastener
member
forces
given

been
truss

146.

model

by

making

given

by

the

in

distribution

forces

fastener
forces

of

fasteners

the

This

forces

fastener

by

by

as

be

will

the

using

full

frame

to

local

the

from-the

considering

may lead

given

applied

However,

directly,

model

joint.

be

can
given

calculated

is

procedure

diaphragm.

truss

simple

at

the

be

can

the

forces

trapezoidal

the

of

procedure

fastener

of

The

forces

fastener

of

method.

element
The same

A. 9.1.

simulation

for

finite

appendix

the

use

distribution

the

of

internal

only

the

overestimated
in

explained

the

values
following

discussion.

The comparison
the

three
9.29

fig.

of

the

fastener

methods of analysis
9.1.
and table

The following

forces,

in

used

observations

-calculated
is given
study

the

be drawn

can

comparison:
1.

the

from

in

the

As illustrated
load,

by

by the

either
deflections

figs.

9.21

(case

3)

vertical

9.23,

under

or horizontal
top flanges

asymmetric
(case 4),

take 'asymmetric
of thebotbom and
The figures
shapes.
between
show an excellent
agreement
the results
The largest
in
of the-three
methods.
error
the simple
truss
to the finite
method in comparison
(figs.
3%
Neveris
9.21 and 9.25)1
element
about
method
thelesspit
identical
2.

Under
(case

is

of

shapes
symmetrical
2), the full

other
3.

The
is
the

two

9.5%,
the

methods

of
comparison
shown by figs.
finite

simulation

element,
is about

method

deflection

under

in

results
asymmetric

almost
load.

at the top flange

element
and the finite
simulation
identical
for thedeflection
of
values
(figs.
Considering
9.17 - 9.21).
flanges
load

the

(figs.
results

the

vertical
frame

method give almost

the top and bottom
joints,
hinged
corner
is about
17owever,

that

obvious

are
(figs.
the

and

the

9-19)-,

identical
almost
9.18 and 9.20).

deflections

9.13,9.14,9.15
the

in

error

9.17

acting

largest

5% (fig-

under
and
error
9--13).

truss
simple
method
"or rigid
joints,
to

that

of

the

uniform
wind load
9.16.,
Compared to
in. the

full

The error

frame
in

the

147.

truss
simple
deflection,
difference
is

analysis
deflection

8.7%

almost
11% at

and about
between
the

for

mid-span
limit
bounds

two
in

significant
of the rigid

66% of
9.13).

about
(fig.

is

method

the
(fig.

corners

model,
hinged
with

that

calculated
The ratio
is

corners
model,
themid-span

37j',o for

about

The

in the
considered
The maximum
case.
(fig.
11.14),
is

loading

this

maximum
9.13).

deflection.
4.

9.29

Fig.

the axial
(case 1)

for

distribution

the

shows
forces

the

in

local

high
that
be noted
joint
and directly

forces

frame

memberfunder
wind
hinged
with
corners.

diaphragm

the

fastener

of

fastener

forces

9.1

Table

critical
to analyse
loading

shows the
fasteners,

of the
by the

comparison

forces-in
three

calculated
trapezoidal
diaphragm.

the

the

at

acting

top

2)

frame
more

the
methods

Under

(case

flange

action
It should

occur at the apex

bottom
flange.

to it in the
opposite
the sheet to
This shows the need to strengthen
those
locations
by adding
at
member connection
fasteners.
5.

and

used

symmetrical

the

acting

fastener
forces.
1,3
In
small
very
cases
produces
shear
the
4
fasteners
apex joints
at
are the most critical.
and
is distributed
In case 1, where the load
uniformly
along
the

by the

the

flange,

bottom
finite

element

force

maximum fastener
is 2.42 kN.
For

the,

calculated
same fastener

of 2.71 M,
simulation
which
gives
a value
ll%jwhile
truss-, method
the simple
by almost
is higher
2.52
M,
is higher
than the finite
of
which
a
value
gives
by
4%
Considering
the
local
only
only.
result
element
the

frame

full

fasteners
simple
fastener.

at
truss

the
finite

joint,

model
This is

by almost
fastener
higher

the

34%.
force

than

simple

the
truss

element

the

internal

forces

Under

asymmetrical
by the
calculated
finite
method
method.

element
again

3.25

of
give a value
higher
than the

finite
loading,

full

value
gives

by the
given
for the same
element
result
the maximum

frame

is
simulation
by almost
17%, while
4% higher

than

the

148.

be concluded
that
the deflections
predict

It

be used to
diaphragm
with
calculating

plane

OF THE MACE UNIT

shown

calculating
Moreover,
the
using

both

a conven-

AND THE TWO-BAY PYRADOME

analysis
of the space frame model'
The overall
shear flexibiforming
the roof
has been

shows a plan
view
the MACE structure.

analyse
the trapezoidal

panels
truss
elements.

by in-plane

simulated
in
loading
considered
model has been solved
a)

the

in

used

9.30

Fig.

of

in

that
it
advantage
can be applied
frame computer
program.

The model

to

method
forces.

the

has

ANALYSIS

used
lity

element
fastener

the

and

can

in
element'method
The full
frame simulation

forces.

finite

the

agrees
with
well
the displacements

9.5.

truss
simple
method
of the trapezoidal
The method has also

accuracy.
the finite
with

sufficient

agreement
the fastener

an excellent

method
tional

the

can

the
for

study
are
two types

hinged

Considering

The four
in

shown
of

end

The

conditions:

between

connections

cases of
9.31.
fig.

the

frame

fully

rigid.

members.
b)

is

that

Considering

such

connections

are

The model used to

9.38.
fig.
in
presented

been

in

considered
A more
be used to

the

the two-bay
Pyradome
study
structure
Two cases of uniform
load have
fig.
9.39.
They are shown'in
analysis.

simplified

model

has

been

proposed.

This

model

approximate
and easily
calculated
values
this
the
In
deflection.
the
model,
overall
shear flexibiof
been
by two
has
trapezoidal
lity
the
sinulated
panel
of
The results
diagonal
panel.
x-truss
of
members formingan
can

this
using
fig.
11-32.

program

offer

model

are

The analysis
described
in

in

given

has

been

chapter

tables

carried
(4) and

9.4

and

9.5,

and

in

the computer
out. using
(15).
the SAP4 program

149.

9.5.2.

Comparison

a.

and discussion

9.5.2.1.

The MACE unit

Uniformly

distributed

9.2

the

of

results

deflections
load

I
the

over

entire

roof

the

between
the theorecomparison
for this
tical
of the displacements
value
and experimental
in the experimental
loading
The increase
of the
case.
value
(joint
the
joint
1) compared to the
deflection
apex
of
Table

theoretical

corresponding
of the bolted
the

apex

value

of

due to movement
was probably
the test.
Such movement at
for the increase
in the
reason

value
during

connection

be the

joint

may also

the

experimental
in comparison

member,
increase

apex

presents

deflection
to

at the
theoretical

the

of the
However,

mid point

value.
member is

as the apex
restrained
the
The experimental
of
sheeting.
skin
action
stressed
deflection
of the hip
at the mid point
of the vertical
value
the two limits
in between
of the corresponding
member lies
The deflection
is mainly
deflection.
theoretical
point
at this
such
by the

due

to

the

restrained
It

is

much smaller

the

action

by the

stressed

skin

member.
of the
action

should

be noted

that

the

in

most cases
the influence

small

quite

of

Such

hip

bending

measured

adjacent

of

values

the

9.4

shows
deflection

b.

values
Horizontal
Jable

mental.,
in the

9.3

of

the

wind

sheeting.

for
and certainly
small
sufficiently
in bolted
to be significant.
joints
level
general
of agreement1be. tween

the

between
comparison
and the corresponding
the x-diagonals
model.

with
values
calculated
that
this
simulation
of
alternative
in a stiffer
the diaphragms
results
estimated

is

deflections-are

of movement
the
in mind,
Bearing
this
theoretical
the test
and the corresponding
results
by the simple
truss
the deflection,
model,
given
Table

bending

the

shear

model

values
is

of

good.

the

experimental
theoretical
It

can
flexibility

which

gives

be seen
of
under-

deflection.
(cases

load

theoretical

the
presents
deflections

analysis.

Apart

from

c,

and d)

comparison
in the two

some high

between
wind

the

cases

experimental

expericonsidered

results,

150.

now be discussed,
which will
between the analysis
and the

the

general

test

results

pattern
of
is good.

of loading,
where the
case IV
beams,
the
two
eave
opposite.
of
displacements
the two horizontal

load

In
the

level
of

value
mid
two

two

the

of

point

opposite

apex

experimental
corresponding
them is
between
The difference
indication
On the

of
other
of the

values
two opposite
equal,
load of

is

variability
hand, the difference

Id'

is

case
to
applied

are equal.
however,,
not
are,
9.5% which is an

values
almost
of the

measured
is probably

are
the

a)

at

may again
'c'
of loading,

of
value
the loaded

experimental

indicate
For
is

method

The
and
diagonal

total

horizontal

the

is
eave joints
theoretical
and
value
displacement
at the

The high increase

eave.
the mid point
of the two
three
factors:

the

for
in

the

and unloaded
some local

the

relatively
likely
always

displacement

the

of

The difference
loaded

c)

the

of

the
beams

eave

diaphragm
The previous
of the trapezoidal
analysis
the
load acting
has shown that,
along
under uniform
(wind load),
truss
the simple
flange
bottom
model has
finite
less
than
the
11%
more
accurate
about
given
element

b)

equal.

experimental
the mid point
of the
be expected
to be

of the unloaded
displacements
at
because

at the
The

results.

experimental
between
the two

values.
displacement
of
at the mid point
60%
the
than
corresponding
greater
about
in the experimental
to 40% increase
drops
mid point

calculated
3-x and 9-x

at

-where
the experimental
only
one eave,
displacements
of the
at the mid point
to the corresponding
theoretical
close

horizontal

members
However,

which

acting

members

displacement

horizontal

eave beams,
In
26%.
about

case
of the

value
two apex

the

is

agreement

comparison

corresponding
is
model,

due

mid-span
level

of
in

deflection.
disagreement

case
of the.

eaves
effect

Ic'

of

for
loading

concentrated

the
may
load.

displacements
involved
there
small
increase
to be a significant
in the
joints.
to movement in bolted

between

the

theoretical
presented

in

experimental

deflections

by the
values,
given
9.5.
table
Again
the

x-

151.

the

comparison

reflects
compared to

model

action
deflections,
load

the

limits,

two

half

be seen,
during
measured

of

However,

This

displacement.

could

9.32

9.5.2.2.

also

fastener

the

MACE unit

The failure
explained
are those
the

panel

in

section
involving
it

structure
It has
(section

finite
with
The individual
calculated

comparison,
lie
in between

at
the

corresponding
be because
the bolts

the

shows

under

results
of
is similar
However,
more

of the roof
is much
theoretical

of the
the high

the
to

unloaded
internal

x-diagonals
discussed
that

as in

stiffness

the
than

other
the

model.

Critical.
of

test

half
unloaded
location
a given

The pattern
of agreement
model.
truss
for
the
model.
simple
above
the
loading,
model shows
of
cases
truss

the

by friction
stick
this
case.

in

produced

simple

the

to

tended

Fig.

from

the

in

displacement
experimental
limit
lower
the
to
of
closer

apex
forces

model.

can

the

joint

the

The
theoretical
corresponding
values.
location
in
of the displacement
at a given
is close
the roof
to the mean value
of the

bounds.

limit

of

the

of
value

experimental
the loaded
two

It

structure.
displacements

the

stiffness

of the deflections
shows, the comparison
load.
The comparatively
of the asymmetric
in the case has shown that
the
obtained
the most critical
loading
represents
condition

the

that

truss

simple

the

9.32

Fig.

asymmetric
for
this

in

load

Asymmetric

c.

under
large

the

increase

is

forces

modes of
9.2-4.3;

a diaphragm
however,

fasteners.

Thus
to

essential

been

shown
9-4.5)
that

in

the

internal

from

assembly
the most

the

model

have been
likely

to

modes

check the safety

the fastener
calculate

of
forces.

the

the

method for
element
fastener
forces
from

calculated

analysis
of the trapezoidal
I
truss
simple
method agr ees well
the fastener
forces.
calculating

of
force

the

MACE unit
distribution

roof
given

have

been

by the

152.

space frame model by making use of the distribution

by the finiteelement
fastener
forces
given
analysis
trapezoidal
panel.
Fig.
apex joints
by uniformly
that

9.34

shows

asymmetrically.
of the
stability
action
forces

of

the

at
is

them

in

the

most

model)

corners

the

forces

that
obvious
is dependent

when the
under this
on the

is
roof
load

stressed
that
high

can be seen
The
at the apex joints.
is between
fastener
critical
(rigid
0.70
kN
corners
and
It

fastener
is
an individual
the comparison
of the displace(fig.
loading
9.32)
has shown that
to the corresponding
are closer
by the rigid
model than to
corners

capacity
However,

asymmetrical
displacements
values

is

structure

The ultimate
M.
4.0
approximately
experimental
theoretical

fastener

It

model).

under

the

sheeting.
roof
in the fasteners

occur
force

of the
value
4.23 kN (hinged

ments

fasteners

loaded
When the structure
2
load of 0.81 kN/m
It can be seen
the stressed
is of a'
skin action

loaded

skin
local

the

in

nature.
Fig.

the

forces

to

loading

such

the

shows

and opposite
distributed

under

secondary

9.33

shape of
of the

given
by the hinged

of

Nevertheless
model.
given
values
forces
fastener
these high local
show the need to strengthen
the
by increasing
the sheet to frame member connection
in the critical
fasteners
to every
corrugation
regions.
the

The

critical

are
stressed

shown

loading

fastener
in

figs.

corners

forces
9.35

and

under
9.36.

horizontal
It

can

wind
be seen

that

is

to the lo. aded panels.

confined
In case 1c'
when the horizontal
wind load is acting
in the most critical
fastener
along
one eave beam, the force
(rigid
2.36
kN
3.22 kN (hinged
is between
model)
and
corners
the

skin action
of loading,

The. critical
model).
they represent
nevertheless
to increase
is recommended
corners

in

the

critical-locations.

fasteners
a potential
the fasteners

are

thus

safe

but

it
weakness
and again
to every
corrugation

153.

9.5-3.

The two-bay
Table

deflections

Pyradome

9.6

the

presents
and the

between

comparison

the

experi-

deflections
theoretical
mental
corresponding
values,
for
in the analysis.
the two cases of loading
considered
Apart
deflections
from the high experimental
of the valley
beam, the pattern
between the experimental
of agreement
and
theoretical

deflections

deflection

mental

of
bolts

is
the

The increase

good.

of

the

experito

beam was probably

due
the two beams of
which were connecting
The fastener
forces
can be calculated
valley

movement of the
the valley
member.
following
the same procedure

used

for

MACE building.

the

CONCLUSIONS
in this
of the work reported
chapter
has
of describing
model capable
a theoretical
the behaviour
roof
structures
under vertical
of the hipped
Under certain
the
loads.
cases of loading,
and lateral
depends entirely
on the stressed
of such structures
stability
The objective
been to present
1.

skin

of

action

the

trapezoidal

sheet

which

panels

form

the

roof.
Three

2.
behaviour

of

The full
loading.
of in-plane
be in excellent
with
agreement
the displacements.
predicting
overestimated
truss
simple
finite
and

the

method in
forces.

of

the

truss
simple
the hipped
roof
The

a good

predicting

the

the

structures

critical
internal

with

fastener
forces

the

study

cases
to

proved

finite

the
agreement
with
both the displacements

been

simple
modeled using
shear flexibility
of the
The comparison
between

roof.
deflections
and experimental
model can be used to express

panels

theoretical

using

shown

has
The MACE structure
the overall
to simulate

panels
trapezoidal

4.

has

frame

been

method in
element
However,
the method slightly
On the other
handpthe
forces.

fastener

method

element
the fastener
3.

used to
under different
has
simulation

have
methods of analysis
diaphragm
the trapezoidal

given

sufficient
forces

shown that
the behaviour

the
the
of

accuracy.

have

by the

has

truss

been

space

calculated
frame model

of

154.

the
using
finite

MACE building.
the

The procedure

distribution

element
(Appendix

roof
the structure

shape of
of the trapezoidal

analysis
A. 9.2).
is

the.
strengthen
highly
stressed

The calculated

loaded

sheet
regions

of calculation
fastener
forces

to

is

based

on
by the

given

of the hipped
panel
fastener
forces
when
shows the need to

asymmetrically
in
frame member connections
by increasing
the fasteners

the
to

every

corrugation
5.

A similar

has been carried

out on the two-bay
study
The theoretical
deflections
calculated
agreement
model have again
shown adequate

Pyradome

structure.

with

the

proposed

with

the

corresponding

experimental

values.

DIAPHRAGrl
LAYOUT

Fig.9-1Diaphragm Action
In Pitched Roof Portal Frame Under
Vertical Load
ZZ2
ENDGABLE
DIAPHRAGM
LAYOUT

I--,

Fl-q 9-2 DIAPHRAG

ACTION IN FLAT ROOF
PORTAL FRAME UNDER
SWAY LOAD

"

U.D.L OVER THE COVEREDAREA

BIAPHRAGM
LAYOUT

LHU>) )UIIUN

Fig9.3Diaphragm

U.D.L. / UNITE LENGTH

iiiiiiii;
IiI

U.0.1-/UNITE LENGTH
IiijiiiiIiiiiiiiiiiIiIIi;

PURLIN

11

4-c2i

Action in Folded Plate Roof

:t1

ENDGABLE

",Z-

1Ux

EHE MEIIBER

AB
(A) SHEETING SPANNING PERPENDICULAR
TO SPAN

(B)SHEETING

SPANNING PARLLEL TO
SPAN

Fig 9.4 Basic Diaphragrvi Arrangaments

IE

(U

0
CL

Ln

(A

cn

Cl
CU
C3

Ln

CA

cw

rn
x
M
7---

10
x

----\

LLJ

x:
%6.-

CD

17-7

1'-----7

rn

Ln
'ri_
Li
C

Ln

CZ)
Ln

rX
C)
Ln

L_____j L---

x
Ln
C_

H!
CY%
Co
r-

00'99LE

Sjl
E
-4

ol
cy

E
C)
C3

E
cn

ru
(U

LL-

C>

C=
(1)

cn
.
iZ-

'

'0
cm

1]
V)

cy cn
E C:
cu=
LLI

M:
LLJ
-j
U-j
LLJ
I-0-4

LL

cu

ci

cu
'cu

cm

GJ

E cu
CU
cii
LLJ

CJ

V)

V)

an
M=
75 c
cu
En V)
C= CIJ
1
4.ru
Li
(U

C3CU

cct

r-n
l=
cm
:i

ai
in
cu

cn,c2c

2 ei
L.! =

CU

-6c=
cu
r=

EE cu
Mi
-22
cn
r=
m
(1)
Co

V)

(U

CU
"
C- =
vi
vi

cu
r=

L2
r=
M

cu 0)
mvi
r-n cil
2.
.
C- r
f=
,

ca.

tn

Z)

-i

>-

0'9SLE

LM
CD
c2
ui

U-

Co

U-

C3

C:)
vi

C:)
CD

UC:)

4: 2

C: )

C:)
CD

fn

E
cn
ru
CL
(0

CD

C>

Z)

-.
vi

cu

C3

ULL
CR

:C
do,

C
LZ

0
<
z
LAj
LLj

uj

LO
tA

u-

.;
"Lj

=W

%AJ
I

(U

V)
=

LLJ

ui
2:

uj

F--

U.
&A-

c12 CID p-

(Li

-j
uj

j
2:

<
LAtAi

'93

qAj
,=

LLJ

Ix
CL
LLJ

C,2

=
LLJ
cg2
=

--J
<
=

vi

C3

tD
-C

ti

jj
tAj

>
eg

tA
cr

CD
C>
c:: b
10
(n

L&J
Z

LLj

_j

LA-

%A.
or
L&A
LLJ
>

CD
LL.

C)

LLJ
=

LW

ix
Z

"j
=
LAj

u. 1

CL. V)

LLj

z
uu-

(/I
cc

=
vw

LLj
z
u-

cn
=

w
LLJ
c12

LAJ
=x

V,

--J
-<
I'-

uj
ii

LAJ
=m

C) V, C3
Z

Uj
V)

Lj
uj

LAJ

V)

LL J
L,

C) uj = uj
I'cl: V)
V, 0ul uj U- uj LL-

cr- C) ex c)
LA .19

>..

cc

Zero Size

-Ix
Go

kx

A4
V)

P4

iy
Fig.9.9 Finite Element Simulation Of Fastener

(4) SPRING ELEMENTREPRESENTS

Kx FOR SHEETTO FRAME FASTENER

MAIN

(2)PRISMATIC MEMBERREPRESENTS
THE TOTALNUMBEROFFASTENERS
IN A SEAM

FRAME
I ---_(6)VERTICAL MEMBER

(5) DIAGONAL

Zerp Sige
Element Represents
Ky For Sheet To Frame Fastener

Fig.9.10Details Of Joint (B)

(2)

Ze6 ize

Fig. 9.11 Details Of Joint ( C)

C'4
(A

LL

0199LE

%4C)
V)

.a
0

C7,

aj%o
a)
Lcn
CD cu
C)
C)
C)
2-1

c:!
CD
CD
10

rn

CT
CL

cu
0
0
IA
ru
C:
cli

C?

LL.

cu
CL

E
(A
CI-4
q0%

V)
(LJ
V)
cu

Lei

-0
0

c:>
vi
-4a
(Li
r=

1
Ill

LU

cu

cu
r=
c:
(U
cu
(Li
>
(A
Z:
ci

-kc

L-

-0

(U
2
tLLJ LL.
E
ro
(Li

c12
-v-

(U
C M
0.

E-=
cii

cz m

01
C: )
CZ>
o
m

-j

L.AJ

im
C)

zC)
LLJ

1:

<

LAJ

LAJ

LAJ

U-

LAJ

t=:

III"1

a-

-j

-'9
L,

>1

cli
ro

CO
q4C:

r-

r-n

L-

(Z)
9-)

-0
rG

m
iz

CD
Ln

CD
LA
ffi

C
Lf%

C)
LM
's

cs

-i

z0

LW
c2

8
x

-J

LAJ

-J
-J

0.
Z

U-

C>
'o

(1)
m

CD

0
x
CU

>(
uJ
0

ej

EX
(U
Li
LA <
cl

fu

C
0

(U
C)

V)

t_j

ICD

CY%

01

iz

C
C
rvi

=0

Lf%

Cb

Ck

Ln

C31

4=0

Ln

CD,
4
=l
e

20 KN Im

M.M

3.00
2.5G
FINITE ELEMENT
2.00
FULL SIMULATION
1.50
_.

1.00

-SIMPLIFIED

MODEL

O.OQ

Fig. 9.15Wind Load Case.,In-Plhne Displacements Of Top Flange

Corner Joints Are Hinged)

2.60KNIm

- MA

ELEMENT
AND
-FINITE
FULL SIMYLATION
_-SIMPLIFIED

MODEL

-0.50
0.00

Flange
Of
Top
Dispiacements
tanel
Fig.9.16WindLoad Caseln-P.
(Corner Joints Are Hinged)

ZU
C,
cu

E
tu

.2Z
cu

tu

zbd

v?

I:

x
cu
c2.

ei
En
c
6.
< - "q
Um
LJ
<

LLJ

c2
-...
ci
-60

hit, c12 En
rC)
<
(IJ

-1

,W
0

CD
m

CD%

c
iz

C=l
9-4

9: 5.

cm
A

CD

Cb
10

E=- tA
-d0)
Li

E
.

-t2
r-

En

<
Li
ci

U-

LLJ
C>

8,
Cc

C=i
%4.V(A
-0
cs
0r
-1

LLCU

L-c:

r.

C)
tA
(A
cu

E
CL)

7%

CGi
Li C=
0
t-j
>CU CL
(A
is
0

C)
d)

co

En
U-

'.0

EI

--a

10.0KN

10.0KN

0.00

-0.10
-0.20
_Q30

Finite Element

---

Full Simulation

---

Simptified

Modet

-0.40
-050
-0.60

Fig.9.19CaseOf TwoVertical Loads 10.0KN Each At Apex Joint

In-Plane DisplacementsOf Top Flange,
(Corner Joints Are Fixed.)
10.0KN

KN
10JO

-0.10
-0-20

Finite ElementAnd
Simplified Model
Full Simulation

_o-3o
-0-40
-aso
-0-60

Fig.9.20CaseOf TwoVertical Loads100KN Each At Apex Joint

In. Piane DisplacementsOf Top Flange
(Corner Joints Are Fixed) ,

LU
LLJ

. b.C:
0

x
Gi
rL

<
<

C>
'r-

CU

cn
a
ro
E
0-0
CL)
cc

10

C)
W
fu

a
CJ

E
aj
CU W,

cn
C:
cu

S
> Es
ca. --,
- Lcu
a. T
C)

o
ai
W
ru

C5,
ir

4=0

C: o

Ln
c),

rn
C: o

Co

C: o

C>

CD

C
4=0

c: b
rn

c>
-t

C: ).
Ln

c:i

cu

4) t-j0
ro

-j

LLJ

M
C)

C)
LLJ

LLJ
Lhi

U. j
CL
LL.

U-

. a-

x
cu

c3-

(IJ

<
M
EZ

<
r_ E
ir0
C>.
ID
(U
C:) C) X
e- Co 11
=
M
C)

%bc:: >

OJ

<
v)
cu
Ecu

>

ti

C: )
CD

Clc
ci,
iz

C)

C>

4=

C>
V-

ICD

Co

C3,

Co

C)
Ln

tn

10.0KN

oc

-0.50
-0.40
-0.30
-010

0.10

Element

-Finite
----,
_.

Full Simulation
-Simplified

Model

0.00
0-10
0-20
030
0.40

-050
0.60

Fig.9.23 Case Of One Vertical Load 10.0kn At Apex Joint

In Plane Displacements Of Top Flange
(Corner Joints Are Hinged)

mm
KN
110D
-0.40
_MO
x
-020
-0.10

QOO

ElementAnd
-Finite
SimplifiedModel

-0.10
----Full

Simulation

-0.20
-0.30
-0.40

Fig.9.24CaseOf OneVertical Load10.0kn At Apex Joint

In-Plane DisplacementsOf Top Flange
(Corner Joints Are Fixed).
-,

-i
C: )

CD

E3
ui

Z;

2:

LL.

Z;;

CU

cn
X C=
CU fu
C3-

<

E
C; -0
(U
En
CID

C) C)
fu
0

C=
(u

(1)
0
a'-

--J E C:
cu
rq
V) CL
tn (IJ
C:

(U
V)
Ln
a,;
ch
iz

c:.

E
E

<

<=)

rn

C).

C>
Cli

C)
MI

CD

-t

Co

Lf%

"

CD

10.
.

CD

c3
c2

H
0

vi
-i
m

Ci-

(U

-0
x

C)

Co

U-

CD

V)

C
-3
j; j

F=
0
(U -%
Li
CIJ Lrg

C: )
CU
V)
ru
Li
,0-

e-

C:)
ir

c:i

c: b

ri
C>

1.1

C: )
rA

C=i

Cb
C)

c22
cz

c>

4: b

C:
rn )

2111-

c:
b
t
,

c>
Ln

cs

-0.-

CU

10.0 kn

mm
0.60
L

A.so
0.40

-030
-0.20
AU

Element

-Finite

Simulation

----Full
--Simplified

Model

).00

-0.10
_020
J. 30

-050
-0.60

Fig.9.27CaseOf A Side Load 10.0Kn At Apex Joint

In-Plane Displacements Of Top Flange
(Corner Joints Are Hinged)

mm
-0.40

100 Kn

0.30
T

-0.20
-0.10

0.10

Element And
_Finite
Simplified Model
Full Simu tation

-0.20
-030
-0.40

0 Kn At Apex Joint
Fig.9.28CaseOf A Side Load 1U.
In-Ptane DisplacementsOf Top' Flange
(Corner Joints Are Fixed)

2.42 2.36
2.0
980

-1.0

950
\1

12.10

KN Im

L-ao
SCALEOF FASTENER
*FORCES
( KN

16-70Kn

9.80

10.10Kn138

1.41

WFINITE

ELEMENTANALYSIS

16.40KN
KN'
I

5.90 KN
KN ,-

Fig. 9.29 Fastener Forces DueTo Wind Load

(Hinged Corners Model)
-

L
Em

Ln
U-

LL.

(U

r-

C: )
C>
e-

-t
o
c2

en
rn

tr%

trt

Co

e4

rrlb

C%

CD
C>

&A

M
rn

c21

Co

(21

r4

C>

c7.

rd2

U-b
C>

10
rr,

o
rt

CD

un

o
c: p
c:-:;

cy
1:: >

LA
r

c>

a!
U-

-E;

C%d
C: )

--

c:

CD

LA-

r0

tri
C>

0
Co

iz

rn

c
r_
f
c3
4-1

CL,
rn

cu
Q

is
t-i

-I

4u
a

E
Z

tn
M

C>
-

:z

2Ne

c:!
4)

914
%0

c>
C>

r4
r-e
c>

l
-0 *
- c>

cc

75
r=

C:)

C>
C>.

01

ca
rn

. (D

VI

rm

--t

:
IMD

C: )

m.
-o

r4
r,

c>

en

Li-

-t

<".

o
Ln

CU

Li.

u-

c
c: b -

ci

-?

clo

ro

c>
C>
c=i

c>
(D.
c>

VI
.

C>;
cD .2
LA
CD

cr

cu

Co

C>
g,
c
r-

CD

-t
c: >
g

-r

C>

o t

C=b

-t

Lm

c>

cp.

4.m

c>

c>

CD
Ck
C: )

en

1 m
LM
r4

C>

tu
r=

rr

cm-

<->

Z>,

C],

f14

ci

c>

C>

CD

C>

r-

-t

c,

1
U-b
r

ci

c>

c2

CD

C: )

-0

r-

e14

r4

Ul%
Cl%

Co.

trt

CO

CO

C: )

rIl

Ln
4: )

4A
ei

r4

(=

cu
c
cu
-i.
vi
m
LL

C>

r.
4
V%

C: )

(D

U-

12

--t

gt

tn

ro

Ln

CD

:3

10

C>

.-E
U-

CY,

(Z
C:)
c>

-r

C>

C>

c>

C: )
c>
- .

C:)

r
r

(=i

Co

cr

-0
cu
x
U-

cy,

c:Dr
b

rrt
cm.
C>

=;

C>

Im

cm
Lr%

c;

C=!

CD.

* 1 c::t
ci 1 CD

c::
CD

1=;

(D
C>

C>

f14

M
0
d=

Co
U-t.
C: )

en

ul

c7.

ei

rn

-t
LA

,
,

V%

CR

C: )

Q%
(D

c>

U-

ru
: 3-

(U

:=

c>
t
14:
rA

-C
tu

sE
UZ 2

cy,
c . Ul
c: f
=

.4>
c: b

r
r-

cc3
C>

1- 1a
x

-t.

10.

CYun

qU-

tAi
vi

cm
C)

C-4

rn

-t

Lri

%o

LL

LL

LL

u-

U-

ci

c2

CD

rIl

co
b

c:

1
(M
Z

.M
ru

t
;Z -.:
r-

Co

LL

U-

-Main
--Members

Frame Members
Representing The
Sheet

Cotumns

DimsIn M.M.

3600

3600

3600

Fig
The Model Used To Analyse The MACEUnit
-930

(b) Asymmetric Load

1.215KN/r)

(cL)UniformlyDistributed Load
2
0.81 KN Im

(d) Wind Load On*BothSides

(C)Wind Load On One Side
KN/m/Side)
(1.23
'
KN/m)
(2.46
Fig.9.31 Cases Of Loading
On The MACEStructure

ly
6"

"

EXPERIMENTAL
DIRECTION
RESULTS

JOINT

DISPL.1Nmm

THEORETICAL
HINGED

12.60

FIXED

9.06

7.50

9.26
10.50

13.90

12.60

10-78

0.95

0.65

0.19

4.17

3.50

3.35

0.11

Q2-8

0.25

oc

0.92

1.05

1.02

535

Table9_2Experimental And Theoretical Displacements

Un'derU.D.L., 0.81- KNIm2

V.

CASEftC WIND ON OLNESIDE"

JOINT

DIRECTION
EEXPERIM

CASE"DIAND ON BOTH SIDW

THEORETICAL
HINGED

FIXED

THEORET
ICAL
U(PERIME

HINGED

FIXED

oc

1.80

223

2.14

25

3.54

-273

oc

243

172

1.68

623

3.92,

295

95

614

k2l

494

3.92

2.95

y,

-0.15

_Q50

-0.55

0.0

1.50

1.62

1.57

-0.05
2.17

1.90

202

1.95

1.75

1.82

176

cc

3.35

4.94

3.23

3.54

2.73

10

42
Z.
2
-OP

0.0

0.0-

-0.10

QO4

-Q04

-0.0
1.82

176

Table 9.3 Experimental And Theoretical Displacements Under Wind

Load Cases

1K

('4
'0

1.46

0.20
"'(0.02)

(0.12
(651)
(837)

Asymmetric Load
0.215 Kn/m2)

U120)
(-1 3.2)

552)

(345)

(39.9)

DISPL IN M.M

(-21.0)

-Ve=UPWARD
(VERTICAL)

7.79
,(-12.5)
(12.41
(7-15)
1.01
\-(0.10)

0.57
0.41

In

Exper
imenTaL
cL.
1
ce

cy

0.63

024
(0-24)

M
(431)

Ui

Cq

(-15.15

0.73
[
3)
.2

N
(781) C,I
!,;

cr,
r1l
LA
(-20.95)
(09)
(-10.7)
5.39
(26-851
(-10.7)
1
(6.79)
(34.54)
(026.8B0)
12378)431)
1

(6 6)

CS

(-42.7)
(-BS.08)

390)

14.90
(-39.0)
(6.86)

(57.6)
V1
-1
(S7. 6)

(98.8)

(7.81)

(58.3)
0.90
I\\-(0.19)

0.69
LJL4) I
_
_,/

0-98
(-2 8)

0.56
014

O.82
0.23)
CID

Hinged Joints C).

Fixed joints
b.Simple Truss Model'

032

0.62

(0

(021)
(233)
93
.
(6.15)
(20-84)

(28-581

0.69

(0.21)/

(-12.34)
12.34

(-7.44)

O.S2
(0-13)
(3.32)

(6.3)

(
(-16.28)

(49.87)

(6.15)
(6

(933)

Fixed Joints

(-40-17)

(-37. 02)

3.99
1
t-744)

c'!

2
((20.84
0)
(18 21)
(2.93)
0.67
A18L7,
LO
-

<D
K

0.77
19)
.-

t-80-47)
(-80.47))
14-08
14.08
F
(f- 37-02)
37.02)
(6-43)
(6.43)

_ (49. S7)
( 3.32)
4679 )
0.77
\LO-21)

0.93
(9.25)/

Hinged Joints

C.Simplifi ed X. Dictgonats Model

Fig-9.32Comparison Between Experimental And Theoretical
Displacements InThe Case Of Asymmetric Load

Y
8

I
JOINT
X-DIAGONALSMODEL

DISPL.IN mm

EXPERIMENTAL
THEORET
I CAL
DIRECTION
HINGED
FIXED
RESULTS

12.60

B-S7

8.20

7.50

4.42

3.91

13.90

8.93

B.S7

0.9s

0.10

0.10

cc

4.17

3.15

2.50

0.11

0.27

0.26

cc

0.92

0.97

0.93

Table.9.4ComparisonBetween Experimental Results And The

2
Results Of The X- Diagonals Model (U.D.L 0.81KNW

CASE'C"41NDON ONE SIDE"

JOINT

DIRECTION
EXPERIME.

CASEVVIND

THEORETICAL

ON BOTH SIDES'
THEORETICAL

EXPERIME.
HINGED

FIXED

HINGED

FIXED

oc

1.80

2.34

2.27

2.65

2.59

Z27

oc

2.43

1.82

1.67

6.23

3.69

3.47

9.85

5.67

5.27

4.94

3.69

3.47

-0-15

-0-41

1
-OL

-Q05

-0-14

0.0

1.50

I.S7

1.58

2J7

1.77

1.77

3c

1.90

1.70

1.96

1.75

1.77

1.77

CC

3.35

3.56

Z27

21+2

2.59

2.27

01

y1

-0-10

0.0

-0.02

-0.02

1-002

0.0

Table.9.5 ComparisonBetween Experimental Results And The Results Of

The X-Diagonals Model(Wind Load Cases)

U.D.L = 0.81Kn / m2
o
o
oo
ooo.........
ooX

Fig. 9.33 Fastener Forces (In Kn) Due To U.D.L

Of 0.81 Kn / M2

Qos

0.015

Asy mmetric Lo ad = 1.215Kn

Q09
.Ln-) -0-74

0.03

Fig.934 Fastener Forces(In Kn)Due To Asymmetric

Load Of 1.215Kni m2

2.46 KNI

0.001

0.14

192

,0

2.15

1006

Fig.9.35Fastener Forces(In k'n)DueTo Wind

Load Case'c'
0.07

Q07

0.9

1.23K Im

06

1.05

0.08

0.08

Fig.9.36 FcLstener Forces (In kn)Due To Wind

Load Case V

1.23KNIm

I1

z
-J

.a
(A

-6.-

tn

V)
C)
0.

i cc

ui

G1

Ow
,

-13

ER

CL

LAJ

tA

IE

ro
-4.
ai

LA
E<
I-E
to

c: 3

LLI
133

trn

an

LL
Ln

In

Lr%
C-

If

LM
C-

4A
Ix
LLA

vi
Z:
C4

E
E

Li

GD

cm

uJ

=
I-

KA

1
LU
V)

&A
x

LA63

cli
.

a2

:
co

LAJ

LAJ

Ln

rn
m

rl-l

MAIN FRAMEMEMBERS
REPRESENTING
THE SHEET

-MEMBERS

COLUMN5
INTERMEDIATEPOSTS

10000

Yo
Fig.938 The Modet Used Analyse The TwoBay Pyradom
y

\AREA(B)

AREA(B)

77

A ZEAIA)

11

00

Illkv
2
at
11
111
111
The
E_ntire
AreaKn1m
Over
Load=0.80
Distributed
v
cL-Uniformly
b-U.D.L. a8O Kn/m Over Area (A) While

Area (B) Is Loaded By 0.40 KNO

Fig.9.39 Vertical Loads On The Pyradome
DIMS IN MM

--- -

T-

16
A27
39

L-/
'12 14X
'12\!
11
6 '11
lo 13
10
58 a 10
5
,
7

49

CASEW

A,

CASE(B)

EXPER.

THEORETICAL
RESULTS

RESULTS

HINGED

FIXED

RESULTS HINGED

FIXED

k06

EXPER. THEORETICAL
RESULTS

12.0

15.95

8.31

5.50

7.0

728

7.19

3.50

7.0

7.96

747

k5 0

-Q94
6.24

4-

12.0

16.01

5.17

1ZO

1598

5.19

90

816

752

9.0

1Q89

945

8.50

10.80

6.68

8.0

12.01

9.50

9.0

922

7.77

50

9.18

55

8.50

9.63

5.35

850

11.03

6.Bl

2.0

0.17

0.16

1.50

0.17

0.17

10

1350

9.28

7.01

13.0

9."

7.18

11

19.50

11.69

9.86

18.0

1064

8.B

12

lz5o

9.20

6.81

11.0

7.30

5.30

13

9.0

8.00

8.02

8.0

11.Bo

9.65

14

9.50

lQol

24

550

6.76

15

12.0

1600*

5.40

16.50

15.97

5.31

16

20

2.04

1.20

2.50

1.26

0.61

8.00

2028
5.44

Table. 9.6 Vertical Deflections-IJ. D.L. And Asymmetric Load

For The Two- Bay Pyradome
f.
in
Be
m.m.
.
o-ve Upward

155.

APPENDIX

The

A. 2.1.

stiffness

second-order
This

cross
second
Renton

sections.

matrix
The

is

valid

for

submatrices

order
stiffness
(9),
are,
EA

matrix

given

skew and doubly

all
K in

matrix

by Renton

,a
eq.

symmetrical
the
of
a
12 .....,,
44
by
2.68 as given

00000
12EI

[a
12]

Z35

6EI

022

6EI
009,3

p1l]

= -P33]

'

[14]

r12]
:--

EI

Y03

04

[a23]

in

which,

)2

y P4
2,

and

0
2EI

00

91.2
6EI

z
13Z2

x0
2EI

24]

(A. 2.2)

P43]

-6EI z
2

=00

4EI

=-

0000

x0

22]

1)502

aa
44]

22]

(A. 2-4)

156.1

2
c0tz+

22

21

2-25-z

values

T=

01,02

xx0

and

T2=xZ

(A. 2.

EI

from

05 can be calculated

and,

AIL

GJ -P2
c

34241

12z

and

...,

Px Z2
EI

7- 12(1-4)

(A. 2-5)

where
tanh

2
pxz2

(GJ

liz
W

uxt

Pxio

2)

(A. 2.7)

157.
APPENDIX
--------------------SUBROUTINE

A. 3.1
BIMOM

------------------------

SUBROUTINE BIMOM(ID
DOUBLE PRECISION FA, EA, CA, DA, FEA, FCA, FDA, ECA,
1 EDA, CDA, F1, El, Yl, Zl, X2, X3
DOUBLE PRECISION CFM, FM, CC(110)
DOUBLE PRECISION Rl, R2, R3, R4, R5, R6, R7, R8, R9, RlO
DOUBLE PRECISION SUMR1, SUMR2, SUMR3, SUMR4, SUMR5, SUMR6,
1 SUMR7, SUMR8, SUMR9, SUMR10
DIMENSION FA(110), EA(110), CA(110), DA(110), FEA(110),
1 FCA(110), FDA(110), ECA(110), EDA(110), CDA(110)
R2(110), R3(110), R4(110), R5(110),
DIMENSION Rl(110),
1 R6(110), R7(110), R8(110), R9(110),
RlO(110)
DIMENSION SUMR1(110), SUMR2(110), SUMR3(110), SUMR4(110),
1 SUMR5(110), SUMR6(110), SUMR7(110), SUMR8(110), SUMR9(110),
2 SUMR10(110)
COMMON/FUNCT/FA, EA, CA, DA, FEA, FCA, FDA, ECA, EDA, CDA, E1, El
1 Yl, Zl, Xl, X3
COMMON/FACTOR/Rl, R2, R3, R4, R5, R6, R7, R8, R9, RIO
COMMON/SUMM/SUMR1,SUMR2, SUMR3, SUMR4, SUMR5, SUMR6, SUMR7
1 SUMR8, SUMR9, SUMR10
COMMON/FORCS/FM(35000),
CFM(35000)
Do 10 I=1,81
Xl=0.0125*(I-1)
X2=Xl*Xl
X3=X2*Xl
Fl=6.0*(X2-Xl)
El=-Fl
Yl=(-3.0*X2+4.0*Xl-1.0)
Zl=(2.0*Xl-3.0*X2)
FA(I)=Fl*Fl
EA(I)=El*El
CA(I)=Yl*Yl
DA(I)=Zl*Zl
FEA(I)=Fl*El
FCA(I)=Fl*Yl
FDA(I)=Fl*Zl
ECA(I)=El*Yl
EDA(I)=El*Zl
CDA(I)=Yl*Zl
IF(I. LT. 2)GO TO 8
CC(I)=-CFM(14*(I-1))
SL=0.50*(CC(I)-CC(I-1))
Rl(I)=(FA(I-1)*CC(I-1)+FA(I)*CC(I)+0.50*FA(I-1)*CC(I)+
1 0.50*FA(I)*CC(I-1))/240.0
R2(I)=(EA(I-1)*CC(I-1)+EA(I)*CC(I)+0.50*EA(I-1)*CC(I)+
1 0.50*EA(I)*CC(I-1))/240.0
R3(I)=(CA(I-1)*CC(I-1)+CA(I)*CC(I)+0.50*CA(I-1)*CC(I)+
1 0.50*CA(I)*CC(I-1))/240.0
R4(I)=(DA(I-1)*CC(I-1)+DA(I)*CC(I)+0.50*DA(I-1)*CC(I)+
1 0.50*DA(I)*CC(I-1))/240.0
R5(1)=(FEA(I-1)*CC(I-1)+FEA(I)*CC(I)+0.50*FEA(I-1)*CC(I)+
1 0.50*FEA(I)*CC(I-1))/240.0
R6(I)=(FCA(I-1)*CC(I-1)+FCA(I)*CC(I)+0.50*FCA(I-1)*CC(I)+
1 0.50*FCA(I)*CC(I-1))/240.0
R7(I)=(FDA(I-1)*CC(I-1)+FDA(I)*CC(I)+0.50*FDA(I-1)*CC(I)+

158.
1 0.50*FDA(I)*CC(I-1))/240.0
R8(I)=(ECA(I-1)*CC(I-1)+ECA(I)*CC(I)+0.50*ECA(1-1)*CC(I)+
1 0.50*ECA(I)*CC(I-1))/240.0
R9(I)=(EDA(I-1)*CC(I-1)+EDA(I)*CC(I)+0.50*EDA(I-1)*CC(I)+
1 0.50*EDA(I)*CC(I-1))/240.0
RlO(I)=(CDA(I-1)*Cr'I-1)+CDA(I)*CC(I)+0.50*CDA(I-1)*CC(I)+
1 0.50*CDA(I)*CC(I-1))/240.0
SUMR1(I)=SUMR1(I-1)+Rl(I)
SUMR2(I)=SUMR2(I-1)+R2(I)
SUMR3(I)=SUMR3(I-1)+R3(I)
SUMR4(I)=SUMR4(I-1)+R4(I)
SUMR5(I)=SUMR5(I-1)+R5(I)
SUMR6(I)=SUMR6(I-1)+R6(I)
SUMR7(I)=SUMR7(I-1)+R7(I)
SUMR8(I)=SUMR8(I-1)+R8(I)
SUMR9(I)=SUMR9(I-1)+R9(I)
SUMR10(I)=SUMR10(I-1)+RlO(I)
GO TO 10

8 CC(I)=O. O
SUMR1(1)=O. O
SUMR2(1)=O. O
SUMR3(1)=O. O
SUMR4(1)=O. O
SUMR5(1 ) =0.0
SUMR6(1)=D. O
SUMR7(1)=O. O
SUMR8(1)=O. O
SUMR9(1 ) =0.0
SUMR10(1)=O. O
10 CONTINUE
WRITE(8,100)SUMRI(81)
100 FORMATUAH
RR1, E15.6)
WRITE(8,101)SUMR2(81)
101 FORMATUAH
RR2, E15.6)
WRITE(8,102)SUMR3(81)
102 FORMATUAH
RR3, E15.6)
WRITE(8,103)SUMR4(81)
103 FORMAMAH
RR4, E15.6)
WRITE(8,104)SUMR5(81)
104 FORMATUAH
RR5, El5.6)
WRITE(8,105)SUMR6(81)
105 FORMATUAH
RR6, E15.6)
WRITE(8,106)SUMR7(81)
106 FORMATUAH
RR7, E15.6)
WRITE(8,107)SUMR8(81)
107 FORMATUAH
RR8, E15.6)
WRITE(8,108)SUMR9(81)
108 FORMATUAH
RR9, E15.6)
WRITE(8,109)SUMR10(81)
109 FORMAT(//5H
RR10, E15.6)
RETURN
END

159.
APPENDIX A-3.2.
Coefficients

Fact.

Kb and Kt

bl

b2

of

the

b3

geometric

matrix

b4

ti

t2

t3

t4

0.01

0.5999

0.0652

0.0501

0.0166

0.30

0.0334

0.0250

0.0083

0.05

0.5995

0.0652

0.0500

0.0166

0.30

0.0333

0.0250

0.0083

0.10

0.5979

0.0650

0.0499

0.0166

0.299

0.0332

0.0249

0.0083

0.25

0.5869

0.0640

0.0489

0.0163

0.292

0.0326

0.0243

0.0081

0.50

0.5513

0.0607

0.0456

0.0155

0.2700

0. '0297

0.75

0.5014
1

0.0561

0.0410

0.0144

0.2396

0.0268

0.0196

0.0069

1.0

0.4463

0.0510

0.0359

0.0132

10.2o63

0.0236

0.0166

0.0061

2.0

0.2642

0.0347

0.0190

0.0094

0.1006

0.0132

0.0072

0.0036

3.0

0.1642

0.0263

0.0

0.0074

0.0495

0.0079

0.0029

0.0022

4.0

0.1091

0.0218

0.0040

0.0063

0.0263

0.0053

0.0010

0.0015

5.0

0.0760

0.0192

0.0012

0.0055

0.0150

0.0038

0.0002

0.0011

6.0

0.0548

0.0173

0.0006

0.0050

0.0090

0.0029

0.0

0.0008

7.0

0.0406

0.0159

-0.0016

0.0045

0.0058

0.0023

8.0

0.0308

0.0147

-0.0022

0.0040

0.0038

0.0018

9.0

0.0238

0.0138

-0.0025

0.0036

0.0026

0.0015
1

10.0
L

0.0188

0.0130

-0.0026

0.0033

0.0019

0.0013

--

...
Table

A-5.2.

Bimoment

'0.0223

coefficients

0.0076

-0.0002

0.0006

-0.0002

0.0005

-0-0003

0.0004

-0-0003
I

0.0003

bl-

t4

II

160.
APPENDIX

A. 4.1

---------------MAIN
SUBROUTINE
--------------------

C
c

C
C

OF SPACE FRAMES
FIRST AND SECOND ORDER ANALYSIS
A, BM, CM, DM, EM, FM, WJ, CFM
DOUBLE PRECISION
JDF
INTEGER*4
RESFIL, DATFIL
CHARACTER*10
NM(1000),
YL(1000),
ZL(1000),
XL(1000),
DIMENSION
JDF(3000)
1 JS(1000),
JA(5000),
JB(5000),
JC(5000),
ITN(5000),
DIMENSION
I QC(5000),
RC(5000)
DM(8000),
EM(8000),
PCA(5000),
PCB(5000)
DIMENSION
NT(400),
MAP(1000,500),
WJ(6000),
A(8000),
DIMENSION
CM(80000)
1 BM(8000),
EY(50),
GR(50),
AR(50),
GQ(50),
DIMENSION
GQR(50),
GJ(50),
1 QS(50),
RS(50)
INTEGER ADDR(1000),
WADDR(1000),
ORD(1000),
NST(1000),
P, Q
GG(50,5),
DIMENSION
FFW(50,8),
PAY(50),
PAZ(50),
1 WLO(1000),
JLO(2000)
COMMON/FORCS/FM(35000),
CFM(35000)
COMMON/BARMS/GG, FFW, PAY, PAZ, JLO, WLO
COMMON/SPMTS/EY,
AR, GR, GQ, GQR, GJ, QS, RS
COMMON/JOINT/XL,
YL, ZL, WJ
COMMON/MISC/A,
BM, CM, DM, EM
COMMON/INTS/NM,
JS, JDF, NT, MAP, WADDR, ADDR, ORD, NST
COMMON/SPMEM/ITN,
JA, JB, JC, QC, RC, PCA, PCB
COMMON/NITS/NJS,
NPMS, NPMTS, NILS, NJOLD, NPREV, NMS,
1
NEISTOR,
NSTOR, ISTOW, NSTOW, IM, JOB, NPRMS, DET
2
NBMS, NBMTS, MODE, JCN, JCDF, CRD, FLAM, NLJS, TOL

3
c
C
C
C
C
C
C
C

MODE 0=
MODE 1=
MODE 2=

SINGLE ANALYSIS - SEVERAL LOAD VECTORS

STABILITY
ANALYSIS AT SINGLE LOAD LEVEL
REPEATED CYCLES TO CRITICAL LOAD

ADD 10 TO MODE FOR INCREMENTAL LOADING USING INPUT

LOAD AS FIRST STEP

9123

.C
C
777

NMTS, NINTS

IOBUFFER=16384
DO 9123 IJK=1,23
PRINT*
CONTINUE
WRITE
WRITE (*. *'("PROGRAMME
PRINT*
PRINT*
PRINT*

$p)
SPACE")')

PRINT*
PRINT*
OF RESULTS FILE
WRITE (*.. '("NAME
READ (*, '(W)
RESFIL
OPEN (UNIT=8pFILE=RESFILpSTATUS='UNKNOWN'pERR=777)

161.
888

PRINT*

PRINT*
WRITE (*,
OF DATA FILE
#'("NAME
READ
(*p'(A)')
DATFIL
(UNIT=9pFILE=DATFIL,
STATUS='OLD'pERR=888)
OPEN
OEN
(UNIT=lOpSTATUS=ISCRATCHI,
FORM=IUNFORMATTEDI)
FORM=IUNFORMATTEDI)
STATUS=ISCRATCHI,
OoPEN (UNIT=11,
NPRINT=O
100 FORMAT(lHl//47X,
34HSTABILITY
ANALYSIS
OF
+ SPACE FRAMES/47Xp34(lH*)
)/43XplH*p43XplH*/
+//////43Xp23(2H*
PROGRAM TO ANALYSE THE TORSIONAL/FLEXURAL
+43Xp45H*
45H* BEHAVIOUR
OF SPACE FRAMES CONTAINING:
+/43X,
lH*/
lH*p43X,
+/43X,
Ell
BARSOUM TYPE MEMBERS WITH SEVEN
+43XP45H*
DEGREES OF FREEDOM.
+/43X, 45H*
+/43X.. l H*., 43X,, l H*/
[21 MEMBERS WITH OFFSET SHEAR CENTRES
+43X, 45H*
AND NEUTRAL AXES.
+/43X, 45H*
lH*.. 43X, lH*/43Xp23(2H*
+/43X,
WRITE(8plOO)
101 FORMAT(I5,18A4)
(A(I),
900 READ(9pl0l)JOB,,
I=1,18)
IF(JOB.
LE. -l)GO
TO 999
102 FORMAT(l3HlJOB
10(lH*))
NUMBER
16/lX,
,
103 FORMAT(lX.. 12HJOB NUMBER
p16/11(lH*))
IF(NPRINT)3,3.4
3 WRITE(8,103)JOB
GOTO 9
4 WRITE(8,102)
JOB
9 CONTINUE
112
770

C
C
C
C
c
C
C

IIPRINT=NPRINT+l
FORMAT(//lH
WRITE(8,112)(A(I),
FORMATU/13H
WRITE(8,770)

18A4/1)
I=1,18)
SPACE FRAMES/lX,

12(lH*))

NMS=O
NPRMS=O
NPREV=O
NJOLD=O
NSTOR=O
NSTOW=O
FLAM=1.0
ICOUNT COUNTS NUMBER OF DIFFERENT LOAD LEVELS
NI COUNTS CYCLES AT A GIVEN LOAD
NIT COUNTS TOTAL NUMBER OF CYCLES
NEG COUNTS NUMBER OF NEGATIVE DETERMINANT CONDITIONS
ENCOUNTERED
ICREV COUNTS NUMBER OF REVERSALS OF SIGN OF CRITICAL
DEFLECTION
ICOUNT=O
NI=Q
NIT=O
NEG=O
ICREV=O
DET1=1.0
CRDO=0.0
WNEG=1000000000.0
NCY=20

162.

301

REWIND 10
REWIND 11
IM=50
CALL MAPPUSTEP)

IF(NBMS. EQ. O)GO TO 7

J=14*NBMS
DO 2 I=1, J
CFM(I)=O. O
2 FM(I)=O. O
7 I=O
IF(NMS. GT. D)CALL SPACEM
IF(NBMS. GT. O)CALL BARS(I,

NIT)

B 210.0
CALL SOLVE(B)
EQ. O. OR. DET1. EQ. 1.0)DET1=DET
IF(NlT.
LT. O. O)GO TO 991
IF(DET.
GT. O)GO TO 301
IF(NINTS.

C
I=10
IF(NMS. GT. O)CALL SPACE(I)
IF(NBMS. GT. O)CALL BARS(I,

NIT)

RECOVER BLOCK DATA FROM TAPE

IF(NPREV. LT. 1)GO TO 200
DO 250 I=1,7
250 BACKSPACE 10
READ(10)NMS, NJS, NMTS, NJOLD, NPREV, NINTS, NE, ISTOR, ISTOW,
1 NSTOR, NSTOW
CALL STORE(7,1, ISTOR-NSTOR, A, l)
JA(l),
JB(I),
JC(I),
QC(I), RC(I), PCA(I),
READ(10)(ITN(I),
1 PCB(I), I=1, NMS)
ORD(I), ADDR(I), WADDR(I), NST(I),
JDF(I),
READ(10)(JS(I),
1 XL(I),
YL(I),
ZL(I),
I=1, NJS)
QS(I),
AR(I), GR(I), GQ(I), GQR(I), GJ(I),
iEAD(10)(EY(I),
1 RS(I), I=1, NMTS)
J=NJS-NINTS
K), K=I, IM), I=1, J)
READ(10)((MAP(I,
J=ISTOW-NSTOW
Do 260 I=1, NSTOW
260 CM(I)=WJ(I)
Do 255 I=1, NSTOW
255 WJ(WADDR(1+NJS-NINTS)+I-1)=CM(I)
I=1, J)
READ(10)(WJ(I),
Do 253 1=1,7
253 BACKSPACE 10
tF(JOB. LT. 30000)GO To 257
CALL STORE(9,1, ISTOR, A, l)
I=1, ISTOW)
IF(ISTOW. LE. 100)WRITE(8,256)(WJ(I),
lOE12.4/))
256 FORMATU/100H
257 B=0.0
CALL SOLVE(B)
GO TO 5
C
C
C
C
C

SEQUENCE FOR SECOND ORDER ITERATIONS

FM(I)
HOLDS MEMBER FORCES AT LAST CONVERGENCE.
LOAD FACTOR = WAL
FM(I)
ARE FACTORED TO PREDICTION SO DIRECTLY COMPARABLE

163.
C
C

WITH CFM(D
CURRENT VALUES OF FORCES ARE IN CFM(D
200 IF(MODE. EQ. O)GO TO 900
NI=NI+l
NIT=NIT+l
Z=DET/DET1
WRITE(8,107)NI,
CRD, DET, Z
IF(CRD.
EQ. O. O. AND. CRDO. EQ. O. O. AND. NI. GE. 2)GO
I3, El5.6,15H
107 FORMAT(/20H
DEFLECTION,
CRITICAL
1 ERMINANT, E15.6,
FlO. 5)
IF(CRD. EQ. O. O)GO TO 315
LT. TOL)GO TO 313
IF(ABS((CRD-CRDO)/CRD).
LT. O. O)ICREV=ICREV+l
315 IF(CRD*CRDO.
GE. 3)GO TO 973
IF(ICREV.
GE. NCY. AND. MODE. EQ. 2)GO TO 973
IF(NI.
GE. NCY)GO TO 900
IF(NI.
CRDO=CRD
IF(MODE. NE. 3)GO TO 15
J=14*NBMS
Do 383 I=1, J
383 CFM(I)=FM(I)
GO TO 15

TO 902
DET

ARRIVES AT LABEL 313 IF CONVERGED AT GIVEN LOAD FACTOR

313 ICOUNT=ICOUNT+l
IF(MODE. EQ. 1)GO TO 900
WRITE(8,108)
108 FORMATU/26H
CALCULATION HAS CONVERGED)
Z=FLAM/CRD
WNEW=5.0*FLAM/3.0
IF(ICOUNT. EQ. 1)GO TO 319
EQ. O)GO TO 319
IF(NIT.
ZO=WWO/CO
Y=ZO-Z
IF(Y. EQ. O. O)GO TO 993
WNEW=FLAM+(FLAM-WWO)*Z/Y
WRITE(8,109)WNEW
109 FORMATUMH
PREDICTION, E15.6)
LT. TOL. AND. ISTEP. EQ. 0)
IF(ABS((WNEW-PRED)/PRED).
1 GO TO 900
IF(WNEW. LT. WNEG)GO TO 319
WRITE(8,147)WNEG
147 FORMAT(/32H PREDICTION TOO HIGH - REDUCE TO, El5.6)
WNEW=WNEG
319 IF(ICOUNT. GT. NCY)GO TO 900
WWO=FLAM FLAM=0.75*WNEW+0.25*Wwo

411
412

IF(ISTEP.
EQ. O)GO TO 411
X=WWO+1.0
IF(X. GT. FLAM. AND. ICOUNT. GE. 2)GO TO 411
IF(ISTEP.
GT. 20)GO TO 411
ISTEP=ISTEP+l
FLAM=X
GO TO 412
ISTEP=O
CONTINUE

C
FACTOR=FLAM/WWO

164.

C
C
C
C

c
C

PRED=WNEW,
C0-2CRD
CRDO=CRD*FACTOR
NI=O
ICREV=O
WAL=FLAM
J=14*NBMS
DO 371 I=1, J
CFM(I)=CFM(I)*FACTOR
371 FM(I)=CFM(I)
FACTOR, FM(l), FM(2),
WRITE(8,110)FLAM,
110 FORMAT(/16H NEW LOAD FACTOR, E12.5,9H
4El3.5)
1 E12.5,10X,
***
TAPE RECOVERY REQUIRED (REWIND)
***
PREPARE FOR RECYCLING
15 DO 10 I=1, ISTOW
10 WJ(I)=O. O
CALL STORE(4,1, ISTOR, A, l)
IF(NLJS. EQ. O)GO TO 7
DO 11 I=1, NLJS
KA=JLO(2*I-11)
K=WADDR(KA)+JLO(2*I)-l
DO 11 J=1, NILS
IW=K+(J-1)*JS(KA)
11 WJ(IW)=WJ(IW)+WLO(I)*FLAM
IF(JOB. LT. 20000)GO TO 7
WRITE(8,385)
J=14*NBMS
I=1, J)
WRITE(8,386)(CFM(I),
FORCE MATRIX/lH
385 FORMAV/13H
386 FORMATUE13.5)
)
390 FORMATM
WRITE(8,390)
GO TO 7

FM(3), FM(4)
FACTOR,

ARRIVES IF NEGATIVE DETERMINANT HAS BEEN FOUND

991 WRITE(8,992)
992 FORMATU/30H
HALTED - NEGATIVE DETERMINANT)
IF(MODE. EQ. O)GO TO 900
C ***
C TAPE RECOVERY REQUIRED (REWIND)
C ***
973 ICREV=O
IF(NIT.
GT. 15U)WRITE(8,974)
974 FORMATU/16H
TOO MANY CYCLES)
GT. 150)GO TO 900
IF(NIT.
NEG=NEG+l
IF(NEG. GT. NCY)GO TO 900
IF(FLAM. LT. WNEG)WNEG=FLAM
IF(ICOUNT. EQ. O)GO TO 8
FLAM=FLAM-0.24*(WNEW-WWO)
FACTOR=FLAM/WAL
GO TO 387
8 FLAM=0.5*FLAM
FACTOR=0.5
387 NI=O
J=14*NBMS

165.
383

DO 388 I=l,
j
CFM(I)=FM(I)*FACTOR
WRITE(8,110)FLAM,
GO TO 15

FACTOR, CFM(l),

CFM(2),

CFM(3)

CFM(4)

C
993 WRITE(8,111)
111 FORMATM/14H
GO TO 900
902 WRITE(8,903)
903 FORMATU/34H
GO TO 900
999
CLOSE (8)
CLOSE (9)
CLOSE (10)
CLOSE 0 1)
STOP
END

NO PREDICTION)

REPEATED ZERO CRITICAL

DEFLECTION)

166.

C
C
C
c
C

APPENDIX A. 4.2
---------------SUBROUTINE
MAPP
--------------------

SUBROUTINE MAPP(ISTEP)
DOUBLE PRECISION A, BM, CM, DM, EM, WJ
INTEGER*4 JDF
YL(1000),
ZL(1000),
NM(1000),
DIMENSION XL(1000),
JDF(3000)
1 JS(1000),
JA(5000),
JB(5000),
JC(5000),
DIMENSION ITN(5000),
1 QC(5000), RC(5000)
DIMENSION EY(50), GR(50), AR(50), GQ(50), GQR(50), GJ(50),
1 QS(50), RS(50), BW(50)
,
WJ(6000),
A(8000),
DIMENSION NT(400), MAP(1000,500),
1 BM(8000), CM(80000)
DIMENSION DM(8000), EM(8000), PCA(5000),
PCB(5000)
INTEGER ADDR(1000), WADDR(1000), ORD(1000), NST(1000),
1 P, Q, S, T
DIMENSION GG(50,5),
FFW(50,8),
PAY(50), PAZ(50),
l WLO(1000), JLO(2000)
,,
COMMON/BARMS/GG, QZZ, PAY, PAZ, JLO, WLO
COMMON/JOINT/XL, YL, ZL, WJ
COMMON/NITS/NJS, NPMS, NPMTS, NILS, NJOLD, NPREV, NMSj
1 NE, ISTOR, NSTOR, ISTOW, NSTOW, IM, JOB, NPRMS, DET
2 NBMS, NBMTS, MODE, JCN, JCDF, CRD, FLAM, NLJS, TOL
3 NMTS, NINTS
COMMON/INTS/NM, JS, JDF, NT, MAP, WADDR,ADDR, ORD, NST
COMMON/MISC/A, BM, CM, DM, EM
COMMON/SPMEM/ITN, JA, JB, JC, QC, RC, PCA, PCB
COMMON/SPMTS/EY, AR, GROGQ,GQR, GJ, QS, RS, BW
IW=NSTOW+l
DO 7 I=IW, 600
7 WJ(I)=O. O
NE=O
C READ ROUTINE FOR SPACE FRAMES
NPRMS=NPRMS+NMS
IF(NPREV. GT. O)READ(9,160)NJS,
NMS, NBMS
IF(NPREV. EQ. O)READ(9,160)NJS,
NMS, NMTS, NBMS, NBMTS,
1 JCN, JCDF, TOL, MODE
160 FORMAT(8I5, FlO. O)
WRITE(8,193)NJS,
NMS..NMTS, NBMS, NBMTS
193 FORMAT(/I5,, 7H JOINTS, I12,14H
SPACi MEMBERS, 14,6H'TY
1 PES, I2,16H
BARSOUM MEMBERS, 14,6H TYPES/)
IF(NPREV. GT. O)GO-TO-3
IF(JCN. EQ. O)JCDF=l
IF(JCN. EQ. O)JCN=l
IF(TOL. EQ. O. O)TOL=0.005
172 FORMATUH MODEI2/)
WRITE(8,172)MODE
DEGREE 0
173 FORMAT(15H CRITICAL JOINT, I5,5X, 26HCRITICAL
1F
FREEDOM, I5/)
JCDF
IF(MODE. NE. O)WRITE(8,173)JCN,
174 FORMATOOH TOLERANCE, F12.6//)

167.

270

ISTEP=O
IF(MODE. LE. 9)GO TO 270
MODE=MODE-10
ISTEP=l
CONTINUE

WRITE(8,174)TOL
18(lH*)//)
175 FORMAT(19H JOINT CO-ORDINATES/lX,
WRITE(8,175)
3 NJS=NJS+NPREV
19,3Fl 0.0)
161 FORMAT(Il
162 FORMATOF10.0)
IW=NPREV+l
Do 401 I=IW, NJS
NST(I)=O
Nm(I)=0
JDF(I),
XL(I),
YL(I),, ZL(I)
READ(9,161)JS(I),
195 FORMAT(6H JOINT, 214,3X, I8,3El8.6)
JS(I),
JDF(I),
XL(I),
YL(I),
ZL(I)
WRITE(8,195)I,
DO 401 J=1, IM
401 MAP(I, J)=O
163 FORMAT(4I5,4F10.0)
IF(NPREV. GT. O)GO TO 304
IW=NMTS+NBMTS
WRITE(8,500)
18(lH*)/)
MEMBER INFORMATION/lX,
500 FORMATU/19H
DO 402 I=1, IW
AR(I), GR(I), GQ(I), GQR(I), GJ(I),
READ(9,162)EY(I),
1 QS(I), RS(I)
196 FORMAT(/5H TYPE, 12,8E14.6)
EY(I), AR(I), GR(I), GQ(I), GQR(I), GJ(I),
WRITE(8,196)I,
1 QS(I), RS(I)
IF(I. GT. NBMTS)GO To 402
PAY(I),
PAZ(I),
BW(I)
J), J=1,5),
READ(9,662)(GG(I,
PAY(I), PAZ(I),
BW(I)
J), J=1,5),
WRITE(8,665)(GG(I,
J), J=1,8)
READ(9,662)(FFW(I,
J), I=1,8)
WRITE(8,661)(FFW(I,
NOTE THAT GG(I, l)
IS POLAR M OF I ABOUT SHEAR CENTRE
665 FORMAT(13H BARSOUM TYPE, 4x, 8El2.5)
661 FORMAT(17H BIMOMENT FACTORS, 4X, 8El2.5)
662 FORMAT(8FlO. O)
402 CONTINUE
304 T=NMS+NBMS
WRITE(8,176)
176 FORMAT(//lH
DO 404 I=1, T
JA(I),
QC(I), RC(I),
JB(I),
JC(I),
READ(9,163)ITN(I),
1 PCA(I), PCB(I)
4El5.6)
197 FORMATUH MEMBER, 14, Il2,3I5,12X,
QC(I), RC(I)..
JC(I),
JA(I),
JB(I),
ITN(I),
WRITE(8,197)I,
1 PCA(I), PCB(I)
J=ABS(JA(I))-NJOLD+NPREV
K=ABS(JB(I))-NJOLD+NPREV
EQ. O)GO TO 403
IF(JS(K).
IW=NST(J)
DO 43 L=1, IW
IF(MAP(J,
L). EQ. K)GO To 403
43 CONTINUE

168.
NST(J)=NST(J)+l'
NMW ) =NM(J ) +J S(K)
MAP(J, NST(J))=K

403

IF(JS(J).
EQ. O)GO To 404
IW=NST(K)
Do 44 L=1, IW
IF(MAP(K, L). EQ. J)GO To 404
44 CONTINUE
NST(K) =NST(K) +1
NM(K)=NM(K)+JS(J)
MAP(K, NST(K))=J
404 CONTINUE
NILS, NINTS
READ(9,160)NLJS,
END OF SPACE FRAME DATA
C PRELIMINARY STORE MAP COMPLETE
C START FINAL STORE MAP OR CONSIDER NEXT REDUCTION
WRITE(8,129)
15HSTORAGE DETAILS/lX,
15(lH*)//)
129 FORMAT(///lX,
200 J=1000
IW=NJS-NINTS
DO 13 I=1, IW
EQ. 0)GOTO13
IF(JS(D.
GE. J)GOT013
IF(NM(I).
LT. 0)GOTO13
IF(NM(I).
J=NM(D
K=I
13 CONTINUE
IF(J. EQ. O)GO TO 57
IF(J. LT. 999)GO TO 19
WRITE(8,113)
113 FORMAT(/26H ELIMINATION
PLAN COMPLEM)
GO TO 21
57 WRITE(8,123)K,
J, S
123 FORMAT(11H LAST JOINT, 3I6)
NE=NE+l
ORD(NE)=K
GO TO 21
19 S=NST(K)+l
111 FORMAT(16H ELIMINATE JOINT, I4,14H
14
MEMBERS
1 9H STORING
14)
WRITE(8,111)K,
J, S
NE=NE+l
ORD(NE)=K
NM(K)=-NM(K)
J=NST(K)
DO 14 I=1, J
N=MAP(K, I)
IF(N. LT. O)GOT014,
IW=NST(N)
DO 18 Q=1, IW
IF(MAP(N, Q). NE. K)GOT018
MAP(N, Q)=-MAP(N, Q)
NM(N)=NM(N)-JS(K)
18 CONTINUE
DO 85 L=1, J
IF(MAP(K, L). LT. 0)GOTO85
IF(I.
EQ. L)GOTO85
S=O
IW=NST(N)
DO 16 P=1, IW

169.
IF(MAP(N, P). NE. MAP(K, L))GOT016
S=2
16 CONTINUE
IF(S. GT. 1)GOT085
L))
NM(N)=NM(N)+JS(MAP(K,
NST(N)=NST(N)+l
L)
MAP(N, NST(N))=MAP(K,
LT. IM-3)GO TO 85
IF(NST(N).
NM(N)=l
2I6)
REQUIRED,
ELIMINATION
FORCED
120 FORMAT(28H
NST(N)
WRITE(8,120)N,
85 CONTINUE
14 CONTINUE
CONSIDER NEXT REDUCTION
GO To 200
STORE MAP LAID OUT COMPILE ADDRESSES
21 S=l

114

22

P=j
Do 20 I=1, NJS
WADDR(I)=P
ADDR(I)=S
P=P+JS(I)*NILS
16,2110)
FORMAT06,9HADDRESS
NST(I),
ADDR(I),
WADDR(I)
WRITE(8,114)I,
LT. 1)GO TO 20
IF(NST(I).
IW=NST(I)
DO 22 J=1, IW
TO 22
J)).
LT. I)GO
IF(IABS(MAP(I,
J)))
S=S+JS(I)*JS(IABS(MAP(I,
CONTINUE

20 S=S+JS(I)*JS(I)
S=S-l
P=P-l
2110)
STORAGE
115 FORMATOOH
P
WRITE(8,115)S,
STORE MAP COMPLETE
FINAL
ISTOW=P
ISTOR=S
DO 50 I=1, NJS
2014)
116 FORMATM
MAP
JS, 15,6H
(MAP(I,
J), J=1,20)
50 WRITE(8,116)NM(I),
IF(NLJS.
EQ. O)GO TO 53
READ IN LOAD VECTORS
916 FORMAT(///16H
15(IH*))
LOADING
DETAILS/lX,
WRITE(8,916)
NINTS,
127 FORMAT(//5H
NLJS, 15,7H
NILS, 15,8H
NILS, NINTS
WRITE(8,127)NLJS,
DO 24 I=1, NLJS
5)
105 FORMAT(2I5,7FlO.
J=1, NILS)
L, (BM(J),
READ(9,105)K,
J=1, NILS)
L, (BM(J),
WRITE(8,194)K,
JOINT, I4, I2,7E15.6)
194 FORMAT(6H
K=K-NJOLD+NPREV
WLO( I) =BM(1 )

JLO(2*I-1)=K
JLO(2*I)=L
DO 24 J=1, NILS
M=WADDR(K)+L-1+(J-1)*JS(K)
WJ(M)=BM(J)+WJ(M)

I5//)

170.
24 CONTINUE
WRITE(8,128)
128 FORMAT(lH ///lH
53 CALL STORE(4, NSTOR+1, ISTOR-NSTOR, A, l)
C ADJUST INTERFACE TERMS TO CONFORM TO NEW ADDRESSES
IF(NPREV. LT. 1)GO TO 256
DO 255 I=1, NPREV
J=NPREV+1-I
K=NT(J+1)-NT(J)
CALL STORE(3, NT(J), K, A, 1).
CALL STORE(4, NT(J), K, A, l)
255 CALL STORE(2, ADDR(J), K, A, l)
256 RETURN
END

171.
A. 4.3
l-'APPENDIX
----------------SUBROUTINE
BARS
------------------

c
c
c
SUBROUTINE BARS(IC, NI)
DOUBLE PRECISION A, BM, CM, DM, EM, FM, WJ, CFM
DOUBLE PRECISION XLG, X, Y, Z, AA, AB, B1, B2, SO, FSS, XX, XXX,
1 BB, CC, DD, CL1, CM1, CN1, CL2, CM2, CN2, CL3, CM3, CN3
INTEGER*4 JDF
ZL(1000),
YL(1000),
NM(1000), JS(1000)
DIMENSION XL(1000),
JC(5000),
JA(5000),
JB(5000),
QC(5000)
1
ITN(5000),
2
EY(50), GR(50), AR(50), GQ(50), GQR(50), GJ(50), QS(50)
WJ(6000), A(8000),
BM(8000)
3
NT(400), MAP(1000,500),
JDF(3000),
4
DM(8000), EM(8000), PCA(5000), PCB(5000),
BW(50), PAY(50), PAZ(50), JLO(2000),
WLO(1000)
5 -GG(50,5),
6 RC(5000), RS(50), CM(80000), FFW((50,8)
INTEGER ADDR(1000), WADDR(1000), ORD(1000), NST(1000)p
1 P, Q, R, T
REAL MX, MY, MZ, MY1, MY2, MZ1, MZ2, MW,MW1, MW2
CFM(35000)
COMMON/FORCS/FM(35000),
COMMON/JOINT/XL, YL, ZL, WJ
COMMON/NITS/NJS, NPMS, NPMTS, NILS, NJOLD, NPREV, NMS, NMTS,
ISTOR, NSTOR, ISTOW, NSTOW, IM, JOB, NPRMS, DET, NINTS
1
-NE,
2 NBMS, NBMTS, MODE, JCN, JCDF, CRD, FLAM, NLJS, TOL
COMMON/INTS/NM, JS, JDF, NT, MAP, WADDR,ADDR, ORD, NST
COMMON/BARMS/GG, FFW, PAY, PAZ, JLO, WLO
COMMON/MISC/A, BM, CM, DM, EM
COMMON/SPMEM/ITN, JA, JB, JC, QC, RC, PCA, PCB
COMMON/SPMTS/EY, AR, GR, GQ, GQR, GJ, QS, RS, BW
IF(IC.
GT. 5. AND. JOB. GT. 999)WRITE(8,150)
150 FORMAT(//14H
MEMBER FORCES/lX, 13(lH*))
DO 459 II=1, NBMS
I=II
EVALUATE LENGTH AND DIRECTION COSINES - MEMBER I
P=ABS(JA(I))-NJOLD+NPREV
Q'4ABS(JB(I))-NJOLD+NPREV
R=ABS(JC(I))-NJOLD+NPREV
XLG=SQRT((XL(Q)-XL(P))**2+(YL(Q)-YL(P))**2+(ZL(Q)1 ZL(P))**2)
CL1=(XL(Q)-XL(P))/XLG
CM1=(YL(Q)-YL(P))/XLG
CN1=(ZL(Q)-ZL(P))/XLG
AA=(CL1*(XL(R)-XL(P))+CM1*(YL(R)-YL(P))+CN1*(ZL(R)1 ZL(P)))/XLG
X=XL(P)+AA*(XL(Q)-XL(P))
Y=YL(P)+AA*(YL(Q)-YL(P))
Z=ZL(P)+AA*(ZL(Q)-ZL(P))
AB=SQRT((XL(R)-X)**2+(YL(R)-Y)**2+(ZL(R)-Z)**2)
/AB
CL2= (XL(R)-X)
CM2=(YL(R)-Y)/AB
CN2=(ZL(R)-Z)/AB
CL3=CM1*CN2-CM2*CN1
CM3=CN1*CL2-CN2*CL1
CN3=CL1*CM2-CL2*CM1

172.
164

FORMAT(/3E15.6,26H
DIRECTION COSINES MEMBER, 15,
1 9H
LENGTH, E15.6)
165 FORMAT(3E15.6)
IF(IC. GT. 5. OR. NI. GT. O)GO TO 449
WRITE(8,164)CL1,
CM1, CN1, I, XLG
WRITE(8,165)CL2,
CM2, CN2
WRITE(8,165)CL3,
CM3, CN3
IF(I.
EQ. NBMS)WRITE(8,166)
166 FORMATOH /1H )
C
c

BUILD

449
450

TRANSFORMATION

MATRIX

DO 450 J=1,196
DM(J)=O. O
CR=RC(I)
CQ=QC(I)
PA=PCA(I)
PB=PCB(I)
DM(1)=CL1
-DM(2)=CM1
DM(3)=CN1
DM(15)=CL2
DM(16)=CM2
DM(17)=CN2
DM(29)=CL3
DM(30)=CM3
DM(31)=CN3
DO 447 J=1,3
Do 447 K=1,3
L=J+3+14*(K+2)
M=J+14* (K-1
DM(L)=DM(M)
447 CONTINUE
DM(74)=-CL3
DM(75)=-CM3
DM(76)=-CN3
DM(4)=-CR*CL2-CQ*CL3
DM(5)=-CR*CM2-CQ*CM3
DM(6)=-CR*CN2-CQ*CN3
DM(18)=-CR*CL1+PA*CL3
DM(19)=-CR*CM1+PA*CM3
DM(20)=-CR*CN1+PA*CN3
DM(32)=CQ*CL1+PA*CL2
DM(33)=CQ*CM1+PA*CM2
DM(34)=CQ*CN1+PA*CN2
DM(106)=CL1
DM(107)=cmi
DM(108)=CN1
DM(120)=CL2
DM(121)=CM2
DM(122)=CN2
DM(134)=CL3
DM(135)=CM3
DM(136)=CN3
DM(151)=CL1
DM(152)=CM1
DM(153)=CN1
DM(165)=CL2
DM(166)=CM2

IN

DM(14*14)

173.
DM(167)=CN2
DM(179)=-CL3
DM(180)=-CM3
DM(181)=-CN3
DM(109)=-CR*CL2-CQ*CL3
DM(110)=-CR*CM2-CQ*CM3
DM(111)=-CR*CN2-CQ*CN3
DM(123)=-CR*CL1+PB*CL3
DM(124)=-CR*CM1+PB*CM3
DM(125)=-CR*CN1+PB*CN3
DM(137)=CQ*CL1+PB*CL2
DM(138)=CQ*CM1+PB*CM2
DM(139)=CQ*CN1+PB*CN2
DM(91)=1.0
DM(196)=1.0
LE. 5. AND. NI. LE. O. AND. II. LE. 1)WRITE(8,399)(DM(J)
IF(IC.
1
J=1,196)
399 FORMAT(/22H
TRANSFORMATION
MATRIX/14(/lH
14F8.4))

BUILD

STIFFNESS

MATRIX

I=ITN(II)
DO 451 J=1,196
451 A(J)=O. O
E=EY(I)
AA=AR(I)
GZ=GR(I)
GY=GQ(I)
GK=GJ(I)
ECW=E*GQR(I)
GO=GG(I, l)
Gl=GG(I, 2)
G2=GG(I, 3)
G3=GG(I, 4)
G4=GG(I, 5)
J=14*(II-1)
FX=-CFM(J+l)
QY1=-CFM(J+2)
QZ1 =-CFM(J+3)
MX=-CFM(J+4)
Myl=-CFM(J+5)
MZ1=-CFM(J+6)
MW1=-CFM(J+7)
QY2=-CFM(J+9)
QZ2=-CFM(J+10)
MY2=-CFM(J+12)
MZ2=-CFM(J+13)
MW2=-CFM(J+14)
PY=-PAY(I)*FLAM
PZ=-PAZ(I)*FLAM
ZO=RS(I)
Yo=QS(I)
Bl=(Gl+G2)/GY-2.0*ZO
B2=(G3+G4)/GZ-2.0*Yo
SO=GO/AA+YO*Bl+ZO*B2
FSS=FX*SO
XX=XLG*XLG
XXX=XX*XLG
A(1)=E*AA/XLG

IN MEMBER COORDINATES IN A(14X14)

174.
AA=1.2*FX/XLG
BB=-FX/10.0
CC=FX*XLG/7.5
DD=-FX*XLG/30.0
MY=MYl-MY2
MZ=MZ1-MZ2
MW=MW1-MW2
A(8)=-A(l)
A(16)=12.0*E*GZ/XXX+AA
A(18)=0.6*MY/XLG+0.05*QZ1+0.55*QZ2
A(20)=-6.0*E*GZ/XX+BB
A(21)=-0.05*MY-0.05*QZ2*XLG
A(23)=-A(16)
A(25)=-0.6*MY/XLG-0.55*Qzl-0.05*QZ2
A(27)=A(20)
A(28)=-0.05*MY-0.05*QZ1*XLG
A(31)=12.0*E*GY/XXX+AA
A(32)=-0.6*MZ/XLG-0.05*QY1-0.55*QY2
A(33)=-6.0*E*GY/XX+BB
A(35)=0.05*MZ+0.05*QY2*XLG
A(38)=-A(31)
A(39)=0.6*mz/XLG+0.55*QY1+0.05*QY2
A(40)=A(33)
A(42)=0.05*mz+0.05*QY1*XLG
A(46)=1.2*GK/XLG+12.0*ECW/XXX+1.2*FSS/XLG+0.3*(2.0*MY/
1 XLG+QZ1+QZ2)*Bl+0.3*(2.0*MZ/XLG+QY1+QY2)*B2+(FFW1*MW
2 /XLG+FFW5*MX)*BW
A(47)=0.55*MZ+0.1*QY1*XLG+0.45*QY2*XLG
A(48)=-0.55*MY-0.1*QZ1*XLG-0.45*QZ2*XLG
A(49)=-GK/10.0-6.0*ECW/XX-FSS/10.0-0.05*(MY+QZ1*XLG)
1 *Bl-0.05*(MZ+QY1*XLG)*B2-(FFW3*MW+FFW7*MX*XLG)*BW
A(51)=-0.6*MY/XLG-0.05*QZ1-0.55*QZ2
A(52)=0.6*MZ/XLG+0.05*QY1+0.55*QY2
A(53)=-A(46)
A(54)=0.05*MZ-0.05*QY1*XLG+0.1*QY2*XLG
A(55)=-0.05*MY+0.05*QZ1*XLG-0.1*QZ2*XLG
A(56)=-GK/10.0-6.0*ECW/XX-FSS/10.0-0.05*(MY+QZ2*XLG)*
1 Bl-0.05*(MZ+QY1*XLG)*B2-(FFW3*MW+FFW7*MX*XLG)*BW
A(61)=4.0*E*GY/XLG+CC
A(63)=-MZ*XLG/15.0-QY1*XX/60.0-0.05*QY2*XX
A(66)=-A(33)
A(67)=-0.05*MZ-0.1*QY1*XLG+0.05*QY2*XLG
A(68)=2.0*E*GY/XLG+DD
A(70)=MZ*XLG/60.0+QY2*XX/60.0
A(76)=4.0*E*GZ/XLG+CC
A(77)=MY*XLG/15.0+QZI*XX/60.0+0.05*QZ2*XX
A(79)=-A(20)
A(81)., --0.05*MY+0.1*QZ1*XLG-0.05*QZ2*XLG
A(83)=2.0*E*GZ/XLG+DD
A(84)=-MY*XLG/60.0-QZ2*XX/60.0
A(91)=GK*XLG/7.5+4.0*ECW/XLG+FSS*XLG/7.5+(MY*XLG/15.0
1 +QZ1*XX/60.0+QZ2*XX/20.0)*Bl+(MZ*XLG/15.0+QY1*XX/60.0
2 +QY2*XX/20.0)*B2+(FFW2*MW*XLG+FFW6*MX*XX)*BW
A(93)=0.05*MY+0.05*QZ2*XLG
A(94)=-0.05*MZ-0.05*QY2*XLG
A(95)=-A(56)
A(96)=MZ*XLG/60.0+QY1*XX/60.0
A(97)=-MY*XLG/60.0-QZ1*XX/60.0
A(98)=-GK*XLG/30.0+2.0*ECW/XLG-FSS*XLG/30.0-(MY*XLG/60.0

175.
1 +(QZ1+QZ2)*XX/120.0)*Bl-(MZ*XLG/60.0+(QY1+QY2)*XX/120.0
2 )*B2-(FFW4*MW*XLG+FFW8*MX*XX)*BW
A(106)=A(l)
A(121)=A(16)
A(123)=0.6*MY/XLG+0.55*Qzl+0.05*QZ2
A (125) =-A (20)
A(126)=0.05*MY+0.05*QZ1*XLG
A (136) =A (31
A(137)=-0.6*MZ/XLG-0.55*QY1-0.05*QY2
A(138)=-A(33)
A(140)=-0.05*MZ-0.05*QY1*XLG
A(151)=A(46)+(PY+PZ)
A(152)=-0.55*MZ-0.45*QY1*XLG-0.1*QY2*XLG
A(153)=0.55*MY+0.45*QZ1*XLG+0.1*QZ2*XLG
A (15 4) =-A (49)
A(166)=A(61)
A(168)=-MZ*XLG/15.0-0.05*QY1*XX-QY2*XX/60.0
A(181)=A(76)
A(182)=MY*XLG/15.0+0.05*QZI*XX+QZ2*XX/60.0
A(196)=A(91)+(QZ1-QZ2)*XX*Bl/30.0+(QY1-QY2)*XX*B2/30.0
Do 452 J=1,14
Do 452 K=1,14IF(J. GE. K)GO TO 452
A(J+14*(K-1))=A(K+14*(J-1))
452 CONTINUE
IF(IC. GT. 5)GO TO 480
FORM T. K. T(TRANS)

C
C
C

DM, A, BM)
CALL XMULT(14,14,14,
DM, A)
CALL XTRAN(14,14,
BM, A, EM)
CALL XMULT(14,14,14,
IS NOW IN
MATRIX
TRANSFORMED STIFFNESS
INTO 7X7 BLOCKS AND ADD INTO STORE
IF(JS(P).
DO 456

EQ. O)GO
J=1,7

EM.

SPLIT

INTO

TO 454

DO 456 K=1,7
456
1101

460
1102

463
462

A(K+7*(J-1))=EM(K+14i(J-1))
SORTSQ, lOI8)
FORMAT(7H
JS(p),
GE. 20000)WRITE(8,1101)P,
ADDR(P)
JDF(P),
IF(JOB.
SORTSQ(7, JDF(P),
NE. 7)CALL
A)
IF(JS(P).
JS(P)**2,
A, l)
CALL STORE(1, ADDR(P),
EQ. O)GO TO 459
IF(JS(Q).
IF(Q. LT. P)GO TO 454'
DO 460 J=1,7
DO 460 K=1,7
A(K+7*(J-1))=EM(K+98+14*(J-J))
''
SORTG, lOI8)
FORMAT(7H
GE. 20000)WRITE(8,1102)P,
Q, JS(Q),
IF(JOB.
JS(P),
JDF(P),
JDF(Q)
NE. 49)CALL
7, JDF(Q),
IF(JS(P)*JS(Q).
SORTG(7, JDF(P),
A)
K=0
IW=NST(P)
DO 462 J=1, IW
J)).
IF(IABS(MAP(P,
LT. P)GO TO 462
J)).
IF(IABS(MAP(P,
NE. Q)GO TO 463
T=K
K=K+JS(P)*JS(IABS(MAP(P,
J)))
CONTINUE

176.
T=T+JS(P)*JS(P)+ADDR(P)
CALL STORE(1, T, JS(P)*JS(Q),
CONSIDER TERMS FOR END 2

A, l)

454

IF(JS(Q).
EQ. O)GO TO 459
DO 465 J=1,7
DO 465 K=1,7
465 A(K+7*(J-1))=EM(K+105+14*(J-1))
IF(JOB. GE. 20000)WRITE(8,1101)Q,
JS(Q), JDF(Q), ADDR(Q)
IF(JS(Q).
NE. 7)CALL SORTSQ(7, JDF(Q), A)
CALL STORE(1, ADDR(Q), JS(Q)**2,
A, l)
IF(JS(P).
EQ. O)GO TO 459
IF(P. LT. Q)GO TO 459
DO 467 J=1,7
DO 467 K=1,7
467 A(K+7*(J-1))=EM(K+7+14*(J-1))
IF(JOB. GE. 20000)WRITE(8,1102)Q,
JS(Q), JDF(Q), P, JS(P), JDF(P)
NE. 49)CALL
IF(JS(P)*JS(Q).
SORTG(7, JDF(Q), 7, JDF(P), A)
K=O
IW=NST(Q)
DO 469 J=1, IW
J)). LT. Q)GO TO 469
IF(IABS(MAP(Q,
J)). NE. P)GO TO 470
IF(IABS(MAP(Q,
T=K
J)))
470 K=K+JS(Q)*JS(IABS(MAP(Q,
469 CONTINUE
T=T+JS(Q)*JS(Q)+ADDR(Q)
A, l)
CALL STORE(I, T, JS(Q)*JS(P),
GO TO 459
C STIFFNESS MATRIX TERMS ENTERED - NEXT SEQUENCE EVALUATES
C
C

MEMBER FORCES
480

IW=NILS*JS(P)
DO 484 J=1, IW
484 EM(J)=WJ(WADDR(P)+J-1)
UNSORT(7, JDF(P),
IF(JS(P).
NE. 7)CALL
NILS, EM)
IW=NILS*JS(Q)
Do 486 J=1, IW
486 BM(J)=WJ(WADDR(Q)+J-1)
UNSORT(7, JDF(Q),
NE. 7)CALL
NILS, BM)
IF(JS(Q).
Do 488 K=1, NILS
Do 488 J=1,7
L=NILS-K
EM(J+14*L)=EM(J+7*L)
488 EM(J+7+14*L)=BM(J+7*L)
END 1 AND END 2 NOW CONSECUTIVE
C
MEMBER DISPLACEMENTS
IN
DM, BM)
CALL XTRAN(14,14,
A, BM, DM)
CALL XMULT(14,14,14,
NILS, DM, EM, A)
CALL XMULT(14,14,
100 L=II+NPRMS
Do 489 J=1, NILS
(A((J-1)*14+K),
489 WRITE(8,170)L,
K=1,14)
170 FORMAT(I4,7El5.6/4X,
7El5.6/IH
)
C EVALUATION
OF MEMBER FORCES COMPLETE - COLLECTED IN*CFM

490
459

DO 490 J=1,14
CFM(14*(II-1)+J)=A(J)
CONTINUE
WRITE(8,167)

EM

177.
167

FORMAT(lH )
IF(IC. LT. 5)GO TO 211
IW=NBMS*14
I-'21, IW)
WRITE(11)(CFM(I),
C STORE CURRENT MEMBER FORCES ON TAPE 1
211 IF(JOB. GT. 10000. AND. IC. LT. 5)CALL STORE(9,1,
RETURN
END

ISTOR, A, l)

178.
APPENDIX A. 4.4
----------------SUBROUTINE
SOLV(B)
-----------------------

SOLVE(B)
SUBROUTINE
A, BM, CM, DM, EM, WJ
DOUBLE PRECISION
JDF
INTEGER*4
ZL(1000),
NM(1000),
JS(1000)
YL(1000),
XL(1000),
DIMENSION
1
JDF(3000)
JB(5000),
JC(5000),
JA(5000),
QC(5000)
ITN(5000),
DIMENSION
1
RC(5000)
GQ(50),
AR(50),
GQR(50),
GJ(50),
GR(50),
EY(50),
DIMENSION
BW(50)
1 QS(50),
RS(50),
WJ(6000),
A(8000),
MAP(1000,500),
NT(400),
DIMENSION
1 BM(8000),
CM(80000)
EM(BOOO), PCA(5000),
PCB(5000)
DM(8000),
DIMENSION
ORD(1000),
NST(1000),
WADDR(1000),
INTEGER ADDR(1000),
1 P, Q, R, S, T
PAY(50),
PAZ(50),
FFW(50,8),
GG(50,5),
DIMENSION
JLO(2000)
1 WLO(1000),
COMMON/BARMS/GG, FFW, PAY, PAZ, JLO, WLO
YL, ZL, WJ
COMMON/JOINT/XL,
NPMS, NPMTS, NILS, NJOLD, NPREV, NMS, NMTS,
COMMON/NITS/NJS,
1
NE, ISTOR, NSTOR, ISTOW, NSTOW, IM, JOB, NPRMS, DET, NINTS
2
NBMS, NBMTS, MODE, JCN, JCDF, CRD, FLAM, NLJS, TOL
JS, JDF, NT, MAP, WADDR, ADDR, ORD, NST
COMMON/INTS/NM,
BM, CM, DM, EM
COMMON/MISC/A,
JA, JB, JC, QC, RC, PCA, PCB
COMMON/SPMEM/ITN,
AR, GR, GQ, GQR, GJ, QS, RS, BW
COMMON/SPMTS/EY,
BLOCK
INVERSION
BASIC
263 DO 60 I=1, NE
TO 240
LT. 5.0)GO
IF(B.
K=ORD(I)
STORE INVERT
KBB; WRITE BACK INVERT,
EXTRACT AND INVERT
IN BM
T=ADDR(K)
BM, l)
CALL STORE(3, T, JS(K)*JS(K),
DET)
BM, A
CALL XINVT(JS(K),
'
LT. O. O)RETURN
IF(DET.
A, l)
CALL STORE(2, T, JS(K)*JS(K),
AND WRITE BACK TO WB
KBB)*WB
(INVERT
EVALUATE
R=WADDR(K)
IW--;JS(K)*NILS
DO 63 J=1, IW
63 DM(J)=WJ(R+J-1)
NILS, A
JS(K),
DM, BM)
CALL XMULT(JS(K),
DO 64 J=1, IW
64 WJ(R+J-1)=BM(J)
EQ. NE. AND. NINTS. EQ. O)GO TO 60
IF(I.
STORE BLOCKS END TO END IN CM
EXTRACT KBA BLOCK BY BLOCK,
90 NS=O
NB=O
T=ADDR(K)+JS(K)*JS(K)
R=O
IW=NST(K)
DO 65 J=1, IW
J))
N=IABS(MAP(K,
IF(MAP(K,
J). LT. O)GO TO 66
NB=NB+l

179.
NS=NS+JS(N)
NT (NB) =N
IF(N. LT. K)GO

To

67

CALL STORE(3, T, JS(K)*JS(N),

CM, R+l)
R=R+JS(K)*JS(N)
66 IF(N. LT. K)GO TO 65
T=T+JS(K)*JS(N)
GO TO 65
67 IC=ADDR(N)+JS(N)*JS(N)
IV=NST(N)
DO 69 P=1, IV
P)). LT. N)GO To 69
IF(IABS(MAP(N,
P)). NE. K)GO TO 70,
IF(IABS(MAP(N,
Em, l)
CALL STORE(3, IC, JS(K)*JS(N),
IU=JS(K)
IY=JS(N)
DO 71 Q=1, IU
DO 71 L=1, IY
71 CM(R+Q+JS(K)*(L-1))=EM(L+JS(N)*(Q-1))
R-=R+JS(N)*JS(K)
P))),
70 IC=IC+JS(N)*JS(IABS(MAP(N,
69 CONTINUE
65 CONTINUE
IF(B. LT. 5.0)GO TO 91
IF(NS. EQ. O)GO TO 60
EVALUATE KAB(INVERT KBB)WB AND SUBTRACT FROM WA
NS, CM, DM)
CALL XTRAN(JS(K),
CALL XMULT(NS, JS(K), NILS, DM, BM, CM)
NR=O
DO 82 J=1, NB
DO 72 P=1, NILS
IW=JS(NT(J))
DO 72 Q=1, IW
AA=WJ(WADDR(NT(J))-l+Q+JS(NT(J))*(P-1))
WJ(WADDR(NT(J))-l+Q+JS(NT(J))=AA-CM(NR+Q+NS*(P-1))
1 *(P-1))
72 CONTINUE
82 NR=NR+JS(NT(J))
EVALUATE KAB(INVERT KBB)KBA AND SUBTRACT BLOCK BY BLOCK
FROM KAA
CALL XMULT(NS, JS(K), JS(K), DM, A, BM)
CALL XTRAN(NS, JS(K), DM, A)
CALL XMULT(NS, JS(K), NS, BM, A, CM)
tj C=0
DO 73'Q=1, NB
NR=O
DO 74 P=1, NB''
TO 76
LT. NT(P))GO
IF(NT(Q).
IW=JS(NT(Q))
IV=JS(NT(P))
DO 75 N=1, IW
DO 75 M=1, IV
75 BM(M+JS(NT(P))*(N-1))=CM(NR+M+NS*(NC+N-1))
IF(NT(Q).
NE. NT(P))GO
TO 77
CALL STORE(5pADDR(NT(P)),
JS(NT(P))*JS(NT(P))OBMl)
GO TO 76
77 IC=ADDR(NT(P))+JS(NT(P))**2
IW=NST(NT(P))
DO 79 N=1, IW

180.

80
79
76
74
73

C
C

IF(IA8S(MAP(NT(P),
N)). LT. NT(P))GO
IF(IABS(MAP(NT(P),
N)). NE. NT(Q))GO
CALL STORE(5, IC, JS(NT(P))*JS(NT(Q)),
IC=IC+JS(NT(P))*JS(IABS(MAP(NT(P),
CONTINUE
NR=NR+JS(NT(P))
CONTINUE
NC=NC+JS(NT(Q))
CONTINUE

TO 79
TO 80'
BM, l)
N)))

REDUCTION OF JOINT K COMPLETE

END OF BASIC INVERSION BLOCK
GO TO 60

C
C

EVALUATE DISPLACEMENTS IN REVERSE ORDER OF REDUCTION

240 K=ORD(NE-I+l)
IF(I.
EQ. 1. AND. NINTS. EQ. O)GO TO 98
GO TO 90
C EXTRACT KBA BLOCK BY BLOCK INTO CM, PUT (INVERT KBB) IN
C BM, WA IN DM
91 IF(NS. EQ. O)GO TO 98
BM, l)
CALL STORE(3, ADDR(K), JS(K)*JS(K),
NR=O
DO 93 L=1, NB
DO 94 P=1, NILS
IW=JS(NT(L))
DO 94 Q=1, IW
94 DM(NR+Q+NS*(P-1))=WJ(WADDR(NT(L))+Q-1+JS(NT(L))*(P-1))
93 NR=NR+JS(NT(L))
RESULT IN CM. DISPLACEMENTS
EVALUATE (INVT. KBB)KBA(WA)
IN DM AND WJ
JS(K), NS, BM, CM, A)
CALL XMULT(JS(K),
NS, NILS, A, DM, CM)
CALL XMULT(JS(K),
IV=JS(K)*NILS
DO 92 J=1, IV
DM(J)=WJ(WADDR(K)+J-1)-CM(J)
92 WJ(WADDR(K)+J-1)=DM(J)
98 CONTINUE
60 CONTINUE
c
C EVALUATION OF JOINT DISPLACEMENTS COMPLETE
C
STORE(9,1, ISTOR, A, l)
IF(JOB. GT. 30000)CALL
IF(JOB. GT. 30000. AND. ISTOW. LE. 100)WRITE(8,256)(WJ(I),
1 I=1, ISTOW)
lOE12.4/))
256 FORMATU/100H
IF(B. LT. 5.0)GO TO 261
LT. 1)8=0.0
IF(NINTS.
IF(NINTS. LT. 1)GO TO 263
141-FORMAT(I4,7El5.6/)
TRANSFER ALL BLOCK DATA TO TAPE
NSTOR=ISTOR+1-ADDR(1+NJS-NINTS)
NSTOW=ISTOW+1-WADDR(1+NJS-NINTS)
155 FORMATMOH RETAINING, 2I6/)
NSTOW
WRITE(8,155)NSTOR,
WRITE(10)NMS, NJS, NMTS, NJOLD, NPREV, NINTS, NE, ISTORpISTOW
1
NSTOW
pNSTOR,
CALL STORE(6,1, ISTOR-NSTORpA, l)
WRITE(10)(ITN(I),
JA(I),
JB(I),
JC(I),
QC(I), RC(I), PCA(I),

181.
1 PCB(I), I=1, NMS)
WRITE(10)(JS(I),
JDF(I),
ORD(I), ADDR(I), WADDR(I), NST(I),
1 XL(I), YL(I),
I=1, NJS)
ZL(I),
WRITE(10)(EY(I),
AR(I), GR(I), GQ(I), GQR(I), GJ(I),
QS(I),
1 RS(I), I=1, NMTS)
J=NJS-NINTS
K), K=1, IM), I=1, J)
WRITE(10)((MAP(I,
J=ISTOW-NSTOW
I-21, J)
!JRITE(10) (WJ (I)
C
C

MOVE ALL INTERFACE TERMS TO BASE OF STORAGE

DO 272 I=1, NINTS
S=NST(I)
K=I+NJS-NINTS
NM(I)=NM(K)
NST(I)=NST(K)
JS(I)=JS(K)
JDF(I)=JDF(K)
XL(I)=XL(K)
YL(I)=YL(K)
ZL(I)=ZL(K)
NT(I)=l+ADDR(K)-ADDR(1+NJS-NINTS)
P=NJS-NINTS
Q=O
T=NST(I)
IF(T. LE. O)GO TO 271
DO 84 J=1, T
N=IABS(MAP(K, J))
IF(N. LE. P)GO TO 59
Q=Q+l
J)/N
MAP(I, Q)=(N-P)*MAP(K,
GO TO 84
59 NST(I)=NST(I)-l
84 CONTINUE
271 IW=Q+l
IF(IW. GT. S)GO TO 272
oO 85 J=IW, S
35
J)=O
MAP(I,
,
272 CONTINUE
NT(NINTS+1)=NSTOR+l
DO 86 I=1, NSTOW
86 WJ(I)=WJ(WADDR(1+NJS-NINTS)+I-1)
NSTOR, A, l)
CALL STORE(8, ADDR(1+NJS-NINTS),
NJOLD=NJOLD+NJS-NPREV
NPREV=NINTS
STORE(9,1, ISTOR, A, l)
IF(JOB. GT. 30000)CALL
IF(JOB. GT. 30000. AND. ISTOW. LE. 100)WRITE(8,256)(WJ(I),
RETURN
261 WRITE(8,140)
140 FORMAT(///24H
DISPLACEMENTS AT JOINTS/lX,
24(lH*)/lH
Q=7
IF(NBMS. LE. O)Q=6
DO 23 K=1, NJS
IF(JS(K).
EQ. O)GO TO 23
P=K+NJOLD-NPREV
IV=JS(K)*NILS
DO 22 I=1, IV
22 BM(I)=WJ(WADDR(K)+I-1)
IF(JS(K).
NE. Q)CALL UNSORT(Q, JDF(K), NILS, BM)

I=1,

ISTOW)

182.
DO 21 J=1, NILS
N=Q* 0-1)
+1
IW=N+Q-1

21 WRITE(8,141)P,
(BM(M), M=N, IW)
IF(P. EQ. JCN)CRD=BM(JCDF)
23 CONTINUE
END

183.
SUBROUTINE

STORE

---------------------SUBROUTINE
STORE(IT,
IADDR, ISIZE,
ARG, II)
DOUBLE PRECISION
ST, ARG
DIMENSION
ST(16000),
ARG(800)
COMMON/XST/ST
IF(ISIZE.
EQ. O)RETURN
IT
GO TO(1,2,3,4,5,6,7,8,9),
1 DO 10 I=1, ISIZE
10 ST(IADDR+I-1)=ST(IADDR+I-1)+ARG(II+I-1)
GO TO 99
2 DO 11 I=1, ISIZE
11 ST(IADDR+I-1)=ARG(II+I-1)
GO TO 99
3 DO 12 I=1, ISIZE
12 ARG(II+I-1)=ST(IADDR+I-1)
GO TO 99
4 DO 13 I=I, ISIZE
13 ST(IADDR+I-1)-'20
GO TO 99
5 DO 14 I=1, ISIZE
14 ST(IADDR+I-1)=ST(IADDR+I-1)-ARG(II+I-1)
GO TO 99
6 WRITE(10)(ST(IADDR+I-1),
I=1, ISIZE)
GO TO 99

7 READ(10)
(ST(IADDR+I-1),
I. =l, ISIZE)
GO TO 99
8 DO 15 I=1, ISIZE
15 ST(II+I-1)=ST(IADDR+I-1)
GO TO 99
140 FORMAT(/12H PRINT STORE)
9 WRITE(8,140)
902.4)
179 FORMAT(I5,2H
IW=(ISIZE+8)/9
00 144 I=1, IW
J=9*I-8
(ST(J+9*(I-1)),
J=1,9)
144 WRITE(8,179)J,
WRITE (8,141
141 FORMATOH
99 RETURN
END

184.
SUBROUTINE
-----------------------

UNSORT

SUBROUTINE
UNSORT(JS,
IJDF, NILS,
DOUBLE PRECISION
B, C
INTEGER*4
IJDF, JDF, K, IL, IM
IT(7)
DIMENSION
13(49), C(49),
COMMON/US/C
JDF=IJDF
M=O

2
8
7
3

IL=JS
IM=10
K=IM**(IL-1)
DO 5 I=1, JS
IT(I)=O
IF(JDF. LT. K)GO TO 5
M=M+l
IT(I)=2
JDF=JDF-K
K=K/IM
IB=O
Do 3 J=1, JS
GT. 1)GO TO 8
IF(IT(J).
Do 2 I=1, NILS
O
C(J+is*(I-1))=O.
GO TO 3
IB=IB+l
Do 7 I=1, NILS
C(J+JS*(I-1))=B(IB+M*(I-1))
CONTINUE
IW=JS*NILS
Do 4 I=1, IW
B(I)=C(I)
RETURN
END

B)

185.
SUBROUTINE
SORTSQ
---------------------SUBROUTINE SORTSQ(IQ, IJR, TEIIM)
DOUBLE PRECISION TERM
INTEGER*4 IR, IJR, K, L, M
DIMENSION TERM(49), I(7)
L=IQ
M=10
IR=IJR
-K=M**(L-1)
DO 3 J=1, IQ
I(J)=O
IF(IR. LT. K)GO TO 2
IR=IR-K
I(J)=2
2 K=K/M
3 CONTINUE
IC=o
DO 6 J=1, IQ
LT. 1)GO TO 6
IF(I(J).
DO 5 KK=1, IQ
IF(I(KK).
LT. 1)GO TO 5
IC=IC+l
TERM(IC)=TERM(KK+IQ*(J-1))
5 CONTINUE
6 CONTINUE
RETURN
END
--------------------SUBROUTINE
SORTG
--------------------SUBROUTINE SORTG(IAQ,. IAR, IBQ, JBR, TERM)
DOUBLE PRECISION TERM
INTEGER*4 IAR, AR, IBR, JBR, K, L, LB, M
IB(7)
DIMENSION TERM(49), IA(7),
L=IAQ
LB=IBQ
M=10
IAR=JAR
IBR=JBR
K=M**(L-1)
DO 6 J=1, IAQ
IA(J)=O
IF(IAR.
LT. K)GO TO 2
IAR=IAR-K
IA(J)=2
2 K=K/M
6 CONTINUE
K=M**(LB-1)
DO 3 J=I, If3Q
IB(J)=O
IF(IBR. LT. K)GO TO 4
IBR=IBR-K
IB(J)=2
4 K=K/M
3 CONTINUE
'LC=O
DO 7 KK=1, IBQ
IF(IB(KK).
LT. 1)GO TO 7

186.
DO 5 J=1, IAQ
IF(IA(J).
LT. 1)GO TO 5
IC=IC+l
TERM(IC)=TERM(J+IAQ*(KK-1))
5 CONTINUE
7 CONTINUE

RETURN
END
SUBROUTINE

XTRAN

SUBROUTINE XTRAN(M, N, B, C)
DOUBLE PRECISION B, C
DIMENSION B(M, N), C(N, M)
DO 2 I=1, M
DO 2 J=1, N
J)
C(J, I)=B(I,
RETURN
END
XMULT
SUBROUTINE
-------------------SUBROUTINE XMULT(L, M, N, A, B, C)
DOUBLE PRECISION A, B, C
DIMENSION A(L, M), B(M, N), C(LN)
DO 2 I=11L
DO 2 J=1, N
C(I, J)=O
DO 2 K=1, M
K)*i3(K, J)
C(I, J)=C(I,
J)+A(I,
RETURN
END
-------------------SUBROUTINE XINVT
-------------------B, C, DET)
SUBROUTINE XINVT(IP,
DOUBLE PRECISION B, CAP, AT
DIMENSION B(49), C(49)
IF(IP.
EQ. 1)GO TO 9
IW=IP*IP
DO 2 I=1, IW
2 C(I)=o
DO 3 I=1, IP
3 C(I+IP*(I-1))=1.0
IW=IP-l
DO 5 IQ=1, IW
K=IP-IQ+l
AP=B(K+IP*(K-1))
IK=K-1
DO 5 I=1, IK
AT=B(I+IP*(K-1))/AP
DO 5 J=1, IP
B(I+IP*(J-1))=B(I+IP*(J-1))-B(K+IP*(J-1))*AT
5 C(I+IP*(J-1))=C(I+IP*(J-1))-C(K+IP*(J-1))*AT
MATRIX NOW REDUCED TO UPPER TRIANGLE
DO 6 IQ=1, IW
AP=B(IQ+IP*(IQ-1))
IK=IQ+l
DO 6 I=IK, IP

FORM

187.

AT=B(I+IP*(IQ-1))/AP
DO 6 J=1, IP
B(I+IP*(J-1))=B(I+IP*(J-1))-B(IQ+IP*(J-1))*AT
6 C(I+ip*(j-1))=C(I+IP*(J-1))-C(IQ+IP*(J-1))*AT
MATRIX NOW REDUCED TO DIAGONAL
DET=1.0
DO 7 I=1, IP
DO 12 J=1, IP
12 C(I+jp*(J-J))=C(I+IP*(J-1))/B(I+IP*(I-1))
7 DET=DET*B(I+IP*(I-1))
GO TO 8

9 DET=B(l)
C(1)=1.0/B(l)
8 RETURN
END

FORM

188.

APPENDIX A. 5.1.
Calculation

Three

cases

were

stiffness

warpinR

of

by Khan

considered

and

factor

carry-over

(79)

Tottenham

and

as

follows:
(i)

Beam fully
The boundary
at
at

The warping

the

at

restrained
conditions

are:

x=o:

ex =0v

and

x=k:

6x =0.

and

p is

by,

stiffness

given

far

end

x=1
x=o

and

carry-over

(ii)

Beam with

far

The boundary
at
at
The warping

The

carry-over

kZ
kZ

k9,
- sinh

end

restrained

conditions

for

b)

(A-5.2.

are,

B=o

(A-5-4.

b)

(A. 5.5.

p is

by,

equal

case

0x

a)

and

C is

this

rotation

and

x=Z:

)2 s inh Ict
kk _(kk
cosh kZ - sinh

against

(A-5.4.

6x =o.
given

(A. 5-. 3.

6x=1

x=o:

factor

IcT

ex =ov

stiffness

(A-5-l.

C is,

factor

sinh
T-9, cosh

a)

EIW

k9, (kR, cosh kZ - sinh kt)

+ cosh kZ)
kk + 2(l
kksinh
the

(A-5-l.

EI
kP.

to

zero

in

this

case.

189.

(iii)

Beam

far

with

The boundary

conditions

at

x:

The warping

and

B=o,

and

is

stiffness

given

k9,
k9, sinh
cosh k9,

The carry-over

APPENDIX A-5.2.
The

These

functions

=1

(A-5.6.

a)

T=o

(A-5.6.

b)

(A

by,

w
pl

to

equal

zero

are

in

1-4

given

the

(i

(2 =s

and

where

matrix

a)

(A. 5.8. b)

+ C)

(A-5.8.

c)

(A. 5.8. d)

functions

S1+

case.

(A-5.8.

4=s
the

this

5.7.

by,

3 = sc

where,

in

bimoment-warping

(1
C)
+72
+
2S
=

are,

functions

constant

case

EI

C is

factor

this

of

0=o,
xx

o:

at

free

end

S and

2-r coth
tanh T-T
2T
sinh

C are
2T)

2T
sinh
2T - 2T cosh

2 VI'GJIEIw

given

by,
(A. 5.9. a)

2T

(A-5-9.

(A-5-10.

b)

190.

APPENDIX A. 9.1.
Calculation

diaphraRm

of

flexibility

Notations
Length

the
b

of

corrugations

Depth

in

panel

of

direction

direction

the

to

perpendicular

to

parallel

the

(mm).

corrugations
alva 4

in the
(mm).

panel

to

factors

Correction

for

allow

intermediate

members.

perpendicular
Non-dimensional

for

constant

sheeting

distortional

flexibility.
E

Modulus

Poisson's

Sp

Flexibility

of

of sheet
(mm/kN)
-

sh

Flexibility
:

Number
Number

np

Total

Factor

sheet

of

seam fasteners

a sheeting

or

a perpendicular
A

Cross

sectional

Angle

of

Height

Net

of

members

of

material.

perpendicular

per

widths

member

(mm/k. N).

panel.

per

perpendicular

seam line.
members.

the profile
is fixed
upon whether
as
the
decking
on
number of sheet to
and
in a sheet width.
member fasteners
2)
(mm
of an edge member.
area
of a diagonal
to the direction

inclination

respect
model with
to the corrugations.
Pitch

to

of

depends

frame

a seam fastener

of

number

the

of

ratio

fastener
Ss

elasticity.

(mm).

corrugations

of profile
thickness
of

(mm).
sheeting.

member

of

perpendicular

the

191.

A. 9.1.2.

2-. 5
cad
Et2.5
Tables

for

dimensional
A. 9-1.3.

a,

to

distortion

the

of

sheeting

profile

a4

9.1)

b2

the

correction
K are
factor
due

Flexibility

2a(l

V)
Etb

1.2

A. 9.1.4.

due

Flexibility

Flexibility
perpendicular
2a-S

2.1

factors
presented
to

shear

a1 and a4 and
in reference
strain

in

for

the

non-

(13).

the

sheet

[I+L2h/d)]
(A. 9.2)

due to movement at
member fasteners

the

sheet

to

-P
p

(A. 9-3)

b2
A. 9-1-5.

2.2

due

Flexibility

movement

in

the

seam fasteners

2S,: S (n, h - 1)
). p
2n
sSp+1npS
due

Flexibility

A. 9.1.6.

to

9-4)

to

axial

strain

in

the

frame

2a3
-3E-A
2
f
A. 9.1.7.

force

Area

in

As shown in
the diagonal

E-A d
Y'd

9-5)

diagonal
simulatinR
of a sheet nanel

members

fig.

l. d.

member

A-Cose

is

the

vertical
by,
given

the

ove

component

snear

of

the

(A. 9.6)

L92.

then
the

when Q=1.0

CosO

A=C

frA,

subsituting
takes
diagonal
member

kN

the

area of
the form,

Pd

(A. 9.7)

Cos e

EC

F ig.

(A. 9-7)

that

be noted

A. 9.1.

has been
equation
should
derived
for the case when the two edge chords
are parallel.
the
is
due
to
the
In this
of
edge
members
strain
axial
case
In the trapezoidal
bending.
in-plane
of the truss
panels
simulating
the inclined
the diaphragm,
edge member has its
own sharc
in the panel.
force
As the axial
in carrying
the shearing
It

large
is
member
stiffness
the diagonal
member, the vertical
caused by shear can be neglected
member

in

this

of

can be calculated

using

to

comparison

that

of

component
of the strain
and the area of the diagonal
(A.
9.7).
equation

APPENDIX A. 9.2.
Calculation
truss
model

F,

F,

of

and FL are

vF

forces

is
the
the

The

from

the

of-the

results

simple

fastener
-4-

finite

method;

the

component
of
vertical
the sheet and
between
force
ICI
frame members at joint
,

FD Cos e-

1+

the

by the

given

element
V

forces

fastener

[(
-1
F

F +F

maximum fastener

(A. 9.8)

FV

+F

(A. 9.9)

Fig.
force

at

joint

ICI

takes

the

A. 9.2.
value,

(A. 9.10)

193.

REFERENCES

National
2.

and Gere, J. M., "Theory

Book
McGraw-Hill
2nd Edition,

Bleich,

Co.,

New York,

"Elastic

Stability

Book
N. S.,

Trahair,
thesis

to

presented

P. T.,

Brown,

and

University

for

the

of
Institution

No.

1,

Journal

No.

July

Livesley,

"The

Application

to

Buckling
Division,

some problems
Engineer,
vol.

J. D.,

"Stability,

Journal
August

of
1962,

1968,

Frames",

R. K.,

K. H.,

Solution

Integral

N. S.,

Division,

Chu,

of

Trahair,

101,

ST4,

Doctor

fulfilment

Transactions,
No.

CE10,

2,

Cantilevers",
of Propped
Institution
of Engineers,

Apr.

Vol.

No.

of

Vol.

ASCE,

Renton,

in

Australia,

in

Buckling

Analysis",

10.

P.,

Vacharajittiphan,

Structural

Sydney

Engineering

and
Plane

Computer

9.

CE10,

Vol.

Australia,

8.

of

Civil

Transactions,

Engineering

Lateral

Structures",

Frame

"Finite

Stability

"Elastic

N. S.,

Trahair,
Civil

7.

of Elastic
Co., New York,

pp. 193-196.

1968,

Oct.

Engineers,

of

degree

N. S.,

Trahair,
Equations",

Differential

of

the

of the requirements
Nov. 1967.
Philosophy,

6.

1961.

Structures",
of Metal
1952.
N. Y.,

Strength

"Buckling

F.,

McGraw-Hill

5.

D. C.,

Washington

1961.

N. Y.,

4.

Edition,

2nd

S. P.,

Timoshenko,
Stability",

3.

Beams",

Elastic

Walled
V. Z. , ?'Thin
Foundation,
Science

Vlasov,

of
34,

ST7,

PP-94-100.
"Analysis
of
1975,

the

PP-1497-1516.
Digital

of an Electronic
Analysis".,
Structural
No.

Space

-of
the Structural

1,

January

Frames

of
Structural

The

1956,

pp. 1-12.

by Computer

Division,

ASCE,

Vol.

88,

pp-81-103.

R.
H.,
"Large
Deflection
Rampetsreiter,
and
the
Journal
Structural
Frames",
Space
of
of
98, No. ST12, December 1972, pp. 2701-2722.
ASCE, vol.

194.

Instability
Z. , and Naim, M. M. , "Elastic
Structural
the
Journal
Frames",
of
Space
1980,
July
ST7,
No.
106,
Vol.
ASCE,

Razzaq,
Unbraced
Division,

of

pp-1389-140012.

ST3,

Vol.

108,

No.

Aly,

G. A.,

and Sato,
for
matrix

Stiffness
Journal
Feb.
14.

of
March 1982,

of
1984,

PP-559-577.

of "Generalized
A. B11,
Beams - by Chaudhary,
Thin-Walled
ASCE, Vol.
110, No. 2,
Division,
Structural

the

Discussion

N.,

PP-421-422.
Stlbe
gedrckter
zentrisch
LuftfahrtBereich",
im elastischen

"Drillknicken

R.,

Kappus,

ThinASCE,

Division,

Structural

the

Journal

Beams",

Walled

13.

A. B.,

Chaudhary,

for

Matrix

Stiffness

"Generalized

Profil
offenem
1937.,
Forschung,

N. A. C. A.

Translated'in

mit

Mein. 851,

Tech.

1938.
15.

W. P.,

Rodden,

Formulation
Journal

of
Astronautics,

16.

Approach

6,

June

Structural

Analysis

Journal

the

91,

vol.

Journal

G. W.,

Hicks,
Journal
No.

"Matrix

No.

Using

American
vol.

B. J.,

of
1963,

"Consistent

J. S.,

Archer,

Problems",

19.

Institute

No.

of
Astronautics,

18.

1,

ST6,
"Finite

the

3,

of
December

1963,

Element

"Discrete

Journal

Analysis",

of

and Astronautics,

Matrix

Formulations

Finite

Element

10,

of
October
of

1965,

Element
1967,

pp. 71-80.

Techniques",

Aeronautics

and
pp. 1910-1918.

1965,
Structural

Stability
ASCE,

pp-141-157.

Elastic

Division,

for

Division,

Structural

Structural

of
ST6, December

and
pp. 225-227.

PP-1437-1439-

Formulation
the

Aeronautics

Aeronautics

Institute
No.

Beam",

Timoshenko

of
January

J.,
and Padlog,
Instability
Structural

to

1,

Hartz,

No.

1,

vol.

American

vol.
17.

Loaded
of an Axially
Institute
the American

R. H.,

Gallagher,
the

J. P.,

the

of

Coefficients

P. G., "A Matrix

and Bhuta,
Transverse
Influence
Structural

Jones,

Buckling
ASCE,

vol.

Analysis",
93,

195.

20.

Analysis
Element
P. V. , "Finite
R. H. , and Marcal,
Structural
the
Journal
Structures",
of
nonlinear
of
1968,
94, No. ST9, September
ASCE, vol.
Division,
Mallett,

pp. 2081-210521.

Krahula,

Bent

the

Journal

of

J. L. , "Analysis
of
Method",
Element
Finite
and Astronautics,

Aeronautics

Twisted

and

American

of

No.

5,

vol.

Using

Bars

Institute

7,

1967,

June

pp-1194-1197.
22.

for

and Structures,

Kraj'cinovic,

D.,

Structures",

Journal

96,

vol.
24.

Barsoum,

25.

Journal

27.

28.

and Klingner,
Journal
Beams",

Steel
96,

No.

of
Computers

99,

N.,
No.

Division,

ST9,

1970,

pp-1919-1932.

Tall,

and
Journal
ST12,

L.,

Frames",

10,1979,

"Linear

Stability

91,

G.,

"Torsional-Flexural

Analysis

of
ASCE,

Buckling

of the Structural
1965, pp-103-124.
ST4, August

Journal
No.

Journal

PP-517-529.

Division,
of the Structural
December 1973, pp. 2439-2458.

A.,

vol.

Torsional

Lateral

vol.

and, Structures,

ASCE,

Division,

Structural

T.,
"Elastic
and Usmi,
Unbraced
and Braced Planar

ASCE,

13,

Buckling

Lateral

"Elastic
the

of
September

and Winter,
Members",
Thin-Walled

Chajes,
of

R.,

W. A. M.,

Beam-Column",

29.

Engineering,

G.,

Tebedge,

vol.

in

Element
K. C., "Finite
and Rockey,
the Buckling
of Columns and Beams",
Science,
j ournal
of Mechanical
vol.

for

Buckling
of

Methods

Analysis
Problems",

Stability

Numerical

of

Element

D. A.,

Powell,

Alwis',

"Finite

pp-335-352.

,International
1971, PP-945-949.

vol.

of the Structural
January
1970, pp. 107-121.

R.,
and Gallagher,
and Torsional-Flexural

Nethercot,

of

of

of Thin-Walled
ASCE,
Division,

Analysis

R.,

Solutions

26.

Force

ST1,

2,1970,

Technique

Journal
639-662.
pp.

5,1969,

vol.

"Matrix

Element

International

No.

of Torsional
International
vol.

Discrete

Assemblages",

Thin-Walled

Solids
23.

"A Consistent

D.,

Krajcinovic,

196.

30..

Torsional"The
J.
T.
Poskitt,
E.
K.
Horne,
M. R. , Hoh,
,
, and
Members",
Prismatic
Thin-Walled
Behaviour
Flexural
of
13,
Science,
Mechanical
Vol.
Journal
of
International
641-657.
pp.

1971,
31.

Culver,

Journal

Equations",
vol.
32.

Buckling

Flexural

Solution

"A Direct

J. D.,

Renton,

of
April

ST2,

No.

92,

Bending
of the Biaxial
ASCE,
Division,
the. Structural
1966, pp. 63-83.

Solution

"Exact

C. G.,

The Structural

Axially

of
Engineer,

38,

vol.

Torsional-

the

of
Loaded

Bars",

Thin-Walled
9,

No.

1960,

September

pp. 273-276.
33.

Pekoz,
of

Thin-Walled

Structural

the

of
May 1969,
34.

M. G.,

Salvadori,

Lee,

G. C.,

lity

of

36.

Aug.

Wales,

of

Nethercot,

D. A.,

the

118,

Literature
Research

Lateral

elastic
vol.,

of
ASCE,

Division,

Instabi-

on the Lateral
Bulletin,
Council

Thesis

of

Beams Supported
to

presented

the

K. C., "A unified

approach
and Rockey,
The
Structural
beams",
buckling
lateral
of
49, No. 7, July
1971, pp-321-329.

and Kitipornchai,
Stepped
I-Beams",

Buckling

pp. 1165-1182.

1953,

1970.

N. S.,

Trahair,

I-Beams",

Stability

Ph. D.

Sheeting",

University

of
October

1960.

"The

D. A.,

Engineer,
38.

of
Welding

63,

No.

Nethercot,

to

ST5,

No.

95,

vol.

Buckling

vol.

"A Survey

by Corrugated

37.

Journal

Load",

Eccentric

ASCE,

"Lateral

ASCE,

Beams",

Series

Under

Division,

Buckling

"Torsional-Flexural

PP-941-963.

Transactions,
35.

G.,

and Winter,
Sections

T. B.,

vol.

97,

No.

S.,
Journal
ST10,

Lateral

"Elastic
of

the

October

Structural

1971,

pp. 2535-2548.
39.

Kitipornchai,
of

Tapered

ASCE,

vol.

and Trahair,
I-Beams, " Journal
S.,

98,

No.

ST3,

N. S.,

"Elastic

the

Structural

of
March 1972P

pp. 713-728.

Stability
Division,

197.

40.

N. S., "Stability
and Trahair,
Journal
Cantilevers",
of
Beams
and
symmetric
January
No.
ST1,
98,
ASCE,
Division,
vol.
ral

Anderson,

J. M.,

Mono-

of
the

Structu-

1972,

pp. 269-286.
41.

Properties
S.,
"Buckling
N.
Trahair,
and
Structural
the
Journal
I-Beams",
Monosymmetric
of
of
106, No. ST5, May 1980, PP-941-957.
ASCE, vol.
Division,

42.

Nethercot,

S.,

Kitipornchai..

beams with mixed end

51, No. 4, April
vol.
43.

T. M.,

Roberts,

of

Roberts,

'T. M.,

45.

Flint,

of

Mechanical

A. R.,

"The

Austin,
Buckling
tions,

47.

Influence

of Elastically
ASCE, vol.
120,
N. S.,

"Stability
The Journal

Restraints",
Australia,
48.

S.,

Yegian,

W. J.,

Trahair,

Strains

Science,

of Beamstf, The Structural

September
1951, pp. 235-246.
46.

of
Engineer,

and Stability
Section",
International
23,1981,

vol.

of

pp. 297-306.

if
Azizian,
Z.
G.,
Nonlinear
Analysis
and
Bars of Open Cross Section",
International

Thin-Walled
Journal

Order

"Second

of Open Cross
Science,
Mechanical

Journal

Buckling

"Lateral

The Structural
conditions",
1973, pp-133-139.

Bars

Thin-Walled

44.

K. C.,

and Rockey,

D. A.,

vol.

37,

No.

25,1983,

vol.

of Restraints
Engineer,
vol.

pp-565-577.

on the
29, No.

and Tung, T. P., "Lateral

End-Rebtrained
I-Beams",

April

1955,
of

Stability
9,

Transac-

pp-374-390.

I-Beams

of the
6, June

of

with
Institution

1965,

Elastic

End

of Engineers,
pp. 157-168.

M. G., "Lateral
Buckling
Salvadori,
of Beams of Rectangular
Proceedings
Under Bending
Cross Section
and Shear",
1951,
Mechanics,
Applied
lst U. S. National
Congress
of
-p-403-406.

49.

Trahair,

N. S.,

Rigid-Jointed
of

Engineers,

pp. 171-180.

"Elastic
Structures",
Australia,

in
of I-Beam Elements
The Journal
of the Institution
1966,
July-Aug.
7-8,
No.
38,
vol.

Stability

198.

50.

N. S.,

Trahair,
lar

Institution

Australia,

52.

N. S.,

Stability

"Elastic

Structures",

"Flexural-Torsional
A. J.,
Thesis
to
presented

Illinois,

at

the
53.

of
ASCE, vol.

Hartmann,
No.

55.

A. J.,
August

Hartmann,

A. J.,

Buckling",
96,

Trahair,

2,

in

degree

Buckling
the

of
EM2, April

"Experimental

Planar
of

fulfilment

of
Philosophy.

of

"Flexural-Torsional

Journal

Lateral

of

University

partial
of Doctor

W. H.,

No.

95,

ASCE, vol.

the

Division,

Study

Structural

1966,

Buckling

of the Structural
1967, pp. 11-26.

Journal
No.

92,

"Elastic

Journal

ST4,

vol.

the

for

1964,

and Munse,
Planar
Frames",

Division,

Beams",

in

A. J.,

Buckling

54.

Urbana,

requirements

Hartmann,

No.

Beams",

Continuous

of

Division,
of the Structural
ST6, June 1969, pp. 1295-1312.

Hartmann,

CE10,

vol.

Journal
No.

Transactions

pp. 167-172.

1968,

Trahair,

of
Engineering

Civil

Engineers,

of

October
51.

Beams",

Continuous

Rectangu-

Narrow

Buckling

"Interaction

pp-37-59.
Continuous

of

ASCE, vol.

of Flexural-Torsional
Division,
ASCE,

of the Structural
July
1970, pp-1481-1493.

ST7,

56.

Buckling
Lateral
of "Elastic
Beams - by Hartmann,
A. J. 11, Journal
of Continuous
of the
Structural
ASCE, vol.
Division,
94, No. ST3, March 1968,
pp. 845-848.

57.

Nethercot,

N. S.,

D. A.,

approximations
Engineer,
vol.
58.

Dux,

P. F.,

Laterally
Division,

Discussion

and Trahair,
for
54,

elastic
No. 6,

"Lateral

N. S.,

The Structural

beams",
June

buckling

1976,

pp. 197-204.

Buckling
S., "Elastic
of
and Kitipornchai,
Journal
Continuous
I-Beams",
of the Structural
1982,
ASCE, vol.
108, No. ST9, September

pp. 2099-2116.
59.

Trahair,

N. S.,

Journal

the

ST12,

"Restrained
Structural

of
Decembe.r 1969,

Elastic
Division,

pp. 2641-2664.

Beam-Columns",
ASCE, vol.

95,

No.

93,

199.

60.

P.,

Vacharajittiphan,
Buckling

Lateral

of

and
Portal

of the
ST5, May 1973,

No.

99,

vol.

"Elastic

Journal

Frames",

ASCE,

Division,

Structural

N. S.,

Trahair,

pp. 821-835.
61.

P.,

Vacharajittiphan,

in

Buckling

Lateral

ASCE,

Division,

Structural

and Trahair,
Frames",
Plane

N. S.,

"Analysis

Journal

of
ST7,

101,

vol.

No.

of

the
1975,

July

PP-1497-1516.
62.

the
63.

Frames",

dissertation

University,

in

64.

65.

Birnstiel,

C.,

of

No.

presented
fulfilment
of

to

fulfilment
of

68.

space
tional

frames

Civil

Engr.

and
using

Conference

ST2,

practice",

Rao,
A.

S. S.,

Wheaton

Crosby
"The
& Co.

Finite
Ltd.,

and
Structural

of the
1974, pp-547-564.

March

Influencing

"Factors

A collection
tin-walled

of

of

1967.

"The

K. I.,

cdmputer
techniques",

sparse
matrix
on Space Structures,

K.,

"Warping

Journal

Structures",

Majid,

University

requi-

Division,
of the Structural
Feb. 1980, pp-491-504.

"Thin-walled

Zbirohowshi-Koscia,
to

69.

A.,

the

Philosophy.
N. S.,

ST3,

State
of

of
and strength
on the stability
Windus Ltd.,
London
Chatto,
structures,

Jennings,

Space-

Oklahoma

papers

67.

Philosophy.

of

Rigid-Jointed

of

No.

Thesis

partial
Doctor

S. B.,

Journal

106,

degree

Iffland,

and

A. H.,

Chilver,

100,

vol.

Stability",

ASCE, vol.

in

and Trahair,
Joints",

I-Section

Division,

1970,

Doctor

of

P.,

at
ASCE,

Frame

66.

degree

Vacharajittiphan,
Distortion

Frames",

presented
in partial

1965,

the

Space

"Stability

E..

Citipitioglu,

of

the

of

requirements

rements

of
in

University,

Lehigh

to

"Analysis

S.,

Morino,

1966,

analysis
InternaDept.

of

Surrey.

"Thin-walled

Lockwood

Beams from

& Son Ltd.,

Element

Method

Exeter,

first

in

theory

London

1967.

Engineering".

edition

1982.

of

200.

70.

71.

McGraw Hill,

London,

First

1982.

Brebbia,

A. J.
"Computational
C. A. , and Ferrante,
,
Problems",
Pentech
Solution
of Engineering
1979.
2nd Edition

the

London,

R. V.

Southwell,

No.

1970,

EM6, December

Theory

Methods
Press

of

1941.

Plot
R. H. , 11Southwell
Division,
the Structural

Leice ster,
Journal
of

the

to

Introduction

"An

,
2nd Edition

Elasticity",
73.

"Computer

Edition

for

72.

R. G.

Tickell,

R. J. , Sawko, F. , and
Engineer",
Methods for Civil
Cope,

Beam-Column"

f or

ASCE,

,
95,

vol.

pp-945-965.

74.

Method for Predicting

S. T. , "The Southwell
Structures1l,
Quart.
Journ.
Loads of Elastic
Critical
XIV, Pt. 2,1961,
Math.,
Mech. and Applied
Vol.
PP-137-153.

75.

Roorda,

Ariaratnam,

the

Journal
No.
76.

Structural

of
EM6, December
J. M.,

Davies,

1967,

ASCE,

of Unbraced
Conference

"Stability

U. S. A.,

Missouri,

Structures,

Plot",

Southwell

92,

vol.

PP-37-48.

Speciality

International

on the
Division,

Thoughts

"Some

J.,

Pallet

Racks".

on Cold-Formed
18-19,1980,
November

Fifth
Steel

pp-409-427.
77.

Pergaman
78.

W., "The
and Merchant,
London 1965.
Press Ltd.,

M. R.,

Horne,

R. C.,

Coates,

79.

Black,

M. M.,

Hill,

in

and use
H. V., Crosby

Frames",

"Structural
and Kong, F. K.,
Ltd.,
London 1980.
Thomas Nelson

Analysis

Beams",

Open-Section
design

"The

of

M. G.,

Coutie,

2nd Edition,

Analysis,

Stability

Thin

building,
Lockwood

and Design
Steel
walled
Edited
Ltd.,

of Thin-Walled
Structures.

by Rockey,
1969,

K. C.,

Their
and

pp. 173-209.

80.

H. M., "Torsion-Bending
M. M. and Semple,
Analysis
Beams",
Thin-Walled
Int.
J. Mech. Sci.
of Continuous
1969,
Pergamon Press
vol. 11, pp. 791-810.

81.

Walker,

Black,

Sections",

A. C.,

"Design

International

and Analysis
Textbook

of Cold-Formed
Company Ltd.,
London

1975.

201.

82.

Khan,

A. H.

distribution
structures

Engrs.,
83.

Feb.

85.

2, Dec.

Part

S.,

Medwadowski,
Journal

84.

H.
"The method of bimoment
Tottenham,
and
,
,
thin-walled
the analysis
for
of continuous
Proc.
Instn.
Civ.
to torsion",
subjected

of
1985,

the

J.,

1977,

pp. 843-863.

"Warping

Division,

Structural

Distribution",

Moment

ASCE,

110,

vol.

pp-453-466.

Davie s, J-M.
Paper TN 454

Torsion-Bending
to

be published

Analysis
Proc.

ICE,

Made Easy"
Part

2.

P.,
Analysis
"Ultimate
Load
Osterrieder,
and
Structures
Thin-Walled
Beam
Three-Dimensional
with
of
Elements",
Stability
Using Finite
Cross Sections
of

Ramm, E.,

metal
report,

Paris
structures
pp. 201-210.

16-17

Nov.

1983,

Preliminary

No.

2,

202.

REFERENCES (II)
1

Young, J. G. and Lawson,

R. M.
Davie s, J-M-,
,
,
"MACE Nursery
University
Load Tests".
of Salford,
unit
Report
Department
Engineering
Ref. No. 74/51,
Civil
of
Bryan,

E. R.

December
2.

1974.
"Calculated
R. M. , Young, J. G.
,
folded
plate
of a nursery
roof
Conf.
Proc.
of Building
on the performance
1974, pp-31-44.
Glasgow Mar.,

Davies,

J. M. , Lawson,
behaviour
of

observed
school".
Structures,
3.

BS449:

of
5.

of
Engineering

Civil

Nilson,

A. H., "Shear

Journal
No.
6.

of

Nov.

ST11,

Ammar,
Steel

the

A. R.,
Shear

Ref.

Report

No.

of
Division,

A. H.,

and Nilson,.
Diaphragms,

Parts

"Analysis
I

Proc.,

Diaphragms",
formed

8.

Rolla,

steel
Mo.,

Bryan,

E. R.,

9.

of
2nd Specialty

structures,
Oct. 1973,
"The

CONSTRADO Monograph,
England,

Light

"Analysis

A. H.,

1980.

Gauge Steel"

ASCE, Proc.

86,

vol.

of Light
Research

and III',
Engineering,
Civil

350 and 351, Department

of
N. Y. Aug, 1972,
Ithaca,
University,

Nilson,

Dept.

1960.

Nos.

7.

use of structuTests.
Loading

Light

Diaphragms

Structural

the

on Pyradomes",
March
80/142,

Tests

"Loading

Salford,

University

A.

Appendix

building".

in

steel

ral
4.

part

for

"Specification

2,1969

and

Apr.

Gauge
Reports
Cornell

1973.

Gauge Steel

Shear

Conference

University

of

on coldMissouri-Rolla,

PP-325-363.

Stressed
Crosby

Skin

Design

Lockwood

Buildings",
of Steel
Staples,
London,

1973.

Davies,
J. M. , and Lawson, R. M. , "The Shear Flexibility
Proc.,
3rd International
Corrugated
Steel Sheeting",
Steel Structures,
Specialty
Conference
on cold-formed
Rolla,
Mo., Nov. 1975,
University
Missouri-Rolla,
of
PP-535-568.

of

203.

10.

Davies,

R. M., "The Shear

and Lawson,
Sheeting".
International
Metal

Deformation

J. M.,

of Profiled
Numerical

Methods

in

Engineering,

vol.

Journal

for

12,1978,

pp-1507-1541.
11.

Shear

Flexibi-

University
Sheets",
of Salford,
of Profiled
Report
No.
Engineering
Ref.
Department
of Civil

82/173,

Davies,

J. M.,

11A.General

Solution

for

the

lity

Aug.
12.

1982.

Davies,
of
No.

the

J. M.
ItSimplified
,
Division,
Structural

ST11,

Nov.

1977,

Diaphragm

ASCE, Proc.,

E. R., "Manual
J. M. and Bryan,
Granada publishing,
Design",
Diaphragm

14.

Davies,

Davies,

Journal
No.

15.

the

of
ST7, July

"Calculation
Structural
1976,

vol.

Journal
103,

pp. 2093-2109.

13.

J. M.,

Analysis",

of Steel
Division,

Skin
of Stressed
1982.
Great Britain
Diaphragm
ASCE,

Vol.

Behaviour",
102,

pp-1411-1430.

SAP4: A computer program for

University
systems.
of linear
Computer Center.

and dynamic analysis

static
Regional
of Manchester