MOMENT CONNECTIONS FOR VIERENDEEL TRUSSES

OF SQUARE HOLIDW STRUCTURAL SECTIONS

MOMENT CONNECTIOMS FOR VIEBENDEEL TRUSSES

OF SQUARE HOLLOW STRUCTURAL SECTIONS

!SJ
YONG KIMG LOO, B.Sc. (Eng.)

A Project Thesis
Submitted to the School of Graduate Studies
in Partial Fulfilment of the Requirements
for the Degree
Mast3r of Engineering

HcHaster University

1973

MASTER OF EIJGI11EERIJ:TG
(Civil Engineering)

( 197 3)

TITLE :

Moment Connections for Vierendeel Trusses

McMASTER UNIVERSITY

Hamilton, Ontario.

of Square Hollow Structural. sections

AUTHOR t

Yong King Loo, B. Sc. (Eng. )

(National Chung-Hsing University)

SUPERVISOR :

Dr. R. M:. Korol

NUMBER OF PAGES :

.lx , 120

SCOPE AND CONTENTS :

A research programme is presented for the analytical.

evaluation of the deflections of Vierendeel trusses comprised
of

semi~rigid

connections under panel feint loadings. The

semi-rigid connections are made of two unequal width square

ESS members welded at right angles.

As the fle.id.bility of the

joints increases when the width ratio is less than 1.0, the
joints are tmable to develop the moment capacity of the web
member and excessive deflections limit functional capability
of the truss. Hence, several types of joint reinforcement are
recommended.

A yield line method is attempted to estimate the

strength capacity of the joint with and without reinforcements.

In adtli tion, a plate analysis forms the basis for estimating

elastic joint stiffness for evaluating anticipated deflections

at mid span.

ii

ACKHO'RLEDGEMENTS

I wish to express my sincere gratitude to my

research supervisor Dr. R. M. Korol for his guidance and
encouragement during the course of tbis research programme.
Financial assistance given by the Civil Engineering
Department of McMaster University in the orm of a teach
ing assistantsbip is thankfully appreciated.
Also, I am grateful to the Computing Center of the

University for allowing me to use its excellent comiuting

facilities.
T~.is

research was sponsored by CIDECT (Comite

International pour le Developpment et 1 1 etude de la

Construction

Tubulai~e)

with direct motivation of the Steel

Comrany of Canada, to whom I extend my sincere thanks.

iii

TABLE OF COl\TTEUTS

CRA-"PTER
I

~
DJTRODUCTION

1.1

Hollow Structural Sections

1.2

Classification of Connections

1.3 Objective of Study

IT

METHOD OF ANALYSIS FOR VIERENDEEL TRUSS WITH

SEMI-RIGID COIDJECTIONS
2.'1

Vierendeel Truss

2. 2

Method of ..iL11a.lysis for Vierendeel

2.2;1

Static Matrix (A)

2. 2;2

Deformation Matrix (BJ

2.2.3

Stiffness Matrix

2. 2; 4

M~atrix { P}

15

2~ 3

Matrix Operations

16

2.4

Limitations On Deflection

17

2.5

Strength Limitations

19

2.6

Design Procedures

21

III

(s)

10

12

THE BUCKLING PROBLEM OF THE HSS MEMBERS

SOME

~ISTAXTCE

AWAY FROM THE JOINT

3.1

General Introduction

3~

Local Bu.ckling of J?late Element of HSS

Compression Flange

iV

36

36

CEAJ?TER

IV

PAGE

STRUCTURAL BEHAVIOUR OF UNEQUAL-WIDTH

COlTIIBCTIONS

48

4~1

Elastic :Behaviour

4~'2

The Strength of Joints Estimated By

Plastic Method

Experimental. Collapse Loads

1 F...e;

11

49

Comparison Between Theoretical. And

4;3

48

51

I11TFLUEnCE OF SOME MEMBER SIZING ON THE

BEHAVIOUR OF

5;:1

JOil~S

General Introduction

59

59

5.2 The Influence of Chord Width-to-tbicknesa

Ratio, b/tc

60

5.3 The Influence of t 0 /tw

61

JOINTS WITH RETITFDRCEMEMTS

66

6.1
6. 2.~1

General Introduction

66

Analysis of Flange-Plate Reinforced

Joint

66

6. 2.:2

Shortening of The Reinforcing 'Plate

67

6.2.!3

Deflection of The Connected :Plate of

Main Member
6.2eL~

69

Simply Supported Rectangular Plate

Subjected to Uniform Line :C.oad

70

Simply Supported

6~5

Rect~ngular

Subjected to Parabolic Line Load

74

Chord Flange Stiffener

76

Ruanched Reinforcement

77

Plastic Mechanisms for Reitiorced

Joints

VII

Plate

SUMMARY A.'rffi C01'TCLUSIONS

77

97

APPENDICES

Semi-Iigid Connection Equation

100

rr

Computer Programme

104

IIr

Loading Equations Expressed in Trigonometric

Series

107

Nomenclature

115

List of References

118

VI

ILLUSTRATIONS
FIGURES

1.1

Classification.of HSS Connectio!l!3

12

Relative Rotation Versus Width Ratio)\. Under

Elastic Condition
A Vierendeel Truss With Panel Point Loads

2.c2

A Diagram Showing Unbalanced Forces and

Displacemen.ts

23

24

End i":!oments

24

Free Body Diagrams of Vierendeel Truss

25

Effects of

x19

(also

x20 )

on Internal End

Rotations
2.6

2~1

2~5

-PAGE

26

Bend:tng of a Prismatic

M~mber

ab .

Semi-Rigid Connection With Relative Rotation

28

i,foraent-Rotation Curve for a Se!!!i-rig.i.d

Connection

28

2.s

A Set of Solution for Eanel Load P=l kip

29

2.9a

xlot of Deflection Versus J. for Truss A

30

Plot of Deflection Versus J for Truss B

31

2.1oa

Stress for Panel Point Load of 4.2k for Tniss A 32

2~10b

St:cess for Panel Point Load of 2. 5k for Truss B 32

3.. l

A Rectangular Plate Subjected to Compressive

Axial Stress

45

VII

FIGu7.ES

-PAGE

A...
b

3.2

Plate Coefficient K as Function of d/b

4~1

Detail Dimensions of Joints

54

4~2

A Yield-line Patte:rn for Applied Moment

55

Alternative Yield Fatte:rn for Applied Momen.t

55

Theoretical Applied: Moment for Collapse

57

Applied Moment Versus Relative Rotation.

58

Load 'Deflection for b/t 0 EUect (Drexel)

64

Load Deflection for t 0 /\y Eff'ect (Drexel)

65

Flange Plate Reinforcement

84

Chord Flange

Stif~ener

84

Haunched Reinforcement

84

5.2

6.2a

Reinforcing Plate Subjected to Hodified

85

Concentrated Load
Tri.angular Pinned Truss in Plate

85

6.2c

Trianglar Pinned Truss

86

6~'3

A Rectangular Plate Subjected to Tra..."?1.sverse

87

Line Load
6~4

6.6

6~8

A Simply-supported Plate and its x-y Axes

88

A Plot of Relative Rotations Versus

90

-:!\...

Load-Deflection Curves

92

Haunch Cuttings

93

Defomecl Fla:1ge Plate Reinforcement on the

Tension

Si~e

of Joint

VIII

94

FIGURES

6.9

J?AGE

Relationships Between Shearing Stress: and

Strain for Rig:,id :Plastic Material

6~10

94

Case (i) Yield Line Pattern for Chord Flange

Stiffener
Case

95

(ii)

95

6;~12

Case (iii)

A.

Conjugate Beam

103

Uniformly Distributed Line Load

113

Parabolic Line Load Acting on a Plate

114

A Parabolic Curve on y-q. axes

114

95

TABLES
2~1

3~1

Static Matrix (A)

33

Deformation Matrix (B)

34

Central Deflections and Code Li.nitations

35

Estination. 0 Buckling Stress

47

Detail Dimensions of" Joints

54

Lengths and A.YJ.gles of Rotation of. Yield Lines

56

Joint Details

63

6.1

Relative Rotations With Various of

6~2

Joint Dimensions

.1\.

89

91

Moment. Capacities of

IX

~Joints

and Web Members

96

CHAPTER I

INTRODUCTION
1.1 HOLIOW STRUCTURAL SECTIONS
The new structural shape of Hollow Structural
Sections (HSS) have gained wide popularity during the last
fev1 yea;rs.

This is mainly because of a number of advantages

namely:
(i)

HSS provide outstanding strength in proportion to

their weight, especially under compressive and
torsional loading.

This advantage is much better

than those of conventional steel shapes.

(ii) The flat and smooth surfaces of rectangular and

square sections f;Tovide easy attacbment for a

branch ESS member, such as bean to column and web
member to chord :nember in trusses.
(iii)The smooth and clean

a~peara.nce

of the sections

is aesthetically pleasing.

(iv) The ease of painting is also a benef'it.

Besides

that, it gives protection to the interior surface

from corrosion if ooth ends of a member are com
pletely ;::ealed.
( v)

In rnul-ti-storey buildings these sections can be

used in conjunction

Yli th

a ci-rculating fluid to

provide fire resistance.

In order to use these sections properly, the be

b.aviour of welded connections made from HSS members must
be known.

Unfoxtunately, the literature on HSS connections

has not yet been adequately available . I'n this vo.lume

attention is focused on the connections of unequal width square*

RSS members welded at :right angles and

sub~equent

truss deflection and stre.ngth<for :design

~~pu.rpo~se.s.

-~,t-~t: some lj;g~t.--W:ill -b_e ~:cas.t .. on.~t.hia: 1nt~:resting

1.2

behaviour on
It ,is

ho.~d

problem.

CLASSIFICATION OF CONNECTIONS
On the basis of structural behaviour, the HSS conn

ections can be divided into two categories, namely the equal

width connections (d/b = 1) and the unequal-width connections
( d/b < 1), whe:re d and b are the widths of the web and chord

members respectively.
These two categories are shown in Fig. 1.1

For the equal-width connections, most of the load

applied on the branch member (o:r web member) is directly trans
ferred from the web plates of the branch to the web plates of
the main member (chord member) as they are ill the same plane

joined together by weld

pl:J.tes of tLe brg,::1ch
the main r:1ember.

ma~eri~l.

~.1ill

Also, the load on the flange

not be applied to the flange plate of

Instead, :iost of the load is tra.nsmi tted

t~'1r

ough the corners of the cor1nection as the flange plate of the

main :nem.bex has

2w

low stif.fness in couparison with t1ie web pls.te

of the main member.

3
7lhen the .1id ths of tt:.e sections are unequal ( d/b < 1) ,
the behaviour of the joint becomes the plate :p:r:oblem with
102.. ding

mainly resisted

b~r

flexure of the cor:nected plate.

The strength 2..nd stiffness of this type of connection are

drastically reduced.
As joint flexibility .increas~s rapidly with d/b, the
joint can no: longer. develop the yield .moment. or. the :Plastic

moment of the cross-section as it does when d: b.

A graph, shown in :Fig. 1.2, is plotted with relative

rotation versus width ratio (constant applied moment) to

show the change of flexibility of the unequal-width connection

with :respect to width ratio.
J3oth ty Jies of connections are basically semi rigid (AISC
The fo:rmer type, hmvever, vlill have a

type I I I com1ection).

larger :rigicli ty in most practical situations and would be, more

or less, expected to behave
:ri ,Qlu
connec1;i ons
....

1. 3

::?

~s

AISC type I connection (i.e.,

( 1)

03JECTIVE OF STUDY

A constant-height Vierendeel truss composed of square

hollovr structural sections with unequal width of joints will

be referred to in the

investig~tion

of semi-rigid connections

made of ESS.
T1li

:cesea..Tch v:ork was ::ioti vated by Stelco (Steel

Co:nr-2.n:r of Canad2) which .fahricates ESS Vie:!'endeel trusses ..

While sufficient information eyists for designers for equal

~nidth

connections, (2) t:~~e S'3.2Je is i:lOt t:t:e case for unequal

width connections.

For tbis :reason, design infoi"Illation is

:requi:::....,ed i f tl:e aesthetic and economical advantages, together

vri th :CTany others, of these ESS trusses are to be

e:~:ploited.

A prime objective of thls research is to assess the

flexibility of a typical Viexendeel truss composed of square

HSS under working load.s to determine whether excessive deflect

ions might pose a problem.

Hence, while a comi:anion research

study also conducted at Mc1Iaster is to assess experimentally

the ultimate moment capacity and flexibility of various types

of connections in HSS, this study will be confined to an an

alytical evaluation of
( 1)

The deflections of a Vierendeel truss comprised of

semi-rigid connections under panel point loading.

(2)

Strength capacity and behaviour of HSS connections,

and,

(3)

The effectiveness of joint stiffeners.

It shoilJ_d be emphasized t:tat Vierendeel trusses have in

general strength :properties and aesthetic qualities that are

attractive to the designer.

There are cases, however, when de

flection limitations may rule out such a stxuctural alternative.

The case oi a deep truss of limited sfa,n, i.e. floor to loor
depth and exterior wall to wall length in office buildings is a

possible application. Two examples will be used to provide con

tinuity of the various cl:lapters of tbis volume..

Both examples

illustrate the need for considering deflection and strength.in

the design process.

--~--------~-I~
Equal Width HSS Connection d

= b-

Unequal Width ESS Connection d < b

Fig. 1.1

Classification of HSS Connections

O.;a

1.0

o.4

Width Ratio, A {A-=d/b)

Fig~'

i;:2

Relative Rotation Versus Width Ratio J\..

Under elastic Conditi.on

CHAPTER II
IiIBTHOD OF A..l'fALYSIS :FOR VIERIDillEEL TRUSS
WITH SEMI-RIGID CON1IBCTIONS
2.1

HERENDEEL TRUSS

A Vierendeel t211ss is a hyperstatic frame composed of

a series of rectangular or trapezoidal panels without diago;ial
members.

The truss was first proposed by Professor Vierendeel

in 1896.

The main function of the truss at that time was f()r

bridges.

In recent years, Vierendeel trusses have been used as

roof trusses .in low-rise buildings such as schools and gymnasiums.

It is also a commonly used structure for
bridges.

overhea~

pedestrian

The Vierendeel truss to be studied in tbis volume is

the one with rectangular panels as shown in Fig. 2.1.

2. 2

~IETHOD

OF AI!"ALYSIS FOR VIERENTIEEL TRUSS

The method consists of putting the slope-deflection

equations in matric form in a sinple and systematic way. ( 3

The explanation of the method is :probably best given

by ilh1strating an e:rample as followsi
Conside::~

ing

~t

~~nel

a Vierencleel trti.Ss with concentrated loads act

poi~ts

as shown in Fig. 2.1

but. is internally indeterminate to the 24th degree.

The

total degree of indeterminacy could easily be deter:nined

by experience or by the fornula

= 3m + r - 3j
n = the degree of indetermi:oacy
n

where

=total

j :

( 2.1)

number of members in the structure

numbers o:f reactions

numbers of joints

Consider the Vierendeel truss in Fig. 2. 2.

As there

are 18 joints (including the supJ;orts) in the truss, there are

18 possible unbalanced moments acting at these joints.
also, there are 8

And

* ;possible unbalanced linear forces acting at

certain joints as shown in Fig. 2.2

The unbalanced moments are

assumed to be in clockwise direction and the unbalanced linear

forces are assumed ei t.her in vertical or horizontal directions

These unbalanced forces (moments and linear forces), denoted as

P with subscripts indicating their individual locations, are due

to the applied loads acting on the truss.

* The
number of possible unbalanced linear forces in a structure
can be determined by considering the number of deg:reesof free
dom in joint translation in the structure. According to the
formula (4) , the totai.number of joints in translation in a
structure is

s = 2j

where

( 2f

= :.leg:ree
,...
j
nmnbe1"
f ::: number
number
h

2h

+ m)

of freedom in joint translation or sways

of joints including supports

of fixed supports
of binged supr.,orts
r = number of roller supForts
l!t = number of members in the structure

As a result of tbese possible unbalanced forces due

to the exteni..al loads, the truss respoms

with joint rotations

and joint tra..11slations wbich are acco:cdingly in the same direct

ions as the unbalanced forces as shown in Fig. 2.2

Fig. 2.3 shows the end moments, denoted as M with sub
scripts, acting on the ends of all members.

in a clockwise direction.

They are all assumed

(The end moments can also be

re~arded

as 'distributed moments' from the point of View of the Moment

Distribution Method).
2.2.1

STATIC

i~~TRIX

Every joint or part of the truss can be isolated as a

free-body-d.iagra.:n so long as the internal forces are in eq

uilibrium with the exteJ:nal forces.
of joints in Figs (2.2) and (2.3).
~Moments=

Consider every same pair

By using the conditions that

O, the following equations can be written as,

P2

--

P3

= M4 + M5 -t-1i3s

:
l?1_o
--

r1i2 + Ivi3 T M36

(2.2)

...
I

= l'il32+ !Yl49

The expression for P19 can be found by considering

the free-body diagram of the top chord continuous member as
shown in Fig (2.4.t).

Using the condition that the sum of all

horizontal forces must be equal to zero, i.e. I:Fx

P19 = - l/H (1\I33;.M34+M35

M49;-:i150)

=O,

we get

10

Similarly, the expressions for

P20 , P 21 P 2 5 can
be obtained by considering the equilibrium cond.i tion o:f the

vertical forces of the free-body diagram of the individual web

member serarately as shown in Ri.g. (2.4.c).

Let {P} be a column matrix showing values of unbalanced

momenta at the joints and unbalanced linear forces.

{v}

Let

be a column matrix showing values of end moments.

Re-writing the ex:Pressions for

P1, P2, P3

P~6

into matric form, we get,

{p}
[A)
{tr}
(26Xl) (26X50) (50Xl)

(2. 3)

(A) is denoted as the static matrix which is defined

as a matrix expressing the balancing moments at the joints and

balancing lateral :forces in terms of the end moments.

The static matrix (A) is tabulated in Table ( 2._J.)

2.2.2

DEFORMATION MATRIX (BJ

Deformation matrix [BJ is defined as a matrix wJxLch

expresses the elastic rotations,.Q, at the ends of all members,

as caused by the end moments M in terms of joint rotations or

translations,

x.
{e}= [B]{x}

( 2.4)

( B) may be established columnwise by considering the

eff'ects of each joint rotation and t:ranslation X on all the

elastic end rotations ff

(BJ is called the defonnation

11

matrix because it is solely based on the effects of the

disturbing external joint rotations or translations on the
flexural defo:rmation at the ends of the members.
Consider, for instance, the effect of

x2 , x3 ,

~'

x26 all equal

x1

(keeping

to zero) will be

and

x2 (keeping all other X's

Similarly, the effect of

zero) will be
and

Turning to the effect of horizontal joint translation,

x19

, as.shown in Fig. (2.5a).

0'33

=9'34
:: 9'43

':::I

8'35 = 6'36

From \vhich

::a

i.o.

= 9'49 = 0'50 :: -,-119

where H is the height of the truss.

The effect of each vertical. joint displacement, for

x20

instance_,
9'1
""
0

3 --

, as shown in Fig~: ( 2. 5b} , is .

fi2

s'17

= e'ia = -

li,O

4 -- \719 - l"V20 -- + ..1.!2..

L x20

0ol

t:J;

where L is panel length.

By considering the effects of each joint rotation

translation in this way, the deformation matrix [B) can be

or

12

established column.wise.
Table(2.2) shows the established matrix (BJ for the
Vierendeel truss.

A comparison between Tables (2.1) and (2.2) shows that

the de.fo:rmation matrix (BJ is the txanspose of the static matrix
(A J

Tbis characteristic is always found in all cases SOilved

by this method.

Therefore,

(2.5)

Renee, it is not necessary to establish directly

ma~rix

(BJ as it can be indirectly obtained by the transpose of the

static matrix (A)

Sometimes, it is worthwhile to write out

a few elements of (BJ to check the correctness of'

2. 2. 3

(A J ~:

STIFHTESS t11ATRIX (SJ

-

The internal end moments Ma and Mb at the ends of an

initially straight prismatic member ab (as shmvn in Fig.2.t6)
can be expressed in terms o.f the two internal end rotations
&a and

~b

by using the slope-deflection equation;

1t[

~{.b

....

2EI

n.

+ --r;4EI

-r;-~a

o
0

(2.6)

where E and I are the Young's modulus and the moment of in$rtia
of the member respectively, and are constant throughout its
length. The elastic end rotation

~is

considered rositive when

13

the rotation is clockwise.

The total angle of :rotation of semi-rigid connected
member ab subjected respectively to couples Ma and r.rb is,
(Appendix

:r] [ 5J

( 2. 7a)

=-MaL + MbL + Mb
6EI

3EI

( 2.7b)

JM

where J is the joint modulus

Solving eqns. (2.7a)
get (Appendix I)

i;:r

fira

= 6EI

(2.7b) for Ma and M0 , we

and

[1)[8]
2L

4L~2-L2

2L

'

~
oa

6EI
.

a.,_

4L'2-L2 "o
I,

( 2.8)

M0 = 6EI L, 2~ ~2 0'b + 6EI 41 1 2_ L2 9'a

4

where

L'= L+3EIZ

and

=1.0/ J

modulus J ,
* property

The joint
of the joint.

or the rotational spring constant ~s a

Mathematically speaking, it is the app
ied moment M divided by.the relative rotation 0 of the connection
when behaviour is elastic,i.e. M/ The relationship bettmen
M and ~ is shown in Fig. 2. 7 The joint modulus is usually
measured in in-kips fer radian.

14
When J=oo for the case of rigid connections, Eqn.(2.8)
coincides with Eqn.(2.6)
Hence, the slo:pe-de.flection equations :for the chord

members are given by Eqn.(2.6). Whereas, for the web members,

Eqn.(2.8) are used.
J3y using Eqns.(2.6).

, the relationship

and (2.s)

between the. end moments and the end rotations for ea.ch meniber
in the truss can be written as ;

M1

M.
2

=4~I 9'1 +
=

2EI fJ
L

~I

0'2

4EI fY
L
2
4EI

rv

Lo3
2EI 9'3
L

2EI

4EI 9'4
L.

( 2.9)

15

A:rranging Eqn.(2.9) in matrix notation, we have

{ M} = (s) {9'}
(50Xl) (50X50)(50Xl)

( 2.10)

[SJ is call the stiffness matrix in which the.internal

-

end moments are expressed in terms of the internai end rotations.

( S) is a square matrix of order (50X50), its elements are given

as the coefficients of { &} in Eqn. ( 2.9)

Matrix {P} had been defl.ned in Section 2.2.1 as a

column matrix showing values of unblanced moments at the joints
and unbalanced linear forces.

Now, returning to the real situa

tion, the truss is only subjected to loading in which all P's

are zero except P20 , P , P , P , P24 , P25 , P26 of equa.1

21
23
22
magnitude acting at the Panel r.oints. Reference is made to
Fig. 2.1

Therefore, matrix { P} can be quickly established as

0

{p}
{ 26xl)

'"O

.1..20

P21
P22

p23

"'"26

"P

( 2.11)

16
The matrix { P} can also be extended to include many
different kinds.of loading conditions. Each column of matrix
(PJ is for each loading condition. If there a.re foundifferent

loading conditions, the matrix (P) will then have four columns.
Therefore (P) is no longer a column matrix if the loading
condition is a combination of loads.

Three ma.trices (A), ( s)and

{ P},

which have been diseussad.

in previous Sections, are al.l input matrices which bave to be

established before solving the problem.

The solutions, or the

output me.trice~ are the displacement matrix

-

{x}

and the inter.oa.1

end-moment matrix { M}
The operation of the matrix calculation is given as
follows.
From the static equilibrium condition
{ P }=

f A) { 1t}

(2. 3)

From the condition of deformation

{e'}= [:a-) { x}

: (AT] tx}
Since (B)= (AT)

(2.4)
(2.4a)
(2.5)

From the slope-deflection equations

{ r~r}

=[s J{e-}

( 2.10)

17
Substituting Eqn.(2.4a) into Eqn.(2.10), yields
( 2.12)

Substituting Eqn.(2.12) into Eqn.(2.3), we get

( 2.13)

Rearranging matrix equation( 2.13), we obtain the output

matrix iX}
( 2.14)

Substituting the known matrix {X} into Eqn.(2~12), we

finally obtain another output matrix

With the results o:f {X} and

{M}.

1M} ,

the analysis is thus

complete.
A set of the solutions for a Iarticular case(see the
detail in Fig. ( 2. Sa)) is plotted on the truss as shown in .

2.4

LIMITATIONS ON DEFLECTION
A computer p:rogramme [Appendix

II] has been set up to

analyse 'the truss with joint moduli varying from J

= 1 x io4 to

( in-kips/rad. ) and with constant panel J;>Oint concentrated

loads. The

dimensions of the adopted HSS are 4X4 for the web

members and BXS for the chord members. Thickness of 1/4

tt

and

1/2 " are used. The panel length and the height oi the truss is

lx10

18

8 ft. in the case of

n thick material and 13 ft. for

i- n

material.

It is recognised that, with the adding of semi-rigid

connections on the joints, the bebaviour of the truss becomes
flexible. That is to say, when the modulus decreases, the
flexibility of the joint increases. As a result, the maximum
deflection of the truss at the central I-Oint becomes more and
more conspicuous. With certain limits on deflection set fo:r:

the tniss, the limits of joint modulus and loads can be found.
Figs. ( 2. 9) are :plots of the central deflection versus
joint modulus to show the iri_fluence of joint modulus on the

deflection of trusses. As can be seen from the Fi.gures. the

cu:rves start to become flat when J> lx10 6 in-ki:ps/ra.d.

Therefore, the connections can be regarded as rigid when J

6
exceeds lx10 in-kips/rad (6)
The limltations on deflection set oy National Building
Code (NBC)

(9] are as follows,

(i)

~ max. < 360

for plastered ceilings

(ii)

~maz. < 3~

for floor bea111s

19

(iii)

~maJ<.< ~~o

(iv)

Smax.< 180

for asphalt roofing

for metal or elastic membrane

type roofing

where L is the span. of a simply supported beam or truss.

With certain limitations on deflection in mind, the

designer can choose bis joint modulus within the required limits.

From Eqn.(2.13), the deflections and rotations for any

joint can be found. O.f Primary interest is

x23

the central vertical deflection of the t:russ.

which represents

Fxom the results

of the computer programme (Appendix , II) and employing the

particular examples of the Vierendeel t:rnsses whose geometries

Table 2.3 illustrates the deflections

were specified earlier,

at mid sian with rigid( J. =oa) and semi-rigid( J

connections for 1

ki~

limiting deflections
2.5

=ix104in-k/rad)

loading at each panel point. The NBC

~or

the specific trusses are also given.

STB:ElTGTR LIMITATIONS

The computer programme refer:red to earlier (Appendix

II ) also defines the elastic stresses at all member ends

according to the formula
../

'!'!~

i:;1

h!.

...c

Q::-t

( 2.15)

where is the outer fibre stress in ksi at the c:ri. ti cal

20

point considered
M is the associated bending moment in in-kips

and

is the pertinent axial force in kips

is the cross sectional area in in. 2

is the section modulus in in.3

-JI

For the two trusses A and B

considered, Figs.(2_.10a)

and ( 2. lOb) indicate the stresses at each cri tica.l Faint for
panel I-Oint loadings of

4~'2

kips and 2.5 kips respectively for

materials tba.t yield at 55ksi. Note that Egn. ( 2.15) bas been
.used in each case limiting the loading such tbat the maximum

stress is equal to the maximum value at the most highly stressed

section. In each case the critical section is the exterior web

member.

For truss A, Plate slenderness is sufficiently small

that the critical buckling stress is approximately equal to

the yield stress. For truss B, although the critical buckling
is somewhat less than yield because of increased slende:tness,

there is very little reduction in the maxi.mum stress. A more

complete description of the critical stress for plate buckJing
is described in Chapter III.
web

mem~)er

Note that the flexibility at the

ends alters the stress somewhat. For semi-rigid

connections, the chord members cany a greater sbare of the

load than do the web members. Consequently the critical r.oint

mentioned r.ias a slightly reduced stress.

A simplified design for the examples will follow in the
~
ft

Truss A is 8 ft. deep using 1/2 in. thick BSS

T:russ B is 13 ft. deep using 1/4 in. tlTI.ck HSS

21

next Section permitting the designer to employ joint flexi bill ty

as a Parameter.
2.6

DESIGN FROCEDDP-ES
For purposes of indicating the method design, the

simplest i:ossi ble geometry and sizing of members are assumed.

The top and bottom chords are assumed to be parallel and of the
same cross section. All web members are assumed to be of iden
tical section, although it was evident from the previous Section
that stiffer end web members would have improved the overall

stress balance of the truss. However, since a deflection crttertcn

is normally im:r:ortant, the examples are merely an attempt to

illustrate the need to consider both strength and deflection in

design.

Sections 2.4 and 2.5 describe deflection and strength

criteria in the design process. For truss A, it is evident

from

Pigs~(2.10)

that a limiting panel fiQint loading of 4.2 kips

is necessary to prevent failure at a section (viz. sections

34 and 50 in Fig.(2.3)
Deflection criteria are normally based on working stress
levels rather than computed mrudmum values. Consequently to
relate the panel

r~int

loadings from a strength roint of view

to deflection limitations, a :reduction to 60% of capacity is

required. IThen the deflection limitation of L/180 is imposed

unde:r working loads, it is evident from Table 2.3 that for

J =oo the actual deflection for 2.5 h.-ips(i.e. 4.2/1.67 kips)

22

is less than the liuliting value. However, for J

the actual deflection of

0~521

=r~10

ft. exceeds tb.at of

in-k/rad

L/180

and is therefore not acceptable. A value of J ~ 2.5xl04"-k/rad

just satisfies the

de~lection

c:ritexion.' Note that the stress

at sections 34 and 50 axe reduced somewhat and hence an itera

tion process may be attempted to refine the correct :panel JX>int
loading and associated J value. This was not pursued in tbis

study because the stress changes were rather small.

For truss B, similar results are evident. The limiting
panel point loading is 2.5 kips to cause a stress of' nearly

55 ksi at the critical section. Again, from either Table 2. 3

or Figs.(2.10), when J = 5x104 in-k/rad., the limiting deflect
ion of .L/180 for 1.5 ki:ps(=2.5/l.67 kips) per panel point is

reached with little change in the stresses throughout.

Truss B

Truss A

Chord Member
Web member

a" x a" x -t-"

.a
a 0 x B"x t"
~ 4" x 4" x !"
.
4 " x 4 ft x i:tf

Panel length

8 ft.

Truss 'Depth

8 ft.

g. ft.

13

ft~

II

lp

ip

ip

f,p

}p
~.

@f!::

tp

I
tp

8L

Fig,. 2;:1__ V-1.e:rendeel. Tru.srf W1 th Panel. Point

i,Load~
f\)
\..I~

24

Pig~l 2~~2

Ac Diagram showing Uhba1aneed Forces and Ili.splacements

(Displacements X's a.re in the same directions as

unbalanced forces pt s)

25
L

>)

Fl.g. (2~-4.'a) Vierendeel Truss With

Panel Point Loading

j
__.,
M3,-t-~
H

-L...

MD+~
\-\

---
I

!::'l:it-'1a
H

----fl"

I
__..,..

-----

~'i-~

f'"l..... "'\.a.

M.,,-ti"'\11:

rt

I
__,._

I
___...,

I
_,...

Mg-+"1,.

M,,-..H4S

"

ttl;'\-o
t;

Fig~ 2:4~:0 Free J3ody Diagrams For Web Members For ZF

=O

....P.,

26

'JI./j+o34'

~,r
I 6

_...

'" &~~

11

..;Hf1
I

I
I

36

....

~f'\&as

38

~'""-;,1

. .,,(i&""'
J

I
I

"""' 11'1'.0a,

.....

~'K
/ ..,~

c...

I
I

~''#\ 641

7).7
Fig-. 2.5.a

,
\\/~e
It~

-1,

'I"

:1~

....

~" 843

~~, 6-+&.
I

I
I

....'f\8.t,S

,"". &.1

"

A/
~

Effect of x19 on Internal End Rotations

While Other X's a.re Zero

...,_...

.L

~-lt'.--"'-\
t - ! ... ~ ~/
~

~4

.,.e

e-;,

&'10

./
l
-ft,7J..--..Y.__,~~---(
'

&;s

;!ii;

' te:

xw
Fig.

2~5~b

Effect of

on Internal End Rotations

Wbile Other x! s are Zero

:Rig'. 2.6

Bending of' .A l\tember ab

28

+
-- -

- _::_?---_

- -- -

~-- -:::z,-

Fig.: 2.7 .a Semirigid Connection with

Relative Rotation

Relative Rotation. <P

Fig. 2. 7 ;b

Moment-Rotation Curve For

A Semirigid Connection

g'

.,/);

-~

if
B@g' = '4-'

Ca) Truss with Panel Load P

= 1 Ki:p

(b) Moment Distribution (:ft-Krrs)

2.3

A Set of Solution For :Panel Load P -: 1 K

<(

~
0..

~ -----~~-..-..._._-~- --~-~-----~- ---~--

I

------'

,...

31
-

~----,...--

...-._, --

'~~''~-~ -~~

'
.. -

----,~~

:'.'\

....

------------..

....

t-r

OJ
""'...
:s

,l
h
~

- - - ~ - . -- '
J

:!

.~

-t
"
~
(W

.
0

a:
l

...0

'

s
7

(\J

II

32

13.02

10.5

4~

62

54~96

55~02

13.44

3.8 6

l
2

-)

2.1

23.94

44.52

( 39~9)

(42.00)
(9 .. 66)

( 30. 24)

2~'1) (19~74)

(16. 38)

(13.02) (4.62) (20.16) (13.86)

(0~ 29)

( 23.1)

It

Fig. 2.1oa : Stresses For Panel Point Loads of 4~'.2 Kips For truss A.
(Note: :Bracketed Hos. Refer to J.=lx.104 ;~~
Plain nos. Refe:r to J= oo )

1.? .. Q

11~0 1~15

54.70

54.725

15.5

6.75

42.0

16.'25 11.5
22~:50

14.75
2.25

13 '
(53.:25)

( 53.75)

(11.75)

(11.25)(1~25)

(41.0)

(22.0)

( 2. 25)

(16.0)(7.5)

Fi.g. 2.lOb : Stresses for Panel I.-0ads of 2.5 Kips for Truss B

11,

-I

~I

r-...;

"> :

~L

~i

~!

~i

~I
._:j

f"'\1

~I

,111\J

~I

-.;l l

Gfl

I1

IA-'.
~

~I
~!

T! J
~I
! I
:r I I

~1

~I

l
I

~~

CJ\ l,

~ :1

]'. 1
~

"it., ij

;i

""'

ri 1~

';

)(

)(

)(

.)<'

!f

)(

)C

-.

)re

)(

~.

>< ff
">f

xI

lI

J(

~I

<i',}.j

)(

~l

<l

.)(

~l
~!

~I

...... i

<J

-..

,.,

<l

Ill'\

"'

.q

-1

,,..""'

.<f

-1 I

<l
4

__.._/rt

-4

<l
<l
<l
'4

-4
<f

33

--

--

)(

~-~

)( >f

x: >f
x

le;.

x
x=j

x
x~

x~

l'1rifM

)(

'1
:} l:~:

.;.....,

1"

~ <:!"'- 0

"'

......

1<

:x
_j_

....

'

......

b!~ ..... ~~~

"1

~ ~L~ ~f 7!f

2 2.

To.b\e

I
I

.3 4

.2.

10 II l,J. 13 14 IE 16 17 /8

19

2fJ

:u 22 23

~ .:i~.2'

1--)(

....

-x

x -)f
6

7
8
9
/0
II

/:L

/3

/4

15
16

17
/8
19
.20
::1/,
.22.

,'
><-X

x-1
x-x
x-x

'

'

33

37
YI

3tl

iI

_L

l ! 11 1 I

l
I

l
I

i
l

[..:tBl

iI

.A

l I I
lI
I

.i.

l l

T!
!

35

;t

Truss A*

Truss B

value

DefilectioIIS:: f:or:_ l~kip at

Panel :Pointa (:Et.)

!Den.e ction......
C:r:i..terton
~<L/360

~<L/3'20.

i> < L/240

..
Pane]~

~ (ft)

0~83

0.85 K.

rr
0~54 A.

!\..

-rr

00,94 K.

K.

Oo61 K.

3.11 K.

1 .. 26 K.

0~96

o.267

1 ..

~<L/1so

?a

1~10 .:."\..
"17"

Table 2.3

'* Truss
jji:.

()

T(.

R_l

1<.

1.67 K.

4.'21 K.

o.355

K.

2::11 K.

2~37

J Value

rp~ +:r* in-Ura..d.

_T_ru~~ A*

o.17a
0.200

:?oint Loads

1
1aOR

K.

00

iJl..

k.,.._

C-Ad

lzJ.04 n
00

1xio4
(:0

n
If

l.x-J04 n
fl
co
1>:rl04

Tl

Ceut:ra.l Deflections and Code Limitations

A is 8 r. deep using

Truss Bis 13

1n

deep using

tbick HSS

i"

tbick HSS

CHA-."PTER III
THE BUCKLIMG :PROBLEM OF TEE HSS MEMBERS

SOME DISTANCE AWAY FROM THE JOINT

In addition to the attention paid in the preceding

Section to the limitation on deflection, one of the important

facts for the design of a long-span Vierendeel truss is the

relationship between the load capacity and the geometry of

the section.

These relationsbips are mostly a function of

instability~'

As the adopted sections are HSS which consist

of thin plate assemblages, the instability problem of the

section becomes more prominent. The top chord member at mid

span is often critical. However, as the last Chapter showed
the ends of the outside web members are stressed most severely

for the trusses examined.

Thus the problem is to find the permissible ratio b/t

of the plate element of the compression flange.

The foregoing mentioned buckling problem is referred

to the section. some distance away from the joint.

The buckling

problem of the section right under the joint is beyond the

scope of t'bis volume and will not be included in the following

discussion.
......,,..

.)0

37

3. 2

LOCAL BUCKLING OF PLATE ELEMENT OF HSS

COMPRESSION FLANGE
TEE GOVEF..NING EQUATION OF PLATE UNDER AXI.A.L LOAD
1

The governing equation of the plate of dimensions a)Q)

under the action of forces in its middle :plane( see Fig.'3~1)
is

(10)
DC -z>4w

'bx4

+2 ~!:~
2+114w4 )+t<xll-w2+4~'tf.w2+21xyo2w ):i: o
..
Y
~x
~y
oxay
0y
J

( 3~'1)

where

=Flexural. rigidity of plate

per unit width

. --E:tl._
-12(1-JJ2)

JJ =Poisson's ratio, and is 0.3 for steel

t =Plate's thickness

=Transverse deflection of plate

4x,~ =Normal stresses in x and y directions respectively
1xy =Shearing stress in the :Plane of the plate
W

Note that the x and y are orthogonal coordinates in the

plane of the Plate,

x being

In the present case

:parallel to the chord member

o~

axis~

HSS compression flange,

only a uniformly distributed compressive stress .&x exists.

Eqn.(3.1) becomes

( 3. 2)

Eqn.(3.2) is only valid within the elastic limit.

When &x exceeds the elastic limit, Eqn. ( 3. 2) bas to be modified.

Beyond the elastic limit, the tangent-modulus Et will be

effective in the x-direction while in the y-direction Young s
modulus E can be assumed to remain valid as oy= 0 Introduc
ing'(=

Et/E,

Eqn.(3.2) is generalised as

( 3.3)
When .Ox is within elastic zone, r( is equal to 1.0

as Et

=E ,

thus Eqn.(3.3) returns to Eqn.(3~2)

The moments. in the x and y directions can be expressed

as
]f
x
M
y

=-

D (

:t

e,2w

~ -ox2

e2W+>'fi-;;y2

<v;r o2w2
+ ~2
)
-ax
ay

GENERAL SOLUTION OF TEE GOVEFJH'NG EQUATION

The general solution. of the governing Equation (3.3)

can be obtained by using the boundary conditions of which the
conditions on the loaded edges are first considered.

(A)

The B.

c.

on The Loaded Edges

The loaded edges are assumed simply supported. The

B.

c.

are thus

39

w] x

=o
=0

~2w J
~2

'

x=a

x=o

=0

( 3.5)

x =a

Where a is the length of the plate in x-direction,

b, appears later, is the width of the plate in y-direction.
The second portion of

of Eqn.(3.4) where Mx
Assume

Eqn.(3~5)

= O and

~2w/~y2

is due to the first

=O

W = Y sin n~x

( 3.6)

which satisfies the above mentioned B.C Y is a function of y.

Substituting Eqn.(3.6) into the governing pa.rtia1
differential equation(3. 3) and solving the resulting ordinary
differential equation leads to the full expression of Eqn.(3.6)
as given below

Wa

c 1 cosh k1Y

+ c2
+

where

sinh k 1 y
c cos k 2 y + c sin k 2 y ) sin np
3
4
( 3.7)

'

'
'

<le

=buckling stress

The constant terms c 1 , .~. c 4 a:re to be determined

by using the B. c. of the unloaded edges

(3)

The B. C. on The Unloaded Edges

The unloaded edges have the following B.

c. ;

40

(i)

The web plates of HSS exert equal elastic

restraint ,"on the unloaded edges. The deflection,

W, corresfQnding to the smallest value of &

an even function of y.

Eqn.(3.7)
(ii)

is

Thus c 2 and c in
4

vanish.

The unloaded edges remain straight.

(iii) The angle of rotation at the edge of the buckling

Plate is equal to the angle of rotation of the

adjoining restraining web plate.

c.

Using the above mentioned B.

leads to the buckling

condition, namely

J/,.+ f

tanh(

~JJt+I n~)
+~~_,,

t Jp.-1 tan(

nty:=

~/P-1

n'f!)

( 3.a)

where c(= a/b and !is the coefficient of restraint. ':)

the edge is rigidly fixed, and

~=oo

=0

when

when the edge is simply

supported.

( C)

Equation for Buckling Stress

4c

A solution of buckling-condition

Eqn.(3.8) for the

e las ti cally re strained plate ( where O< 'f<oo) leads to an

equation for 4 0

( 3.9)

41

or 6

1!. E.fl

(3k
b)

c - 12(1-))2)

where p and q are constants depending on the coefficient of

restraint '$ and k is called the plate coefficient.

The value of k for a square HSS is 5.79 as can be

found from the curve in Fig. 3.2 prepa,red by Hudoba(11J .:

Dete:rminations of Buckling Stress in the Inelastic Range

And the b/t Ratio

In the elastic range where r( 1.0, the buckling stress
can be determined directly from Eqn.(3.9)
In the inelastic range,

on .de

i.e.
where 6 y

=yield

is an unknown as it.: de I-ends

<oy - <Sc )Jc

r(= - - - - <4>y - ~p )~p
stress ;

~P

= pro poTtional

( 3.10)
limit stress

therefore, a method of trial and error will be necessary by

using

( 3.11)

The procedure of calculation is as follows.

For a given HSS section, we can precalculate the value
on the right-hand side of Eg_n. ( 3.11), and then, a table is set

up for the 't- values computed from Eqn.(3.10) with given J . . .

,,,

'1

42

and

-6'0 /,ft

ip and for the cor:resJ.)Onding value of

value of <i'0

!fi

required buckling stress. is found.

When the

in the table is closely approaching the pre

calculated value on the right-hand side of

and ~n
...

Eqn.(3~11),

For HSS steel

the

-'y ::.55

ksi

48 ksi . These values are based on the resul.ts of

te12sile tests carried out by Hudoba

f1J1

and pertain to the

example in Cbapter rr~' Such a table for these values is given

below
4c

48.0
49.0

50.0

51.0

52.0
54~0

54.8
55.0

"'-/H

1~0000

48~0

0.'8750
0.7450

58.0

o.6070

0.4640

0~1600

0.0326

0.0000

52~'5

65.4

76.5

135.0

303.0
00

In the above table it can be seen that 'l'-values vary

from 1.0 to zero. Hudoba showed that when

tm

total strain is

0.57~ that'!':: O.l The corresponding buckling stress d 0 is

about 0.99iy and thus the des:f.o-::n,ed "(value of O.l can be consi

dered reasonable for calculation of the critical b/t ratio in

allowable stress design.
:ponding value of d 0

1t-

be calculated as

o/t
:: 28

For tbis '!-value of 0.1 the corres

is 200, the critical b/t ratio can then

43
where

- O~ 3

29600 ksi

(for steel)

k= 5. 79( for square HSS)

4cf.r:r = 200

ksi.

For our examples of Vierendeel trusses whose members'

dimensions were given in Chapter

Ir,

we can detemine the

buckling stress for the web members by using the following

data and Table 3.1
~y

=55

6p

ksi..

)J

E =29600ksi

t
,.

.de

1.

and

=48 ksi
r

=0.3,

5~79

tt

2
2
=-12(
'lt E
Ct)
l-.V2) b k

605~

219 ksi

-- 2420.8

for

b= 4

'
1.

ksi (fort

n
1t

f1'

material.)
material)

For these calculations, it is evident that assuming

the yield stress of 55 ksi for each case was in close agree
ment with the critical stresses for buckling.

Although the stress of the top chord member was not

critical in the matrix analysis given in Cb.apter II, it is

necessary to check tbat the 8

xi

plate does not have a critical

stress substanially belovr the yield Yalue. Since the behaviour

is purely elastic up to 48 lrni , a simple elastic calculation
is given belovr.

When rr'= l , Eqn. ( 3. 9) gives

44
2

=~

<1.
c

x 29600 ( 1~ 2 x 5.79

12(1-0.09) 32

15l.3 ksi

Since this value exceeds 48 ksi ( pro:r:ortional limit),

it is ap:f)arent that the buckling stress for the inelastic case

will be between 48 ksi and 55 ksi. Hence the top chord for
truss B is not critical.
of the

.,

Truss~

thick material

A need not. be checked because

45

....

,..

..
r

;(

------

--........ ---

.dx

-;:::

J
ng.

3.'l

A Rectangular Plate Subjected to

Compressive Axial Stress:

46

if]}

111~'.
i
~

4-s

+.o

0.2

o.~

Fig. 3~ 2

06

a-s

/.o

12.

14-

/. 6

~l
lB

2-t>

2':2

I
2~

:Plate Coe:ffici~nt K as Function of d/b

2s

47

Table 3.1
&"c ksi.

Estimation of Bu.ekJ.ing

ff

S~ss

..I

~ f f ksi..~'

48.0

l~O

l~O

48~0

49.'0

0~875000

0.93600

52~;5

54;~0

0;~160000

0~';40000

135.0

54~21

0.127450

o. 35700

151~8

54~93

0~011440

0~ 10700

513~;36

54.945

o~bos990

0.09490

578~977

54~950

0~:008170

0.09_060

606~;5_12 ~

54.970

0.004900

0.06300

782.540

54~:990

0.001636

0~04000

lY/4~750

54.995

o.noos1a

0~:02860

1922~900

54.9969

0~000507

0.02258

2435. 600

~(From above Table, the buckling stresses for the

web members of -i!' and iutlti.cknesa material are
54.;95 ksi. and 54.9969 ksi. respectively.)

CH.APTER IV

STRUCTtEAL BSEA VIOVR OF UNEQUAL-7lTDTil

com:rECTIOITS

4 .1

ELASTIC BEEAVTOu"'R

The theoretical and experimental investigation o:f the

elastic behaviour of the unequal-width connections had been
carried out by Redwood (12 J ( 13) .
As mentioned in Chapter I, when the width ratio d/b
is less than 1.0 , tne joint problem is one of plate flexure

with certain loads and boundary conditions.

In addition to the

localised bending just described, the chord member is subjected

to bending and axial load from the rest of tbe truss.

However,

a length of plate about four times the width is considered to

be sufficiently 'longr that end-effects might be ignored.

Therefore, in the \York described in tl1is volume, particular
attention is devoted to the localised bending of the connecting
flange of the c"hord me:nber.
Theoretical and experimental

valt~es

of joint stiffness

obtain by Redwood are plotted in Fig. (4.1) where D and

deined earlier for applied moment.

were

The detail of the joint

used iu the tests is given in Table 4.1

In Fig. ( 4.1), the stifi:'c.ess irwor:porated L:-j_ appropriate

non di::1ensional form is

joint

pa:ra:neter

7'.. (

sl~own

to inc:rease rapidly with the

7'.:== d/b)

At Telativel:/ low loads, the load-displace;rr1ent cul"'\1es

48

4-9

become non-linear. But in most cases, there is a large reserve

of strength rrhich has not yet been fully utilised. Iherefore,
1

the elastic analysis expressed in Fig.(4.1) can give little

in.fornation on the strength of the joint.
Since the end web members are critical for the examfles
used(see Section 2.5), a plastic analysis of the joint is atte
mpted to estimate its collapse load to determine whether stiff
ening ;rocedures are required to establish full moment capacity.

4.2

TEE STF.ENGTH OF JOINTS ESTTIJ,,ATED BY PLASTIC METEOD

Use is made of Johansen's square yield criterion (14}

for tbe Plastic analysis of the joint. Although the square

yield criterion is mostly applied to reinforced concrete stru

ctures, it has been shown by Wood[l4) that the method aPflied

fairly well for steel plates.

A. collapse mechanism is assU!Iled on the loaded plate.
By using the principle of virtual work, the collapse load can

be calcuJ.ated. As the collapse load estimated by this method

is an upper-bound value, the lowest value of all calculated

collapse loads cor:resJ;<>nding to various assmn.ed mechanisms is

the most

cor:rect one.

JOIFT SUBJ3CTED TO APPLIED rmllirE'NT

Initially, the assumed collapse nech2Ji..ism for the

pl:?..te subjected to :noment transmi ttecl f:rom the weo nember is
s.c.o:r:.1. in Fig. ( 4~ 2) in '-'Illich the dotted lines are tt:.e 7ield lines

50

The region of the Plate enclosed by the web member is assumed

to be rigid as it rotates through, say, a unit angle.

The

amount of rotation of plate elements about the yield lines and

their lengths are tabulated in Table 4.2 note that the values
on one side of the center line are numerically equal to the
other side.
The equation of virtual work associated with the mecban

ism is
Ml

2
=2bm.p (1-1\+~+
1-1'.

2 "-tan~

1-7'

+ 21'-cot~)

(4.1)

for a unit angle rotation o:f the square enclosrz:d by the web
member, and where mp is the internal moment per unit length
along the yield line. Using the condition d?!/dd...
tanG(

=Ji -

=O,

we get

7\..

Substituting into Eqn.(4.1) , we obtain the minimun

value of M
M

2
2 + 4
= 2bm p (1-J\+ 1-"
Jl-x.
X

A... )

Lower values of the collapse moment M were attempted

by employing other patterns of mechanism with

fans ' of yield

lines as shown in Fig.(4.3)

The equation of virtual work works out to be

2*

cosie

+ 2~

(1-~) cosd..

1-7\.

51
The minimum value of M is obtained by setting oM/'iJti..
and

oMhJ

equal to zero

In the range

0<7\.-< 0.635
.

Eqn~;(4~2)
.

gives

a smaller

value of M 'tbail that given by Eqn.(4.4)

Equations (4.2) and

showing values of

11
0 5 bmp

(4~4)

are plotted on Fig.(4.4)

with various values

of~ ( note tl:!a.t

~~
is equal to four times the value in the parenthesis of
0 .::>omp

Eqn.(4.2) or (4.4). )

COMPARISON BETWEEN THEORETICAL AND EX1'ERIMENTAL

co1t.~s:g

'LOADS

The experimental results from the tests perfonned in

the University of Bristol (12)

(13)

plotted in Fig.(4.5) is

used to compare with the theoretical results derived above.

Fillet welded connections were made between ixlo square HSS
and were subjected to applied moment. The specimens were of

Grade 16 steel to BS 15 : 1961. The average yield stress was

18.5 ton/in.2 which was used to compute the mp of the specimens.

In using the theoretical equations for calculating

the collapse loads, the width b was taken as the outside dimen
sions of the section minus twice the thickness of the wall.
And also, two values of i\ fQr each tested joint were taken;
one corresponds to the nominal size of the we.b member, the
other is the nominal size plus weld material.

Details of the connections are summarised in Table 4.1

52
In. Fig.(4.5), the estimated collapse loads show some
agreement with the measured loads especially for the smaller
v1eb

member.
\T~en

the web member becomes larger, the estimated

collapse load based on the nominal web size becomes increasingJ.y

conservative. Onthe other hand, the larger load, based on the

nominal size plus weld, overestimates the strength of the joint:
The

latte~

discrepancy is probably as a result of the unconnected

web plates of chord member failing before the connected flange

plate for large sized web members.

For the trusses A and B( see Chapter II) where the c:ri ti cal
sections are the ends of the web members, the results of the
yield line method and experimental results quoted herein are
im};X)rta.nt.

For the truss A, the moment capacity of the 4 "x 4 "x.i-"

web member is 508.75* in-kips, and the moment capacity by yield

line theory is 240 in-ldps. For the truss B, the moment cap:ici ty
for the 4".x 4 "xi" member is 290~'46 in-ki:ps, and the moment

capacity by yield line method is 60.0 in-kips.

From these figures,

we know that the strength capacity of the web member is not fully
utilized~

Furthenn.ore, the experimental results of Redwood (12) [13)

show tbat fox small values of J\ (width ratio) , the joint capacity
is less than the web member capacity.
Therefore, on these two counts, the joints require
stiffe;.U.ng if the panel roint loadings as eornrn1ted in Cl1apter II

are to be realized.

Based on steel that yields at 55 ksi.

53
An extension of the yield line method will be used to

study the strength of reinforced joints in Chapter VI.

54
Table 4~1
Main. Member

Detail lli.mensions of Joints (taJJ~J

Weld

Branch Member

.l.
1
'
5 x 5 x3/16
22X22x0 .16

5 x 5 x3/16
l .x 1 x

5 x 5 x3116 .. ::-~~~~.160 ...

'

~:677

0~'217

0~325

I6

0.412

0.'546

-2..
16

0~60_5_

0~739

o.-1sa

0.920

..

0~'541

5/16

O~iJ.28

A. _{_dlhl_
(d/b)
Exel.'~ Welc ~ Incl. WelJi
:A.

Size(itt-)

7''

5 x"5 x3/16

~x21-~.l6

5 x 5 X3ll6

32X3txQ .19 2

l.

'1

:,

I')

~Finite

lOL.

difference

~!ethod

10
o.\

.a.

4.1

o;~ -.4

o.s o.~

0.1 Q.\ , r.o

7'

Join.ts with applied Moment *

+(Reference is made to [1~(13])

55
b

B ;-----------------~ c

r,

,'1

'

\,

tf

\
I

'

e'

F' I /

't

-------~:/_l-;_:-~ 1 -----~ ~
A

I '..
I

'.,..

I
I

,/'

tJ''
F \J

:l

....i,.

:
I

',
,...,.,_-. ............---"'

/E

I
I

,'

I
I
I

'

o(.'

I
,'

.v

,,.' '

I
\

,,

;/
\I
,
B -----------------~c

Fig.

4~

A Yield-line

Pat~ern.

for A:Pplied Moment

+
Fig. 't~3

Alternative Yield 1?atte:rn

for Applied i:ioment

56

Lengths and Angles of Rotation of" Yield Lines

Table 4.2

,,
-

.AE,DF

7\. <:o.s t!(.

\ - "

"'t-

J\.b
:>ti. s'1nci. ~
I A.

<\- 7'.)
2

ssnoe.

57

100

-6.o

90

M:

0.5bmp 80

70

~11~ b

60

50

40

30

20

10

.1 ;2
F1.g; 4.4

:3

7'.Width Ratio

.4 .5 .6 .7 .8 .9 1.0

Theoretical Applied Moment :for Collapse

58

r.r

~eb member
M

26

,......
.~
,

~2-~ ..6__ ton-in

t'-'

22

Ill

E
0

l:"

20

Theoretical Estimation Based

on nominal size

....... _.,. Theoretical Estimation Based

on nomial size plus fillet weld

18

-u

.!

-a

c
'<(

All square HSS

5x5..x 3/16 ..

2'4

:c:

Moment Ca.paci ty

we.b Member

16

"

3ix3ix.192
.. ,.
..

14

2t-x2~x0.16

12

ff

2~~n>.16

l@

6
4
2

1.0
2.0
3.0
Relative Rotation(ra.d.Xl0- 2),sb

Fig. 4.5
Applied Moment Versus Relative Rotation
(Experimental Results from University

of B:ristol (12) [13) )

29.82 ton-in

.,

7" 7'}
..
13xi
.x0
.16
. .,8
,.
ly.J..xQ .128

58.TI ton-in

24.35 ton-in
13.08 ton-in

2.72 ton-in

59
CF~PrER

THE

n~FLUENCE

OF SOME MEMBER SIZING ON

THE BEHAVIOtJR OF JOINTS

5.1

GENERAL INTROTIUCTION
This chapter presents a brief review of the dimensional

parameters of the connecting members which are important for

fU.11 web to chord member moment transfer.

When certain ratio

limits are achieved, no additional strengthening is necessary

to obtain full

~orki.ng

capacity of the truss.

Chapter VI will

discuss strengthening methods and will attempt to analyze

joints for moment capacity.
It was nentioned ee..rlier that, when the welded
com1.ections betvrnen two unequal-width square ESS at right angles
to one another were formed the problem of the joint was nainly

the bending of the connected plate, and that the strength and
stiffness of the joint were thus greatly reduced.
There might be some improvements foT the joint if the

dimensions of the connected plate were carefully selected to

match the given dimensions of the b:ranch member (i.e. web

member).
The experimental results by Redwood furnished in the
preceding

chapter were all with cor;.sta:::it dimer1sio::is of the

main :nembe:rG

These tests, therefo:ce, have lL.ni ted value in

60

terms of relating slender;.'1ess or thicimess ratios to moment

capacity.

Based on the limited tests carried out at the Drexel

Institute of Technology

[15),

some information in this aspect

can be gathered.

5. 2

THE I'NFLUENCE OF CHORD WIDTE~to-THICKNESS RATIO, b/t 0

The details of the tested joints at Drexel with various

b/t 0 ratio are tabulated in Table (5.1), and the experimental

curves are plotted in Fig. (5.1).

In all these tests, the widths of the two members were

kept constant; one was 4", the other 6tt.
branch member was also kept unchanged.

The thickness of the

The only change was

made to the thickness of the main member to form b/tc ratios

of 12, 16 and 24.

(t 0 is the tbickness of the main member.)

The joints with these ratios are labeled as Jl, J2 and J3

respectively
..~s shown in Fig. ( 5.1 ) , Jl and J2 with the b/t 0 ratio

equal to 12 and 16 v1ere capable of developing the :fully plastic

moment of the branch member even though d/b is less tban 1.0

J3 with b/t 0 equal to 24 was unable to develop Mp or even My,

the yield moment of the branch.
It is conclu:led that, backed with ooservations from

seve:r:al miscellaneous tests, the unequal-width connections

c:~n
considerably-

be st:rengtilened
to develop H..,;J. or even Mu.;,; if
-

the width-to-tb; cmess ratio, b/tc is 16 or less.

61

T-r<f[i:

.>-W

ITT
"PT
C"P
'-' ..u TTf0'<:
l...:.:..:i.:.,
..w

010
t c /+'-'vr
-

In the preceding Section,

attention was :paid to the

fact that the joint with a b/tc ratio eq_ual to or less tl:an 16

is able to develop M,7 or even Mp

.)

Close examination of Table (5.1) shows that another

important parameter affecting the joint strength is the t 0 /tw
ratio.

Where

~v

is the thickness of the web member.

From Table (5.1), the tc/"tw ratios for the above

mentioned two 'strong' joints, Jl and J2, wmch were able to
develop Mp , are greater than 1.0 , whereas tc/tw ratio for
the 'v1eak' joint, J3, is equal to 1.0
Introduce another joint J4, with joint dimensions
matching other joints as shown in Table (5.1) , to note the
influence of t 0 /tn .
The experimental curves from the Drexel tests are plotted,.
in Fig. ( 5 2)

When a comparisoL is made, regardless of all the

:para.:!leters except b/t 0

between J2 and J4, we might quickly

claim that J2 and J4, should have the same joint resistance
because they bave the same b/tc ratio.

But, as a matter of

fact, J4 is stronger than J2 even though the former has a

smaller 7\ (as joint

stiffr~ess

is pro1:ortional to J\).

The Teason fOT this is claimed to be that J4 1::as a

highe:r tc/tw ratio than J2

62

It is therefore concluded tbat the ratio

tc/~v

is

also an important parameter for strengthening the unequalwidth coru:.ections;

the joint.

the higher the

tc/~N

ratio, the stronger

The unequal-width connections can be assured to

develop Mp if

tc/~,~

2.0 and at the same time b/tcE="l6.

Clearly, the examples of the two trusses mentioned. in Chapter II

do not satisfy the first inequality and, in addition, truss B
violates the second inequality.

Web Member

a"

Branch Member

e~~-~-~-~-~-~-~-=-""--.--------------''t

.~~~--~-~-~--~-~--~-~------------~-~-----~

; -

- - -

- -

- -

ll

..

- -

Table 5 .:1
IJOJ..nt
..

No

Jl

Main
Hembe r
,,..
0

i_cl

.- -

---- --- -

_.!._

11

Joint Details

--

/\. _d

Branch
?JemhA~

UYt)
7

-b

x 6 x 0.5
4 x 4 x 0.25
0.667

b/t0

cltvr

12

2.0

J 2

6 x 6 y!J.375
4 x 4 x 0.25

0~'667

16

1.5

J 3

6 x 6 .x 0.25
4 x 4 x 0.25

0~667

24

1.0

J4

0.5

16

2.0

x 8 x 0.5
4 x4 x

0~'25

_; ..:!'..

Ci

1b1:r~ ~o______..---,.~ :.,,

_.!

3->''

we

./

/
; ./@

-@

.------------

/t

:T1

b c

(12),

72

(16)

w'/---
-o

we.

..24

\6

6.
\-i,g. 5i

DeS \ec.:tio n

Load - detlec.l'ion.. Curves

for

J3 (24-)

C)

WC

.2.

WELD

CRl'iCK})

('1n.)
b/tc.

E5fed:

( Drexe\

Re.s.C.:tlr-ch )

'

-f:l.

l 'O
~~rf

+>

...,

..............
~
'-J

l!)
""'
"-'

65

fu'.lL

I""

cJ

'6

ct"

'U)'

..::.i

:1Iii

r;

[11
ct.

'--"

Gt

l1J

.fJ

<1

..

CHAPTER

vr

JODTTS WITH REilIFORCEME:NTS

6.1

GENEBAL nrTRODUCTI01T

When the ratio b/tc is greater tban--16 and-tcftw-is

less than 2.0 , the unequal-width joint is generally unable to

develop the fully plastic moment of the smaller section (web

member) unless

reinforce~ent

is applied to the joint.

Several kinds of reinforcement can be suggested such as

the use of web. flange plates, chord flange stiffeners(mounted on
the ct.ord flange), haunched plates etc. as shown in Fig. (6.1).

Nevertheless, these forms of reinforcement should be consistent

with the requirements of efficiency, feasibility and economy
as a whole.

The following sections are devoted to various forms

of reinforcement.
6.2.1

ANALYSIS OF FLANGE-PLATE REIJ:1FDRCED JOINT

An

appro:d.mate analysis of the F1.8.J.""lge Plate Reinforce

ment is discussed in tbis section.

Co-:.:.sid.er an unequal-width connection :reinforced

1Jy

two

flat plates welded on the flanges of the branch oember as

snown
"

1.ll

"rL!
J.i'...!..g.

....
1)
\,. b.
a

the branch member.

A nonent is applied or:. the joint through

67

The relative rotation of the joint as a result of the

applied moment can be found by dividing the maxiBlli21 vertical
deflection of the chord flange plate by one half the depth of
the branch member.

The vertical deflection of the joint consists of two

parts; one part is attributed to the shortening (or elongation
at tension side) of the reinforcing plate, the other part is

the deflection of the connected plate of the main member subject

ed to line load transm:i.tted from Teinforcement.

The former

part is small as compared to the latter part.

The exact distrt bution of line 102:.d. transmitted from
reinforcement is not fully known.
sane rea.sonable

a:ppro~dmation,

But, it is quite safe, with

to say that the distribution of

loading falls somewhere between two extreme cases; one is the

uniform distribution, the other is of parabolic shape with
zero intensity at the center.

However, the areas of tv10 dis

tributions should be equal to each other.

On the compression side of the joint subjected to

applied moment, the reinforcing plate of dimensions b

assTh~ed

h x t is

to be subjected to concentrated load of magnitude P

as shown in Fig. ( 6. 2a).

ir1 tl1e

Since all motion is supposed to occur

plane of the :pa1)er, a unit thic.l.G.:. ess of the plate is

taken in the discussion.

68

This ty1;e of plane stress problem is best sol-r..recl b;y

the lower bound theoie;n of lil'li t ar:alysi s ( 16

J (17 Jwhich

states that i.f atJ. eq_uili bri um state of st:ress below yield. can

be found which satisfies the stress boundaI-y conditions, then

the loads imposed can be carried without collapse by a sta"ble

body composed

elastic-perectly plastic material.

Instead of the plate, we suppose that a triangular

:Pinned truss is imagined to carry the load inside the body as

shown in Fig. ( 6. 2b).

The forces in the members of the truss

can easily be determined as

c
where

=2

p
cos 0'

'

=- C sin e'

is half the angle made by two

=Ltan
0"
2

inc~ned

( 6.1)

legs.

The cross-sectional area of each member is to be chosen

in such a manner as to give a safe or pennissi ble axial- stress.
Here the stress must be chosen at or below 2k, where k is the
yield stress in shear, if a lower bound on the limit load is
to be found or if the safety of applying P to the plate is to
be realized.

To maximize the lower bound loading, all members of

the truss are taken at yield.

The width of each member and

the J;Tincipa.l stress in the overlap regions are all indicated

in Fig. ( 6.~c)

The compressive force in the inclined legs and the

lower bound :for the limit value of P can be written as

69

= 2k 7'.b cos 0'

'

P :

2C cos f5 ::: 4k 7\ b co s 2 0-"

The vertical projection of the shortening of the

inclined bar is equal to the change in the height of the truss.

For simpllci ty, we assume the change in the height of the truss

is app:roxLmately equa1 to the shortening of the

plate~

Therefore, by using Hooke's law, the shortening of the

bar is

where

C/A

=2k

= b(l-..7\.)/2

sin f1

The change in the height of the truss or the shortening

of the plate is

=kb(l-P.)
E

h
b( 1-A.) /2

--~
2kh

( 6.2)

J:.,

As can be seen from the above expression, the shortening

of the plate is proportional to its length, h.

As uill be seen

later, tbis shortening is small as compared to the deflection

of the cori.J.1ected plate of the chord mernber and can thus be
neglected.
DEFLECTION OF THE CONNECTED

PL..~TE

OF .MAIN MEMBER

Two extrerae c2ses of loading distributions on the plate

70

will be. considered. One is the uniform line load, the other
is parabolic line load with zero intensity at center.

The long edges of the connected plate of the main

member are assumed to remain straight and are subjected to

restraining moments from the web plates. The effect of these

restraining moments is disregarded for the purJ.X>se of counter

balancing the assumption that all plates are rigid vrhich in

reality ba.ve some flexibility.

A length of four times the width of the plate could

be regarded as

sufl.~ciently

long to simulate the chord flange

aamentioned in section 4.1.

So the length of the long edge

is assumed to be of this a.mount.

And thus the loading conditions

at both ends of the plate of this length could be neglected,

and so the deflection is- zero.

The middle of this length is the center of rotation of

the joint where deflection is zero.

This middle line is anti

symmetrical in relation to the forces.

Taldng

one half of the

length of tris long plate into conside:Tation, a rectangular

plate of dimensions bx2b with four edges simply suprorted,
subjected to either u.-riiform
.
line load or rarabolic line load

is the main topic of discussion that follows.

The details of the rectangular .Plate are shmm in Fig.

{6. 3)
6. 2.4

SI:1IFLY SUPPORTED

F~CTANGUT,AR PLATE St.JBJECTED TO UNIFOPJ:!

LINE LOAD
The :pro blen of a rectang-ulo.:c :t:late with all edges

71

simply supported subjected to transverse loading can be best

solved by Navier' s methodf1sJ.
Navier' s method for plate :problems is essentially

explained as follows.
Consider a simply-supported rectangular plate of di
mensions AxB with coordinate axes as shown in Fig. (6.4).

The governing equation [19] is

( 6. 3)
where

intensity of transverse loading, w and D having been

q~

defined earlier.

The boundarJ conditions for all sides simply supported

are

'

w]Y =0= o

=B

;,2w
0~2

=0

= o

x::: A

(6~4)

'

Suppose a deflection, which satifi.es the B.C. is equal

to

W - C11 s-ln zrA4. sin

'j

..i..

\Y

( 6.5)

Substituting the above equation into the governing

equation
(6.5) is

(6~3),

the loading needed to produce tbis deflection

72

nc1

q -

2~4 + ~41
sin A

+ 2 2
AB

. -

7{:X

B4

~
sin

( 6.6)

Similarly, suppose a deflection which satisfies all

B. c. is

= 0mn sin
.. -p;mx sin
. n~

(6.7)
t~is

The loading needed to produce

deflection (6.7)

must be

_
[m4z.t.4 2m4n4~
n4z.t::4]
q - Damn --:t t
2:82 t 4
A

ml{x

n""Jl.Y

sin-sin

(6~'8)

It is seen that a transverse loading of the form

sin -r
7Z.x sin ~l

applied to an all-sided simply sup:ported plate

will produce a deflection which is also of the fonn sinlx sin~1

Hence, a procedure for solving tbis type of plate
problem can be summarised as follows;

(i) Express the transverse loading into trigonometric

series as

qmn' sin

(ii)

m~x
-Ab.

sin

n~;z

The deflection of the plate will then be

Vf=

..

cmn sin

m =1,2,3 n ==1,2,3

m~x

nzz-;..

sin -~

73

where

Turning to the case of a simply-supported plate of

dimensions bX2b , subjected to unifo:rm line load of intensity

% at

J,

:from one edge i s [ Appendix IV

. -" ' L
-<.:><)

- 4q 0 /

q - z.<b :-::.:

00

. n'f!.f n\fx sinm"y,

SJ.~ SJ..ll2 b
m
b
1

..:;,.

(6.9)

n~1,2,3 TIF-l,3,5

-mn

where

4 qo l

nl#:x
m~y
sin-si~
2b

ntt;f'

q
:::a "11b Sl..ll -
inn
'""'
m
2b

Following the procedure described above, the deflection

of the plate can quickly be ound as

(6.10)

where

mn

::. q

mn

[m2

n2J

/D~ -
b2
4b2

74

SIMPLY SUPPORTED RECTA.MGULAR :PLATE SURJEOJ:ED TO

J?ARABOLIC LINE LOAD

Next, consider the simply supported plate of dimensions

bx2b subjected to a parabolic line load

maximum intensity

o~

3qd at both ends and zero intensity at the centre acting at

distance from one edge.

a

The quantity 3q0 is determined in such

way that the area of parabolic line load is equal to that of

the uri.i fonn line load.

The equation [Appendix III] for this parabolic line

load can be exp:ressed as
oO

L
00

q =:;3qboL.
~

m "i' m

m= 1, 3, 5 n= 1, 2, 3.

fl :

~
~

m=1~3,5

where

~n

n~f
. mTV n~x
2 b si~ sin 2 b

1.1 __ H2-, sin:

-\-;--~
~

sin m?;;z sin n"{l;x

b
2b

"'inn

(6~'11)

n=1,2,3

48qo 1 ( 1
2
=~m
4=ZJ.:b
4~0 l_ (
24
m
3 - m2~2

).
)

. n'J.!;f'
sin2b
..

sin

nl!:f

Tb

Following Navie~'s method, the deflection of the

plate can be found as

l~
.1

00

co

==\
~

\_
~

..

m'ttx

..

mn sin -i;- sin

m=l, 3, 5 n=l, 2, 3

n:{{:y
2b

75
where

Substituting

r=

(2-0.~57'\)b, X::: (2-0.'57'.)b and y=o~'5b

into Eqns. (6.10) and (6.12), we can find the deflections -of
the plate at the middle of two different line loads.

The

actual deflection of the plate falls between these two values,

i.e.

act.

<

( 6 .10)

Three terms of either Eqn. (6.10) or Eqn. (6.12) are

sufficient for an approximate value of deflection.
the above mentioned values off, x, y and q 0

By using
M/db = M/J\.b 2

and different?\., values of deflection from Eqns. (6.10) and

(6.12) are obtained. The obtained values are then changed

into relative rotations of the joint by dividing them by 0.5d
or 0.5.i\.b.

The contribution of deflection or relative rotation

made bJ Eqn. (6.2) is small and is invariably

Eqn. (6.12).

small, the

85~

v1i thin

8% of

As the deflection by Eqn. (6.12) is already

value could be neglected.

A graph with relative

rotation <P versus A. is plotted on Fig.(6.5) to show the fle

xi bili ty of the reinforcel joint with respect to 1\..

76

6. 3

CHORD FLANGE STIFFEFER

A stiffening plate mounted to tae chord flange with

its centre coincident with the centre of the joint is shown
in Fig. ( 6.lb)
It has been recognised that the loading transmitted
from the web member to the joint is mostly resisted by the
connecting flange plate when 7\.is less tban 1.0

An additional

plate above the connected plate will automatically tend to in

crease the flexural rtgidi ty of the joint.

This increase will

be by a factor of eight if tbe thickness of the stiffener is

the same as tbat of the chord flange.

A test of tbis type of reinforcement had been carried

out in the Drexel Institute of Technology (15)

The details

of the joints and the test results are shown in Table (6.2)
and

Fig. (6.6) respectively.

In the test, the stiffening plate of dimensions 5x5x5/16

was fillet welded to the 6n face of the main member of

dimensions 6x4xl/4, the brax.. ch member was then centrally fillet

welded to the stiffener.
As can be seen frora Fig.(6.6), the monent resisted by
the joint with reinforcement is
joint without reinforcement.

175~1~

that resisted by the same

Therefore, the chord flange stif'.fen....

er ma;:r provide efficient :reL1forcemex1t for unequal-wid..th connect

ions. In addition, the adding.. of tl1i s stiffener is not object
ionable from an aesthetic

vie~:>ipoint.

77
6.4

HAUtTCEE.D REINFORCEHEITT

The haunch consists of HSS wedges cut at 45.with a

right angle to match that of the web and chord member.
The details of the haunches for reinforcement are

shown in Fig.(6.7).
To

the knowledge of the author, the experimental tests

on this type of reinforcement for unequal-width connection

has not yet been reported.

England

The tests carried out in Corby,

(20) , on this type of :reinforcement were solely for

equal-width connections.
I

In order to obtain design information for haunched

reinforcements for unequal-width connections, an experimental

programn is currently being carried out in

I~cMaster

University

under the supervision of Dr. R.r:I. Korol with the sponsorship

of CIDECT.
6. 5

PLASTIC MECHANISM FOR REI1JFORCED JOINTS

As an extension of yield-line theory described in

Cb.apter 4, an analysis will be made of reinforced joints

loaded to failure.
FL.AF :}.E PLATE

REI1~FORCE:.f.E:CJT

The details of the flance-I;late reinforced joint are

shown in Fig. ( 6.la) in which tvro rectangular or trapezoidal

.fle.t :91ates are welded to the flanges of the branch member.

78
The collapse yield line pattern for t:'ois reinforced joint sub
jected to applied moment is similar to that of the unreinforced
one (Fig. 4. 2)
Like Eqn. (4.1), the internal

vir~ual

work done by the

yield lines when the joint rotates through a unit angle of

rotation is
.
2A2 . 27\ tanc( .
t.J
2bmp ( 1 - 7\ + 1-A.. + 1-.i'\.
_,. 2A. co <Jo..

Considering

th~

( 6.13)

reinforcing plate {Fig.6.8) on the

tension side of the joint, the shearing stress does the work
of lifting two triangular parts of the plate when the joint

rotates through unit angle of rotation. As the plate is assumed

of rigid. plastic material, the lifting only occurs when the
shearing stress reaches its ultimate value as shown in Fig.(6.9).
Therefore, the total work done by the shearing stress
in two reinforcing plates is

(6.14)

and

'f =~
2

d/2
- /\b/2
-(- b-d)
12 ( b-d) 12

are constants.

Eqn.(6.14) becomes
4 <:5y A.b/2 tJdA
2 ( b-d) /2
: Ad~ 7\.b/2 t (1=.
. 2 (b-d)/2
2

Jc:.d. h)
2

( 6.15)

79

Equating the internal work to external work, we get

~r = 2bmp

2
( 1 -7\. + 21\.. +- 2 1\.. tanc(-t- 27\. cotc:L)
1-1\.
1- 'J'\..

With the condition dM/dct:: O , we get tanc(=

J1

idy.i\bth

- 'A

Thus we obtain the minimum M as

M. ::: 2bm p ( l - J\.

~ + ~
2J\.2 +,; fi~~
4?\. )

" h
+' 1~y?\.Ot

( 6.16)

It is obvious that the last term of the above equation

is due to the reini'orcing plates which add more strength to the

joint
CHORD FLANGE STIFFENER

A chord flange stiffener is a rectangular (usually

square) plate of dimensions b x 1 x t welded between the chord

flange and the web member with its center coincident with the
center of the web member as shown in Fig. ( 6.lb)
It is well known that if the fully plastic moment of
a plate of thickness t is mp per unit length, a plate of

thickness 2t should have monent capacity equal to 4mp.

However, complete composite action is unlikely, hence it is
useful to also consider as a limit the case of no bond between
reinforcing plate and chord flange.
la~1 er

be

Plate

"bet~.~;een

li~1-S

For t:b.is case, the tvm

a moment capacity of 2mp

these two extremes.

Our actual case v1ill

80

The yield line pattern for this type of reinforced

joint is essentially the same as that of the unreinforced joint

with only a slight difference. The diagonal yield lines may

propagate beyond the edges of the stiffening plate.

Whether

or not the diagonal yield lines spread beyond the edges of the
s_tiffener is der;endent on the length of the stiffener.

In view

of tbis phenomenon, the yield-line fatterns are classified

into the followil:J:g th:ree cases.

( i)

EXTEITSIVE YIELD LINES SffiEAD

The collapse mechanism with diagonal :ield lines spread

beJ.rond the

edges of the stiffening plate is shown in F.ig.(6.10).

The yield lines occuriug on a single-layered Plate are acted

upon. b:/ mp' such as BB', CC', BC, BC' , B' G' etc The
others occuring on double-layered plates have a moment ca:Pacity
of 2n-r. or 4m...... The angle of yield-line rotations and their
~

.l::'

lengths are the same as in Table 4.2

The eq_uation of virtual work for the 2mp case is
M

= 2bm:-!:'

2
[ 2 7'. tano(, + 7'... tanc(, + 27'..cotc(
1 -:Av
1 - 7\..
l

7\.1 ( 1 + tan2d't) _
' b
1- 7"

d..

+ 2 A? + 2(1-J\)
1-J\...

7'..2

To find the nir.imum M, we let dM/dot1

tan~,

l-7'.

=0

]
( 6.'17)

which yields

81

27\.

- sec O{I 1"1-J\...

~ sec d..
1-/\..
I

+
27'.

or

2A.l tan, .
b( 1-7\)

l-7'.

")1.2
"'"

1-}\. .

27\.
-

sec~,

- 27\.cotei,
2 T'

co sec~",
~

sec~ =O

- 27'.2 tanoc'.,.
1-7\.

2:hl
b( l-?\.)

J
tami -2-J2.
- tanQ.,
t

--

1-J\.

On fu.-""1;her simplifying, the above equation becomes

1 =0

tanU\ - ( 2 +"-)
" ) tan~, - (1-7\)(~ )
2 ( ~b-1

( 6~'18)

Equation (6.18) can also be written as

tan1: = ( __lL )
~b-1

(<. 2-t-~
tan~ ~
2
l

( 1-.A)

For the case being studied a square reinforcing plate

is assumed such tbat l

=b

Also../\.= 0.5 since the web member

is 4 tt and the cho:rd member is 8 " The solution for tanO(,

from the above equation is tanc<.,:: 0.57 and hence

cf;=

2941'

With 1 == b and .A..= 0.5 , Eqn.(6.17) becomes

Substituting
., ..I-

tanc( 1

= o.57

into above equation, we

001.12.ll1

2bm (

6. 33)

In siwilar v1ay, the minimum 11!

( 6.18)
case is 2bm:p(lO.l)

82

( iL)

DIAGOIJAL YIELD LINES THROUGH THE CORNERS OF STIF EErJER

PL.ATE

The yield-line pattern for this case is shown in

Fig.(6.11) There are only four yield lines, namely
BC ,

:s' C' ,

BB

'

and CC ' having moment cai:e.ci ty mp' the others

have 2mn or the no bond state

..i.;

The equation of virtual work is

]if' :::: 2bm

2
[ 3 .7\. tanc<i.+ 37\. cot + 3J\.. + 2(1-J\.)]
1- J\.

.;r.

1-A..

Using dM/dol..l.: O, we get tan~,_ :: Jl-7'

Substituting tanc<~ =Jl-1\.., , we obtain the min~' M as

Tur f\.= 0.5 as our previous example of truss A or B7

the min. M becomes

(iii.)

CONFilIBD YIELD LINES

The collapse yield-line pattern confined to the region

of the stiffener is shown in Fig.'( 6 .12)

As can be seen from

the Figure, only BB' and CC

bave moment ca:paci ty of m:P' the

other haYe moment capacity 2mp

The equation of virtual work is

83

By

The

using dll/do<..3 =-o, we get

min~'

For J'\.::

tano( 3 : Ji(l-'J\.)

tr is thus

o _:5 ,

The 'fan' yield-line mechanism for above three cases

vtas not attempted because of its comlexi ty.

In addition, the

adjustment to M would be slight and therefore the attemft is

not justified.
Table 6.3 compares joint moment capacities with web
me:nber values for the two cases stud.ied.

Comparing the joint

moment capacities obtained for the flange plate reinforcement

and the chord flange stiffer.er we observe tbat flan.:.;e :Plate

reinforcement provides a value of moment oapaci ty intermediate

bet~een

the

extre~es

of full chord plate action and

co~plete

sliding action.* No definitive statement can be made as to

vfrJ..ch method of reinforcew.ent is supe:cior.
t~1at

as

A test series such

currently underway in the A. D.L. will hopefully answer

that auestion..

?o:r the case of truss A, the :reinforce:::ient

methods a11pea:::... to be si;_ffi cient to develo .P the required Hp for

I)

the rm b ;ne:.nber.

This is not the case for truss

the 1/4

plates are tnsufficient to develop the Tequired \Vea l'1eaber

~nonent.

*-----------------------
.
Truss A only
-r.1

~o:r

84
rr
Fi.g~

l]

6.la Flange Plate

Reinforcement

I
I

I*

-- -- 1l- -

-- - Flan~

Plate Reinforcement

Plan View

I
I

- ---

- - - - - - ....,_

Side View

Fig' 6.lb Chord Fla.D.oo-e

Stiffener

---------- ---Chord Flange Stiifener

J?lan View

Side View

Fig.6.lc Haunched
Reinforcement

,"
;

t
I

[0[]
- - - - - - - -Hua...~ched

Reinforcement

Plan.View

Side View

85

Fig. 6.2a Rein.forcing Plate Subjected

To Modified Concentrated JJOad
p

r--I

+ ----1
I

c,..___,c

I
T

- - --'- - - - - - -t ~I

P/i.

Pf2..

I<
Fig. 6. 2b Triangular Pinned Truss

in Plate

86

I
7\ \> cos & sin&

nb ~

+-7\bsin29'

2k~ cos
.

....

7\b + ?\.bsin2e-

P/2

f /( l+
0

...-.......:::----.1

P/2
Triangular Pinned Truss

sin219')

87

Reinforced JIQint Under .Applie'd>:T\ioment

~-~_-_-_-_-2=b==~~-4-b~:>_,/-/"!IE''<_-_-_-_-2~b=====:;-,..~/
Tsolation of Loaded

/I
;(_<

~late

/Ul(/

(.2./,-1:)
2b

>;4 P.-j
I

88

.P;.;_

Simply-supported :Plate and its x-y Axes

89

VI

:1
1i

c<

r)

'

<J

'-ci ~
.....

,.__

~
I.')

-a'='

0
......

\)

:;J

t<

0
<S

0
0

00

"":s0

.:'-'$

()

t<

"'

q..
~

cS

t-:()

<)

()

.....
~C)

t....

(j
JI

>

"""
;

J)

()

()

.......,_/

ut

c0

.J)

d
I)

c<

,.

QI

i;

-;

()

"'"" ,......

C)

:<

<)

<a

a::

,,,

rn

rn

,..

'1)

-t

6()

\il

0)

:0

()

<>

()

ci

ft'

tl

c-:;

()

ti

0
M

~
t)

ft<

~
.E

'0 'ti

/'ft

<l

f,/\

lfl

<)
t

~
;$

-~

e-i

()'?

lC

l
~

-cs

<:.

...J
<
t::

...
-1
~

\..

J...

~
:.::.
<')

.:
0

.Jl
~

I :3 JI
I

90

--cs
d

-....J

~I:
x
~

0.04

'
~
0

t
Qt

Line Load

o;u:;

..r;
<S

&
0.02

Parabolic Line Load

0.1

0.2

Fig. 6.5 :

0.3

0.4

0.5

0.6

0.7

o.s

0.9

Width Ratio( d(n)

1.0

A Plot of Relative Rotations Versus /\..

91

Table 6.:2 : Joint Dimensions

Join.ii

Branch

No

JSR
J5

6 x 4 x0~~25,
6

x 4

~.'25

'J\.-o

b/t 0 tcftw

4X4 x 0.25

0~'67

24

l~O

4 x 4 xa.;25

o.'67

24

1.0

Rei~o:rc

5x5XJf PL
NON'E

92
~
QJ

.Y
\J

f
{)

ci

lO

-"ij'

LO

r,,

-a

J\

:s
0

...

<l

I
J J
0

3@

-(S

_J

\
\

\.""- \ \
"'o ~

93

Rein.forced Joint

Haunch

Fig. 6.7 : Haunch Cuttings

94

Fig~6.8

Defozmed. Pla.noo-e :Plate Reinforcement

On the Tension Side of Joint

F1.g;6.9 Relationships Between Shearing St:ress

And Strain for Rigid Plastic 'tJateria.1

o.' :i.i-----------""" c I

95

case (i)

Fig~ 1 6.11

Case (ii)

~
"',,,..
.i:J..g.
o.. 1 2

Case (iii)

5'

F.lg~ 6~'10

..........._________

~~

~c

Yield :~ine J?atte:rx:s f'o:r Cl::.orct Fl831.:~e s~tif.fener

Table 63

Moment

+ )(

Chord

B
g"" a">< Y~~

Wel>

4 u >'

= 2-l:>lny('l.32)
:::: 2..40

lf\~1'.<p\

60

ifi\<ir

ancl Web

Ca.r"'"c:.1ty of Joint by l'telcl .. hne 1'heo7_ ~

Unre"m sorc.d Jc,nt "'

Truss A
Chord
~ :ic. s'"" v~'

Wttb
4'' x
'/2..

Motnent Co.roc.\ttes- oj Joint

Fl~e.

Chord
Lower

Stifjen<tr

Upper Ext~me

E:l<t\'"lll\\Q

~..---.

M = 2b7ttt" ( 633')
::: 3+4il l 5

M =2.l>m,.-( 1o. l)

in\<~s

=S56

Me""bers

Movnent C~ftlc.'1ty

oJ-

We\> Mem.b~r

Fbn_ae flo-\e Re1n~rce~J (whtre cP:::.*-(d~- dn

M=

ln\<~s

2.\>'lT\f (+.32)
... ;i:&"i\.bt h ~*

= 458

ltik'f&

M = ~t"'"!
= 508'15

11\'cc.~~

Truss

,,

'J

rt

"Yl\t =::

344-

:::: 0. 8 6

*~

87.o

ink.ips/un1t. wicl~"
1hK;p$ /

uni-l. Wicl'\~

in'<irs

fot- v.,_u
(

\39 .. o nKir

no

in~K:fr.

290A6 in-k:r

"\htc.knQ.U 'm.ct\"er;o.\)

~r" V4-' t\,\c:.knus ~'\"el"io.l)

h ::: c! (or 7\b)

d'y ::: 55 \<st

\..0

0\

CHAPTER VII
stm!J.11fARY AND CONCLUSIONS

Connections made between

~No

unequal width square

semi-rigid~

HSS welded at :right angles are regarded as

This condition is caused by purely flexural action of the

connected chord flange due to internal loads transmitted to

the joint by the web

member~

Vierendeel trusses composed

unequal width squaxe

~f

HSS were used a.s examples utilizing this type of connection

and were ari..alysed by a matrix method taking semi-rigidity

into conside:ration.

The trusses were of equal 8@8

widths, parallel chords 8' and 13

tf

f1

web members, 4 x 4 x

if

'

11

Associated chord sizes were 8 x 8 x

panel

apart and uniform size of

for one case and 4 x 4 x

n

A computer programme was set

"
..1-

!
n

and 8 x 8 x

~o

'

.J..

es~ina~e

11

for the other.

t
.J..~

une

:respectively.,

maximum

deflection of the truss subjected to constant panel-p::>int

concentrated loads. With varying joint modulust the maximum

deflections were calculated and were plotted against the joint
modulus to facilitate the design.ers

choice of joint modulus

or dimensions.
Buckling of the compressive top chord :nember nas Ui.""llikely
in

.,

~ne

two examples considered in Cha.pters LL and III. Instead

of the top chord member, the end web members '.Vere cri tica.l

97

98
and needed to be increased in size due to the presence of

high stresses there.

The loading acting on the joint was confined to the
transmission of loads f:rom the web member only. A length of
four times

the width of the chord member could.be considered

as sufficiently long for the chord flange plate analysis for

rotation under applied moment.

The behaviour of the joint becomes non-linear even

at relatively low loads due to the very high stresses at the
corners of the

joint~

However, this yielding is quite

insignificant if the ultimate load capacity of the joint is

required. Thus, the plastic analysis of the joints was attempted
to estimate the ultimate strength of the joints.

As an ordinary unequal width connection is weak to

develop Mp or even My' several reinforcing techniques have been
suggested, such as fla.nze-'Plate reinforcing 2 chord flange

stiffener and haunched plate

rein.forcing~

AI1

approximate elastic

analysis of flange-plate reinforcement was attempted.

An

extension of the yield line method was also used to estimate

the strength of the reinforced joints. Moment capa.cities of

tb.e reinforced and unreinforced joints were then calculated to

compare the moment capacity of the web member. In each case,
it was found tbat the moment capacity of the joint is usually
lovrn:r than the moment capacity of web rn.ember. Eence, reinforce
ment is deemed. necessary if the strength of the web :nember is

to rJe fully utilized.

99

The following conclusions can be made based on the

study descri. bed;

(1) Vierendeel trusses are aesthetically pleasing

and are generally economical in a

nQ~ber

of applications such

as for pedestrian walkways and as roof trusses.

(2) Deflection criteria may be at least as important
to the designer as are strength considerations. Greater economies
are achieved with the use of unequaJ. web members welded to
chord members in HSS material.' However, this type of' design

reduces the overall stiffness because of semi-rigid connections

and may result in excessive deflections under worldng

loads~;

(3) In each of' the two trusses analysed, the web member
width to chord member width was equal to o~-5(i~e. 'J\.). Analyses

were carried out to show that the web to chord welded joints
r1e:re inadequate in two respects, i.e.,

(i) Strength as dete:rmined by a yield line solution,

(ii) Stiffness, from the point of view of deflection.

An adequate design could only be achieved by employing

one of these joint strengthening techniques described

herein~

( 4) Reducing 7\ wbile desirable in material saving may

be offset by costs incu:cred from joint stiffening.

analvsis was attemnted but

~

e~rnerience
~

No cost

in the use of HSS trusses

will tend to suggest o ptir.rum ]\ in relation to other geometrical

pro:pe:~ies

of the members and the associated joint stiffening

procedures most desirable.

APPENDIX I

SE:U-RIGID CONNECTION EQUATION

(5, 7,8)

Consider a prismatic member ab with semi-rigid

conne~

tions of joint moduli Ja and Jb (For defl.nation of J see foot

note on pp. 13) at ends a and b respectively.

The internal

end moments Ma and Mb due to the external loadings or support

settlement somewhere in the structure are acting at ends a and
b respectively.

The total angles of rotation at both ends are determined

by adding the rotations of the member without elastic rotational

spring subjected to Ma and Mb with the additional rotations due

to the e.xi.stance of the elastic rotational springs.
The first part of the :rotations of the member a b without

elastic spring subjected to Ma and ?.1 0 can be determined

Conjugate Beam. Turethod. ( 26) (27)

b~r

the

Consi<ler Fig. A wr.i ch shows the real beam a b and the

conjugate bea.m.

The bending monent diagram of ti1e real beam with

a multi:plie:r l/EI is regarded as 'loading' acting on the conjugate

beam.

To f'ind the reaction force of the conjugate

be&~

at the

support a, we take moments aoout b, thus

R 1
-a

"'r T
3 - ( Tr
) =0

, MhL ) T

+\

3EI .

11,r
.. ~1

L/c..,..,I
O..t!J
I

As the shear force of the conjugate l:eam at any particular

100

101

point is equal to the slope of the elastic line of real beam.

Thus, the rotation at a is

Similarly, by taking moments about a, we get the :rota

tion of the beam at b

The other pa.rt of the rotations due to the elastic

rotational springs are

where Ja and Jb are the joint moduli at ends a and b respectively.

Hence the total angles of rotation at both ends of the

beam are
(A.l)
(A~2)

Taking out the common factors Ma and

M~b

, two above

equations can be re-arranged as

0'.

_ ( LJa t

a -

Erb

3EI

) tf. 3EIJa
a

=- ( L/6Er }

( L/6EI ) Mb

(A~'3)

( A.4)

102

Multiplying both sides of Eqn.(A.4) by 2(LJa+. 3EI) /LJa

and adding it to Eqn.( A. 3) , we get

La : L ~(3EI/Ja)

Let
&
~rb

Lb ::: L

+( 3EI/Jb)

can be w:ri tten as.

I
T,

Similarly, M can be written as

a

2Lb
L
=6EI 4LaLb
S-a
+
6EI
2
.;. L
4LaLb If' Ja
Jb = J , then La = L 0 =L

Ma

0':

::a

Eqns~(A.6)

and (A.7) can be written as

1

rKa =

2L
6EI
'2
4L
-

2 S-a

L
M :

6EI.

41'2 - L2

tr

+ 6EI
+

a.

4T.' 2
2L

6EI

.u

2 Yb

4L'2 - L2

Eqns.(A.8) and (a.9) coincide with Eqn.(2.9)

(A.8)

103

Real Beam Under Actions

of End Moments

Ma.-1

E:t

Conjugate Beam Under

Conjugate Loading

Fig. A

Conjugate Beam

APPENDIX II

CGMPUT~R

PROGRAMME

HRK7,TlOO.
FTN(f~=3)

LGO.

6400 END OF RECORD

PROGRAM TST !INPUT,OUTPUT,TAE5=INPUT,TAPE6=0UTPUT)
THIS PROGRAMME IS USED FOR SOLVING THE VltRE~Oc~L TRUSS WITH SEMI-RIGID
CONNECTIONS AT dOTH E~DS uF THE ~~b MEMbLRS *****************************~
DIMENSION /.i..(26,5iJ),S(lvU> ,A.SAT(26,26) ,p(z6,lj ,x.(26,l) ,f(:JQ,ll'

c
r

'-

CSP~ING{25),Z(25),WEBSL(25i,STk~SS(l3,l),AXIAL(4,lJ,VEKT(4,l>

c
c

DIMENSION INDEX(26J

READ
(5,101> NP,NF,NLC

FORMAT
C3I5)
101
WHERE NP=NO. OF UNBALANCED MUMENTS AND UNbALANCED LINEAR FORCES AT JOlhTS
NF=NO. OF END MOMENTS,
NLC=No. OF LOADING CONuITlONS ON THE STRUCTURE
READ

(5tl02)

102 FORMAT
READ

99 FORMAT
~\J'RITE

((A( I ,J), J=l,Nfi, l=l,NP>

(10F6.3)
(5,99)
P<I,J),

J=ltNLCi,

I=l,NP)

(1F4.Q)
<6'104)

104 FORMAT

Cl3HOTHE MATRIX AJ
DO 105 I=l,NP
(6,106)
It
(A(I,Jl, J=l,NFl

105 WRITE
106 FORMAT (4H Row,13,1x,1P4El6.7/(8X,1P4E16.7})

ltJR IT E ( 6 d 1 J >

( 13HOTHE rl:ATR IX P J

110 FORMAT
DO 111 I= 1, NP

111 WRITE

(6,146)
I,
{P(I,J', J=l,NLCJ

(4H RGw~13,1x,1F4.0)

THE PANEL LENGTH OF THE VIERENDEEL TRUSS IS 8FT AND ITS HEIGHT IS l3FT
THE DIMENSIONS OF CHORD AND WEB ARE 8X8Xl/4 ANO 4X4Xl/4 RESPECTIVELY

146 FORMAT

DO 7 l

71
72
73

I=1,

61,

5(1)=7844.
DO 72 1=4, 64, 4
5(1)=7844.
DO 73 I=Z, 62, 4
5(I)=3922.

DO

74

I=3, 63, 4

74 SCI)=3922.

KK=O

DO 3001
II=l,10

KK=KK+l

EI=30000.*8.18

SPRING(K~>=FLOAT(II>*lO.*l0.*10.*lO.*lO./l.o

Z<KK>=I./SPRING(KK>

WEBSLCKKJ=l56.+3.*EI*Z(KKl

DO 81
1=65, 97,
4

81

S(l)=((6.*EI*2*WEBSL(KK)J/(4*WE6SL<KK~*W~dSL(K~1-156.*156.)l

C/12.0

DO 82

1=68,

B2 S ( I 1= ( (6
Cl U . 0

1' c

f..'-

100,
4

2 * 'w'J t. o ~ L { K i<-.

J J

I ( 4 ~- 1\i t. o ..::i L ( ";( i

-;<-

wt. o ,,'::) L { r-... r'..

J -

.L ::> c

~" J. :.:>

DO 83
I=66,
98,
4

83 5(1)=( (6.*EI*l56.)/(4.*WEBSLCKKJ*WlBSL(KK)-l~6.*lS6.l)/l2

DO
84
I=67
99,
4

34 S(ll=((6.*EI*l56.l/(4a*wEBSL<KKJ*WEbSL(KKJ-l56.*lj6.J};l2

DO 112
I=i,NP

DO 112 J=l,i\lP

AS.ATCI,J)=O.

>J

lOS
DO

112 K=l,NF

Kl=<K-1)/2-*2+1

112

K2=(K+ll/2*2
K3=2*K-l
K4=2*K

ASATtI,J>=ASAT(l,J}+A(I,~>*CS(K3)*A(J,~l)+S(K4J*A(J,K2J}

DO 113 I=l,NP
113 INDEXCI)=O
114 AMAX=-1.
DO 115 I= 1 , NP
IF CINDEX<I)) 115' 116t 115
116 TEMP=ABS<ASATCI,Il)
IF <TEMP-AMAX> 115,115, 117
11 7 I COL= I
AMAX=TEMP
115 CONTINUE
IF
CAMAXl
118' 3001, 119
119 INDEX <ICOL>=l
PIVOT=ASAT<ICOL,ICOL)
ASAT<ICOL,ICOLJ=l.O
PIVOT=l./PIVOT
DO
120 J=l,NP
120 ASAT(ICQL,J'=ASAT<ICQL,J'*PIVOT
DO 121 I=l,NP
IF <I-ICOL}l22, 121, 122
122 TEMP=ASAT{I,ICOL>
ASATtI,ICOL>=O.O
DO 123 J=l,NP
123 ASATCI,J)=ASATCI,Jl-ASAT<ICQL,J'*TEMP
121 CONTINUE
GO TO 114
118 DO 124 I=l,NP
DO 124 J=l,NLC
X<I,J>=O.
DO 124 K=l,NP
124 X(I,J}=X(l,J}+ASAT{I,K)*P(K,Jj
00 127 I=l,NF
Il=(l-ll/2*2+1

IZ=<I+l)/2*2
I3=2*I-l
I4=2*I
DO 127 J=l,NLC
FCI,J}=O.

DO 127 K=l,NP
127 F{I,Jl=FCI,Jl+X{K,J)*(S(l3)*A(K,Il)+S{I4J*A(K,I2JJ
AX I A L AND VER T AHE THE A X I A L FU i~ CC. S

vF

CH 01-.( 0 A "i 0 'w :. o

j

i, i

E ivf d

t.r'\ S

DO 210 .J=l,NLC
AXIAL(l,J}=(.5*7*Pl20,JJ*d.+F(L,JJ+f(lb,JJ l/130
AXIAL{2,Jl=(.5*7o*PC2J,...Jl*l6.-P(2Q,J)*8.+F(4,JJ+F zo,Jji/lj
.r:..., X I AL { 3 , J ) =t 5 * 7 i P ( 2 0 , ..J ) -it 2 L~ - P ( 2 J , J J ~~ ( l 6 o + 6 i + F {
, J ' + F ( 2 ::.: , ...J
AXIAL(4,Jl=<.5*7.*P(20,..J}*32.-P(2C,Jl*(24.+l6.+d. +f{d,J
C+F ( 2 4

,J) } /

13. 0

V~RT(l,JJ=lF<l,J}+f(2,JJJ/8.
V~KT(~,Jl=C~(3,J)+f(4,~J-F(l,J'-f(i,J''/8.

vERT(3,Jl=(F(5,Jl+F(6,J'-F(3,JJ-F(4J' 1;3.
V~kT{4,J)=(f(7,~l+F{8,Jl-F(5,JJ-F(b,JJ)/8o

1-< t. Si..:> C. C T I

) J

J.

VE L Y

106

VERT(5,J)=(F(9,Jl+FClO,J)-F(7,JJ-F(8,J})/8
210 CONTINUE
SM AND SM2 ARE THE SECTIU~ MODULI UF CHukD A~D WEb
SM=18.8

NE.~clEKS

RESPECTIVELY

Siv1Z =4. 09

......

THE CROSS SECTIONAL AR EA OF THE CriC.n<D MElvit:H::R IS 7 5':1 SCil. IN

THE CROSS SECTIONAL AREA OF THE W~~ MEMBER IS 3.54 SLl. IN.
DO 310 J=l,NLC
5TRESS(l,Jj=ABS<F<l,JJ*l2./SMJ+ABS{AX1ALC1,Jj/7.59}
STRES5(2,J)=A8S(FC2,J)*l2/SM)+AB5CAXIAL(l,Jl/759>
STRESS(3,JJ=ABS(f(3,J'*l2./SM'+AB5CAXIAL(2,J'/7.59)
STRESS(4,J)=A85{F{4,JJ*12./SMJ+Aci5{AXIAL(2,Ji/7.59J
STRESS(5,JJ=A85(F(5,J}*l2/SMl+Ad5(AXIAL(3,J>;7.59)
STRE55{6,J>=ABS<F<6,Jl*l2.ISMl+ABStAXIAL(3,J}/7.59J
STRESSC7,Jl=ABSCF<7,Jl*l2./SMi+ABS<AXIALt4,Jl/7.59l
STRESS(8,J)=ABS<F<8,J>*l2/SM)+ABS<AXIALC4,Jl/759)
STRESS(9,Jl=ABS<VERT<l,J>13.54l+ABSCFC34,J>*l2./SM21
STRESS<lO,Jl=ABS<VERT(2,JJj3.54J+ABSCF(36,~)*12/SMZJ

STRESSC11,J>=ABSlVERT(3,J)/3.54l+AbS[f(38,J}*l2/SM2>

STRESS<l2,Jl=ABS(VERT(4,Jl/3.54i+AdSlFl4Q,Jl*l2/SM2J

STRESSC13,Jl=ABSlVERT(5,J)/3.54>+A8S(f (42,Jl*l2./SN2)

310 CONTINUE
~'/RITE

(6,107>

107 FORMAT (13HOTHE MATRIX

s>

DO
108 I=l,NF

Il=<I-1)/2*2+1

I2=(I+ll/2*2

I3=2*I-l

I4=2*I

108 WRITE
(6,109)
r, 11, SCI3'' I2, S<I4l

1U9 FORMAT (4H Row,13,5x,3HCOL,I3,1PE16.7,5X,3HCUL,I3,1PE16.7l

WRITE

(6,1128>

1128 FORMAT Cl3HOTHE MATRIX f}

DO 1129 I =l ,NF

1129 WRITE
(6'1136) I' .<F( I ,J>, J=l,NLCJ

1136 FORMAT
(4rl Row,13,1x, 1Fl4.4 1

WRITE
1125 FORMAT

DO

(6,1125)

<13HOTHE MATRIX XJ

1126

I=ltNP

1126 WRJTE
(6,1136) I, lXCI,JJ,~ J=l,NLC)

WR i TE . ( 6 , 18 2 0 }

1820 FORMAT l54HOTHE AciSOLUTE VALUE OF STRcSS IN

DO 18 2 1
I =1 , 13

1821 WRITE
(6,1830> 1'

1830 FORMAT
WRITE

1840 FORMAT

C52HlTHE AXIAL

1Fl4.4J

FORC~S

Ii~

CHORD AND

I=l,4

00 1851
186U

FORM.~T

1=1,5

{6,1860'

I, (VERTCI,JJ, J=l,NLCJ

(5H VE.RT,

30C;l CONTINUE

STOP

END

0F TRUSS/)

(6,1840)

00 1841

W~ITE

M~~bc~S

<STRESS< ItJJ' ..J=l,NLCJ

<7H STRESS, 13, lX,

1841 1i1iRITE
(6'1850} I,
(AXIAL< r,J), J=l,NLC)

1850 FORMAT
C6H AXIAL, 13, lX, 1Fl4.4J

1851

TH~

14, lX,

1F14.4 1

W~b

MtMb~RS

IN KIPS//)

L"'PFENDIX

rrr

LOADING EQUATIONS EXPRESSED IN

TfilGONOMETRIC SERIES
UNIFORMLY filSTRI:BUTED LINE IDAD
Cbnsider a uniformly distr.i buted line load of: intensity

Clo acting at a location of a plate as shown in F.lg. B

Tl:Us loading equation can. be expressed into double

t:rigonomet:ric series.- One component of the equation is expressed

in x-direction., the other component is exp:ressed in y-direction.

The product of two components is the required double trigonometric

series .

In y-direction, the loading can be :regarded as U.D.L.

o:f intensity q 0 acting on a simply supported beam. For a

st:ructuraJ. element with simply supported ends the half range

series which is most of'ten used is the sine

series~;

Therefore,

the load may be represented as

To find qn multipy both sides of this equation by

sin. n7$.y /b a.i'J.d integrate from O to b.

b

S0

sin m.~y sin. n'!'y dy

b

the resulting equation is

107

Noting that

'

b/2,

when m ~ n

".-7hen m_

=n

108

b , rb

qn(~ =Jo q sin

'E!!::JL.

dy

qb

= n"tt (l~

cos. .n'll:)

From. above equation

q

= 4qo

when n is odd

n~

an.a. qn =a

when. n is even

Therefore, Eqt4~(A.'10) is expressed as

(A.11)

Ih. the x-direction, the loading can be regarded as a

unit concentrated load (because of line load) acting at a

distance

from one end of a simplied supported beam of length

2b. To derive the expression for this unit concentrated load,

we can treat. the loading as a UDL of intensity l/2E between

x =t- and x

=r+t::

(see Fig.B). The loading on the beam is then;

=p-f:.

zero

from

.x

=0

to

1/2

from

=~-t

to

x =p+<c

zero

from

=r' +e

to

x- 2b

Like

Eqn.(.A~10),

the loading may be ;represented as

Zqn sin n~x/2b, multiplying both sides by sin n1!.:X/2b and

integrating betweenO and 2b we get

~~(o)

n~z

sin 2 b

dx i"

=~ 0

2b

~f+E 1

. n~x
sin - - dx f-i2E
2b
-.

2 n7r..,..

q_ sin ~dx
n
2o

2b

n"(t.x
(0) sin - - dx
2b

('t"~

109

1 2b (
n<.t(
-2f nll.
cos -2b

Whence

f-f ) -

n~(

cos 2b , I/J-tf

)J _ %
--

2b
2

Using the cosine combination. formula, this give

I
n~c
q n -- - 2 sin.
l!:ll"{{
2h

and since

,,

sin~-+ n~t
2b
2b

n~

Sl..n

2b

as

z.-+O

. _ 1. .. niIJ
qn --h sin 2b

The expression. for this unit concentrated load is

thus [17J

1h sin.

2'b

({,.a

Multiplying

..

~%

,,

2t .

sin 2 b ..,. sin 2 b sin

Eqn~;(A.11)

with

2<cr '

~+

Eqn~;(A~:l2),

(A.12)

we obtain the

required double trigonometric series for the uniform line load

acting on the plate,

i~e.

n~x)

SJ..n2b

where
Equation (A.13) coincides with Eqn.(6.9) ..

110

1?ARABOLIC LINE LOAD

Consider a parabolic line load at a location.of a

plate aa shown.. in

Fig~'

a.l.~eas

C The

of this :parabolic

distri~

bution. and the U.D.L. in preceding Section.must equal each

other. The intensity of this. pa:rabolic distribution is found

equal to

3~

at both ends and zero at the center.

The double sine series for this parabolic line load

can~

be decomposed into two components, namely x and y components.

Before dealing with the. comJ?Qnent equation, we would

like to find the equation of a parabolic curve as shown in

Fig~'D

where q is the vertical a.xis and y is the horizontal axis.

The general equation for the parabola in Fig,'D) is f21]

< y ~ bI 2

:: 4k

f q - o)

where ( b/2, 0 ) and ( b/2-, k ) are the coordinates of the

vertex and the focus

respectively~'

The value of k can be found

by substituting the boundary value of the curve, namely when

y :::

o,

.!

q 3q0

K is thus found equal to b 2/48q0

Expxess

q in tenns of y, Eqn.(A.14) can be written as

q

= ( y ..:.

b/2 )

12Qo
b?.

(A.15)

Returning to the y-com:ponent equation., we can regard

the loading in y-direction as a parabolic distribution load

acting on a simply supported beam of length b. Let this load

be represented as

111
n 2'llY
s n m"!J:y
q ::: ql ..~,-,i n 7!.b Y ...a.
'q2 si -0-~ + qm i
b ~

(A.16)
Where

above equation

can be found by multi plying_- both sides of

b~

sin

~y/b

and integrating with respect to

y from 0 to b.

Thus
Substituting q as expressed by
equation, we get

%n

=m:cc
24qo r~
.

when m is odd
when

n is:. even

Therefore,

into above

. ; /21[1
m z.<:j

= 4sqo

-m
m. ~
a--m :: O

Eq~( A;~16)

Eqn.(A~il6)

( l..
..:..

{-l)m]

~ )
m <L:2

can be wrt tten as

wbich is the y-component of the parabolic line load acting

on the

plate~

As in preceding Section, the loading in

~-direction

can be regarded as a unit concentrated load acting at a distance

from one end of a sim:ply supported beam of le- 11gth

expression for this u.."'lit load is

1 [ si n 15.(J s;n <:ex
b
2b
2b I
.J.

" 2~\
.... i.. n 2S:e sin
20
,.::J

J.

.2b

2b~t

The

112

The product of the two comronents gives the req_uired

double sine se:ries, thus

(A~'l8)

where

<Iron --

48qo
b<t

1 .(
m

=~
.:. (
b~ m

.:t:.
4

2)
nt2~2

.. n

SJ. .

:.ttf/2b

3 ~ ~m ) sin n';?!:.f/2b
m2~

113
i

w
.,..

---~b---~-1___.dyUnifo:rm Line Load on a Plate

i
i.

uvuwumumumimJ~

w<

,,

------

Line load in y-direction

f'

fx

.'.:l,.p

-1

.,.,.;t

.2.!.

im

-IJj ;,
lllllj

!~

t-

ib

nine L.oad in x-clirection

Rl.g_B : Unio:rmly Distributed Line Load

++

114

Fig. C

Parabolic Line Load Acting on a Plate

Fi.g. D

A Parabolic curve on

y-q axes

APP3IDIX IV

1.

Matrices :

c~J

Static matrix

[BJ

Deformation matrix

p1)

Internal end matrix

{:e}

A colunm matrix showing values of tmbalanced joint

moments and Ui!balanced linear forces

(sJ

Stiffness matrix

{x}

A column matrix showing values of joint rotations

or translations

{&}

A column matrix sb.owing values of elastic end

rotations
A symbol used to indicate the
2.

transr~se

of a matrix

Notations
a

Length of ?late in ::-direction

V:lidth of Plate in ;I-direction; the width of chord

member of square HSS

C

Conpressive force in triangular pinned truss

Width of web 2ember of squa:re HSS

Elastic flexural rigidity of plate of lmi t width,

E I
1 -)?

Elastic nodulus of steel

E~:,,

Tangent

modul~s

115

of steel

116
G

Shear modulus

Height of Vierendeel truss ; height of stiffening

plate

Moment of inertia

Joint modulus (J

Plate coefficient

lenght ;"1 span lenght of beam ;'1 panel lenght 0

~ M/~)

Vierendeel truss
Modified length of the structural member with
semi-rigid connections at ends
mp

Fully plastic moment

o~

plate (per unit width)

M-

Applied bending moment

Load

Maximum intensity 0 distributed load

Plate thickness

x,y

Co-ordinate axes

Transverse deflection of thin plate

1.0/J or /M

Zn

Plastic section modulus

J.,;

'-

The angle made between the diagonal yield line

and

the vertical

The angle made between the diagonal yield line

and the horizontal

Shearing deformation

A~
}

Deflection

Small inc:re::ient

Coefficient of restraint

117

?\.

Width ratio (d/b)

Y
f

Poission's ratio
A certain distance from one end beam --ahere a

concentrated load is acting

.E

Summation

<(

Sress

Buckling stress

efy

Yield stress

<f'f

FropoTtional limit stress

r(

Shearing strss; E/Et

<f

Relative angle of rotation between two members

at joint

3.

Abbreviations

AISC

American Institute of Steel Conctruction

CIDECT

The International Committee for the study and

Develop2ent of Tubular Structures (Comite'
international pour 1 1 etude et le de'veloppe=ient

HSS

HolloTI structural Sections

i',Iin.

Minimum

!'fa.x.

Maximum

1TBC

national Building Code

FDL

Ur.if'o:rml~...

di stri :mted Loa.:.l

118

APPENDIX V
LIST OF REFEEE1WES

(1) Mehrotra, B.L. , "Stiffening in Tubular Junctions - Some

new Concepts", Journal of the structu1a1 Di vision,

ASCE, Vol 97, No. ST9, Sept. 1971 pp. 2453 - 2457
(2) Mehrotra, B.L. , "Matrix Analysis of Welded tubular Joints",
Ph.D. Thesis, McGill University, Canada, Dec. 1969

( 3) Wang, C. K. , "1\-Iatrix Formulation of Slope-deflection

Equationsu, Jouxnal of the Structural Division, ASCE,

Vol. 84, No. ST6,

l~oc.

faper 1819, Oct.,1958

pr:. 1819-1

1819-19
(4) Wang, C. K. , t1Mat:rix zJethods o:f Structural Analysis",

International Textbook Company, 2nd edition, 1970 p:p. 83

( 5)

Gere, J. r~r.

"Moment Di stri butionn, D. Van Nostrand Company,

Inc., Princeton, 1963 pp. 239

J.~.

,ibid,

p~.

241

(6)

Gere,

( 7)

Lathers, rJ .E. , "Advanced Design in Structural Steel",

Prentice - Hall, Inc. 1960

p~.

375

(8) Rathbu..."l, J.C., "Elastic Properties of Riveted Connectionsn,

ASCE Vol. 101, 1936 pp. 524 - 595

(9)

National Building Code of Canada, 1965

119
( 10)

Bleich, F. , "Buckli:c.g Strength of ~1~etal Structures"

I.1cn.y.~nl
- ...
......
:::.t..~

( 11)

,..Tj
!'.l.-

1 ...t...1

"!100i,..
rtonpa-1-~
J)
.l.:)..
ll
ii
~l ~}

'

Inc.,

1952

Eudoba, J. , "Plastic Design Capabilities of Hollow

Structural Sections", M.Eng. Thesis, rJcMe.ster University,

Jan. 1971
(12) Redwood, R.G. , "Behaviour of Joints Between Rectangular
Hollow Structural ::re:nbers", Civil Engineering and Public
Works Review, Oct., 1965, Pf 1463 - 1468
( 13)

Jubb, J.E. and Redwood, R. G., "Design of Joints to :Box

Sectionsn, The Institution of Structural Engineers,
Conference on Industrialised Building ancl the Structural
Engineer, May 17, 1966

( 14)

Wood, R.R. , "Plastic and Elastic Design of Slabs and

Plates", Thames

( 15)

& Hudson, 1961

Cute, TI. , Camo, S. , ancl Ru.rnpf, ,J. L. , HWelded Connections

for Square and Rectangular Structural Steel Tubing",

Research Rer;ort Fo. 292 - 10 , Drexel Institute of
Teclmology, ?l,,i lazlelphia, Pa., Novenber 1968.
( 16)

D:cucker, D. C. and CLen, VI. J?. ,

Di scontint:ous
(17)

to Bound

0n the Use of Simple

T.i

rrri t Lo ad !3 tt,

Dru.cke:r, D.C. , J?rage:r, 77., G-reenberg, E.J., , HE:;:tended

Llath. 9,
(\ ---;-?.
)
'-"I

~?i e ld s

~~

331 - 389, 1952

120
(19)

Timos!1enko, S. and Woinowskj" - Krieger,

Plates and Shellsn

( 20)

~.TcG-raw-Hill

s.,

"Theory of

3ook Co., 1959, Chapter 4.

Preliminary Results of "Rectangular Hollow Structural

Viereneel tToint Tests", Corby Research Center, U.K.

( 21)

Woods, R., "Analytic Geometry" 1939

( 22)

Du.ff, G., "Joint Behaviour of a Welded Beam-Coltt"D.n

Connection in Rectar.!.gu.lar Hollow Sections. n

Thesis,

The College of AeTonautics, Cranfield, England, June 1963

(23)

Mehrotra, B.L. and Redwood, R.G., "Load Transfer Through

Connection Between Box Sections", Journal of the

Engineering Institute of Canada, Aug. - Sept. , 1970

(24)

Redwood, R.G. , "The Bending of Plate Loaded Through a

Ri~d

Eecta.r:gular Inclusiontt,

Ir~ternational

Journal of

:Mechanical Sciences, Vol. 7 , 196.J

( 25)

Reissr.i.er, E.,

nAnal~/sis

of Sheer-lag in Box

Bea~1s by

the

Principle of ;:IinL:.m.ill Potential Energyn, Q. AfI;lied Math.

Oct., 1946
(26)

Wilbur, J.B. and rorris, C.H., "Elementary Structural

.Ar-Lalysisn McGra\'v-Hill Book Co., Inc., New Yor1r 1948

(27)

Borg, S.F. , Genna!'o, J.J., ttMorlern Structural Analysis",

1969