CRANFIELD INSTITUTE OF TECHNOLOGY

SCHOOL OF AUTOMOTIVE STUDIES

Ph. D. THESIS

M. S. I. ALVI

THE STRESS DISTRIBUTION IN THE JOINTS

OF VEHICLE CHASSIS FRA24ES

SUBJECT TO TORSION

Supervisor: G. H. TIDBURY

February, 1978
This thesis is dedicated to my parents,

elders, family members and friends who

have always remembered me in their.

prayers and without their encouragement

I could not have completed this work.

ABSTRACT,

The problem of warping inhibition in the joints, of commercial

vehicle chassis frames is very important for the estimation.
of stress distribution in the joints and the overall torsional
stiffness of frames. The joints being the weakest parts in
the entire frame require that the stress levels should be
estimated when designing a frame for required torsional stiffness.
Plate theory has been applied to-analyse stress distributions
in the joints of a ladder frame-in torsion. The frame consists
of channel section cross members-welded to the web: of the side
members. The degree of cross member warping inhibition in the
welded joints has also been estimated analytically by using
the concept of the strip beam theory. The analytical values of
-
stresses have been compared with-photoelastic measurements on
photoelastic models of isolated joints and a complete ladder
frame.
ACKNOWLEDGEMENTS

I would like to express my sincere thanks to my supervisor,

Mr GH Tidbury, for his timely guidance, inspiration and
encouragement which has played a major part in the completion
of this work. I want to thank the authorities of The School of
Automotive Studies for providing me with the facilities to carry
out my Ph. D.. studies there.
_
I would-also like to express my gratitutde to the Vice Chancellor
of The University of Engineering and Technology, Lahore, Pakistan,
for granting me study leave.

My special thanks to my friend Ing. Dusan Kecman for helping me

to get important information from the work published in Russian
without which this work would have been narrow and almost incomplete.
I wish to show my gratitude to Dr Majid Sadeghi and Mr KP Nair
for their help.

I would also like to thank Messrs Reg Aldridge and John Heathorn"
and other members of the workshop staff who helped in making
aluminium moulds and advised, me on overcoming the difficulties
encountered during the casting of the Araldite models.

Finally, my grateful thanks to my typist, Mrs E Denton, 'who did

an excellent job to bring this work to its final shape.
 CONTENTS
Page

ABSTRACT
ACKNOWLEDGEMENTS
LIST OF FIGURES
LIST OF PHOTOGRAPHS

NOTATION

INTRODUCTION-
CHAPTER 1 LITERATURE SURVEY 1

1.1. ANALYTICAL METHODS, 1

1.1.1 The application of Vlasov's theory

in joint analysis 2
1.1.2 The application of Plate theory in
joint analysis 3

1.1.3 Joint analysis using, simple models. '4

1.2. THE APPLICATION OF FINITEMENENT TECHNIQUES IN
CHASSIS FRAME ANALYSIS 5

1.2.1 Analysis without consideration of

restrained torsion effects. 5

1.2.2 Estimation of the degree of warping

restraint. 5
..
1.2.3 Stiffness matrix including restrained
torsion terms. 7
1.3. OTHER APPROACHES 8

1.4. APPROACH SELECTED FOR THE PRESENT ANALYSIS 8

CHAPTER 2 THEORETICAL INVESTIGATION 10

2.1. BRIEF DESCRIPTION 10

2.1.1 Sequence Calculations 10

of
2.2. BENDING OF ASIDE MEMBER WEB DUE TO CROSS
MEMBER WARPING LOADS 16

2.2.1 loads 16
Cross member flange warping
2.2.2 Cross loads 20
member web warping
2.3. APPLICATION Or PLATE THEORY TO THIN WALLED
BEAMS OF OPEN SECTION 23

2.4 ANALYSIS OF A JOINT. 24

2.5 SIDE MEMBER SUBJECT TO CROSS MEMBER WARPING

LOADS 25

2.5.1 Analysis of a symmetric joint. 25

2.5.2 Analysis of an asymmetric joint. 30
 2.6. SIDE MEMBER SUBJECT TO A COUPLE LOADING 34
2.7. COMPUTATION OF ANALYTICAL SOLUTIONS 37
2.8. THEORETICAL ESTIMATION OF CROSS MEMBER
WARPING RESTRAINT FACTORS 38
CHAPTER 3 EXPERIMENTAL INVESTIGATIONS 46
3.1. INTRODUCTION- 46
3.2. METHOD OF PHOTOELASTIC ANALYSIS EMPLOYED 46
3.3. PHOTOELASTIC MODELS 47
3.4 CALIBRATION OF CT200 ARALDITE PLASTIC 49
3.4.1. Determination of elastic constants. 49
3.4.2. Determination of material fringe
f" 51
value"
3.5. PHOTOELASTIC INVESTIGATION OF A COMPLETE
WARPING INHIBITION CASE 54
3.6. TESTING OF ISOLATED JOINTS 54
3.6.1. Method of support and loading 54
3.6.2. Photoelastic investigation 56
3.7. TESTING OF A LADDER FRAME 56
3.7.1. Method of support and loading 57
3.7.2. Photoelastic investigation 57
3.8. PHOTOGRAPHIC RECORD OF FRINGE PATTERNS 59

CHAPTER 4 DISC USSION OF RESULTS 67

4.1. BEHAVIOUR OF CT200 MODEL MATERIAL 67
4.2. INTERPRETATION OF FRINGE PATTERNS 67
4.2.1. Coloured fringe patterns 67
4.2.2. Black and white fringe patterns 69
4.3. CROSS MEMBER PARTIAL WARPING RESTRAINT
FACTORS 71
4.4. COMPARISON OF EXPERIMENTAL AND THEORETICAL
STRESS VALUES 71

4.4.1. Isolated joints 72

4.4.2. Ladder frame joints 73

CHAPTER 5 CONCLUSIONS AND FUTURE APPROACH 100

5.1. CONCLUSIONS 100
5.2. SUGGESTIONS FOR FUTURE APPROACH 101

REFERENCES 102
APPENDIX I GENERAL BENDING PROBLEM OF A RECTANGULAR PLATE 106
APPENDIX II GENERAL STRETCHING PROBLEM OF A RECTANGULAR PLATE 111
APPENDIX III FLOW CHARTS 116
APPENDIX IV PREPARATION OF PHOTOELASTIC MODELS 119
 LIST OF FIGURES

Fig. No. Page

2.1a A rectangular plate referred to a cartesian

system.. 11
co-ordinate

2.1b Forces and moments on a plate element subject

to bending 11

2.1c Force components on a plate element subject to

11
stretching

2.2a Side member web subject to the cross member

forces. 13
end warping

2.2b Side member web subject to the cross member

. forces. 13
web warping

2.3a A channel section referred to the cartesian

system 15
co-ordinate

2.3b Joint under investigation 15

2.4a A channel section side member subject to a

channel section cross member end warping forces. 17

2.4b A channel section side member subject to a

loading 17
couple

2.5a Theoreticäl distribution of normal displacements

on the side member web. 40

2.5b Theoretical distribution'of the bending moment m22

on side member" web. 40

2.6 Theoretical distribution of the side member web

normal displacement along cross member , end profile
(Joint I) 41

2.7a Theoretical distribution of-the side member web

normal displacement along cross member end profile
(Joint II) 42
2.7b (Joint IV) 42

2.8 (Joint III) 43

2.9 (Ladder frame inner joint ) 44

3.1. Dimensions of the ladder frame analysed. 50

3.2 Calibration of CT200 Araldite for the determination

of elastic constants 52

3.3. Calibration using a tension peice. 53

 3.4 Calibration using a compression disc. 53
.
3.5 Distribution of direct stress along the
free edge of the channel section. 56

3.6. Support points on the ladder frame. 59

.
3.7. Method of attachment of bracket. 59

4.1 The distribution of maximum shear stress on the,

side member flange (Joint I) 83

4.2 (Joint II) 84

4.3 "". (Joint III) 85

4.4. (Joint IV) 86

4.5 The distribution of the shear stress on the

side member flange (Joint I) 87

4.6 '(Joint 'II) 88

.
4.7. (Joint III) 89

4.8 (Joint 'IV) 90

'4.9. The distribution of the direct stress on the

side member flange (Joint I) 91

4.10 (Joint II) 92

4.11 (Joint III) 93

4.12 (Joint IV) 94

4.13 The distribution of direct stress along the

cross member flange (Joint I) 95.

4.14 (Joint II) 96

4.15 (Joint III) 97

4.16 The distribution of maximum shear stress on

the side member flange (Ladder frame joints) 98

4.17 The distribution of direct stress on the

cross member flange (ladder frame joints) 99
 LIST OF PHOTOGRAPHS

Plate Page

3.1 Aluminium Moulds and, Cast Channel Sections 60

3.2 Photoelastic Models of the Isolated Jointes 60

3.3 Parts of the Joint Casting Mould 61

3.4 Aluminium mould and model of the cast joint 61

3.5 Testing of an isolated joint 62

3.6 The ladder frame subject to twisting. 62

3.7 Isochromatic pattern (Joint I) 63

3.8 Mixed fringe pattern (Joint-I) 63

3.9 Isochromatic Pattern (Joint III) 6464

3.10 Fringe Pattern (Joint III)

3.11 Mixed Fringe Pattern (Joint III) 65

3.12 Fringe Pattern (Joint III) 65

3.13 Isochromatic Pattern (ladder frame joint) 66

4.1 Isochromatic, Pattern (Joint I) 75

4.2 Mixed Fringe Patterns (Joint I) 76

4.3 Fringe Patterns (Joint II) 77

4.4 Isoclinic Fringe Patterns (Joint II) 78

4.5 ýý It " II II
79

4.6 80

4.7 Mixed Fringe Pattern (Joint III) 81

4.8 Mixed Fringe Patterns (Ladder frame joint) 81

4.9 Isochromatic Patterns (Ladder frame joints) 82

 NOTATION'

xix2x3 Catesian co-ordinate axes,

ui Displacement components

zig Stress components

nib Force components

qi Shear'force components

mi d Component of moments

E Young's Modulus

G Shear Modulus

v Poisson-Ratio
Value_..
fcI" Material Fringe

Direct stress in the cross member

Z profile.
P. Half the' length of the cross member
t Cross member section thickness.
L Half the length of the side' member.
h Side member section thickness.

ko Cross member flange warping force intensity.

C Strip beam theory constant.

Cav Average value'of C used in this work

K Cross member warping restraint factor.

m7r
am a. LL

¢1 = Coshamx2

sinhamx2
03 x2$1
04
= x202

p Cross member end warping force per unit length

Dimensions of the joint are defined in relevant figures.

Other variables and constants are defined in the text during
theoretical development of the analysis.
 INTRODUCTION

A chassis frame is an essential structural part of any commercial

automobile vehicle with regard to its load carrying capacity.
Besides payload it is subjected to various kinds of loads
including static, dynamic and shock loads, etc. during vehicle
operation. As such a chassis frame should require maximum
attention during the design of a vehicle structure.

Chassis frames in trucks, trailers and semi-trailers generally

(or longitudinal) members
consist of two side and several cross
members arranged in a ladder configuration. Because of this
arrangement they are also described as ladder frames. The side
members and cross members are usually of open channel or'I-sections
but sometimes closed, tubular sections are employed as cross members.

There is a great variety of designs of joints between the cross

members and side members both as to joint configuration and
the method of attachment of cross member to side member. The
design of chassis frames has been based on the bending loads and
torsional loads. The side members are generally designed to resist
bending loads due to payloads and the cross members mainly stand
torsional loads which arise due to many, reasons during vehicle
operation.

A chassis frame is expected to exhibit a torsional stiffness which

should meet the operational requirements of, the vehicle. Because
of this consideration more emphasis has been given to the proper
control of the overall torsional stiffness values in chassis frames.
However the fact should not be ignored that the frame members should
not be subjected to high stresses that could lead to their failure
thus making the frame ineffective.

The weakest parts of the frame are its joints which are also the
regions of high stresses due to many factors. It is necessary
that the desired torsional stiffness of a frame should not produce
very high stress in the joints which could cause the failure of
the joints.

The major factor for the cause of high stresses in the frame joints
is due to the warping inhibition in them. Open section beams warp
in torsion which, when restrained, produce high localised stresses.
The degree of warping inhibition in a joint also controls the overall,
frame torsional stiffness. So it is important for a designer,
before designing a frame for required torsional stiffness, to
estimate the following

1. The degree of warping inhibition in the joints analytically

or experimentally.

2. The distribution of stresses in joints.

Various automatic structural programmes to analyse frames have

been developed which give a good estimation of the overall
.
torsional stiffness but stress estimations in'the joints are .
not accurate enough. A few attempts have been made to investi-
gate stresses in joints by analytic methods. Most information
regarding stresses in joints has been obtained by using experimental
methods on actual frames., Such information cannot be generalised
for joints of other dimensions and configuration. In order to
improve knowledge in this area an investigation has been carried
out in the present work to analyse stress distributions and the
degree of warping inhibition in the joints of a ladder frame in
torsion. The frame consists of channel section cross members
welded to the web of channel section side members. The equations
of plate theory have been employed in this analysis and the theore-
tical stress values have been compared with experimental stresses
obtained by using reflection techniques of photoelasticity on
photoelastic models.
 1

CHAPTER 1 LITERATURE SURVEY

A general survey of the literature published in English indicates

that investigations in commercial vehicle chassis frames have been
conducted mostly for estimating torsional stiffness of these frames
rather than stresses occurring in the area of a joint. In dealing
with frames consisting of open sections, torsional stiffness and
stresses in them are mainly affected by the; amount of warping inhi-
--.
bitions in the joint. The stresses occurring in the cross members
have been obtained from warping inhibition investigations. Research
work in this field has received greater attention in East European
countries and a number of references are available specially in the
Russian language. Probably Vlasov's [1] in thin walled
major work
beams attracted attention in those countries. English translations'
of work published in Russia in the field are not available and the
author reviewed some of the references appearing later in this
section with the help of a colleague conversant with the Russian
language.

The joints of ladder type frames have been investigated by various

research workers employing the following methods.

i) Vlasov's approach for analysing open section thin walled beams

subject to arbitrary loadings.

ii) Wagner Kappus' approach for analysing open section thin walled
beams subject to torsional loads.

iii) Finite element methods available for structural analysis.

iv) Analytical approach based on classical plate theory.

The analysis of entire frames has been mostly carried out by employing
com,, uterised automatic programmes available for a variety of structural
engineering problems. In certain cases, the conventional programmes
have been modified to include special effects associated with the
behaviour of thin walled beams subjected to torsional loads.

The analysis of joints requires an accurate estimation of warping

inhibition in every joint as its value depends on many factors
including methods of attachment employed for connecting cross members
to the side members. The stresses in a joint depend upon the manner
in-which the load is transferred from cross member to the side member
and vice versa. Most of the research conducted in the analysis of
chassis frames depended on the estimation of warping inhibition by
simple experimental tests on isolated joints. Very few attempts
have been made to estimate this effect theoretically. A survey of
the published literature is presented in the following pages. The
author has considered it useful to briefly point out the salient
features of work published in Russian or German for future reference.

1.1. ANALYTICAL METHODS

The torsional stiffness of chassis frames has been estimated in some

 2

cases by analytical methods employing Vlasov's theory of thin

walled beams [ 1'. The amount of partial restraint in a frame
joint of open section members has been determined by purely experi-
mental procedures for the analytical methods in most cases but a
few have employed theoretical estimation of these effects. The main
objective of analytical methods has been torsional stiffness rather
than stress distribution in the joints although in some cases stresses
in the cross members have been compared.

1.1.1. The application of Vlasov's theory in joint analysis

Doich [ 2]
suggested a method of calculating stresses and torsional
stiffness in bus chassis frames. He assumed'that a beam with
elastically restrained ends could be considered as an assembly of
several beams with rigid ends. The fictitious lengths of such beams
were calculated from restrained torsion considerations at the joint.
They were employed to determine angle of twist at any section of the
composite beam when subjected to known torsional moment and bimoments
at each end. Equality of angle of twist at any section was considered
as a condition for the composite beam and he obtained an equivalent
uniform average cross section beam to replace the composite one. The
equivalent beam dimensions were used to calculate stresses/angle of
twist in the beam. But the fictitious lengths of the elemental beams
can only be calculated if the coefficients of warping restraint at both
ends of a composite beam are known. He has suggested that strain gauge
measurements at various points on the actual frame should be made to
determine these coefficients. His method can be useful only for
existing frames where such measurements are possible. Gel'fgat and
Oshnokov [ 3] torsional
proposed a simplified method of calculating
stiffness by considering the equilibrium of bimoments in a joint.
However, they assumed the joints as rigid ones which does over-estimate
the bimoments near the joints and does not help towards an under-
standing of the interaction between side members and cross members.
Hanke [ 4] analysed a ladder frame joint consisting of a channel section
cross member symmetrically attached to the web of a channel section
side member. He considered that in a perfectly coupled joint (i. e.
rigid) the total work done by the stresses induced in the joint due to
torsion was zero. He included the work done due to bimoment stresses
in this equation and obtained a condition of bimoment equilibrium at
the joint to that proposed in [ 3]. The angle of
which was similar
twist per unit length of all members endin in the joint was assumed
to be equal. He used the equations [l]to calculate the distribu-
of
tion of bimoment and the angle of twist etc. for each member of the
joint. He introduced an experimentally determined coupling coefficient
to deal with joints with imperfect coupling to include the degree of
warping inhibition. He has suggested that coupling coefficients
should be determined experimentally by measuring the angle of twist and
bimoment in various members of an isolated joint. Zaks [5] investi-
gated the effect of cross member end warping loads on the side member
in a joint from the consideration of equilibrium of bimoments in the
two parts of the side member assumed to be connected to each other in
the joint. He derived a bimoment equilibrium equation at the joint in
a more general form including the asymmetry of the cross member attach-
ment to the side member. He also assumed a perfect kinematic coupling
between various members meeting in the joint. This is another name for
rigid joints which in actual practice and due to local
seldom exist
 3

deformations of the webs and flanges of the various members the

kinematic coupling does not remain perfect. Zaks used the coupling
[4] to overcome this difficulty. He has also
coefficients of
included the length of the side member in the derivation of the
bimoment equation. The coupling coefficients are affected by the
length of the members if the beams are very short otherwise they
for joints. Zaks Belokurov [ 61
are constant particular and
found that warping in a joint was also dependent upon the direction
in which twisting moments were applied to the cross member and side
member ends in the joint. They investigated the case when equal
torques of the same sense were applied to the cross member and side
member ends in a joint, the warping produced was three times greater
than in the case in which torques were applied in opposite sense to
each other. They concluded that a soft joint would result when the
members were subjected to torques of the same sense while a stiff
joint would result in the other case. The automobile frame joints
would behave as "soft joints" as torques of the same nature were
induced. Warping of any member in the joint depends on the inhi-
bition offered by others which must depend on the displacements
produced due to the applied loads. They found that joints in which
cross members were attached with vertical webs behaved in a much
stiffer manner. compared- to the joints where cross member webs were attached
horizontally. Zaks and Belokurov L7 ] further invesitgated the effect
of the cross member web orientation in the joint on the torsional
stiffness of a ladder frame consisting of I-section cross members.
They found that the torsional stiffness of the frame increased as
the angle between the cross member web and the longitudinal axis of
the side member was increased but the stresses increased too. They
proposed that this was not a proper method of controlling the torsional
stiffness. Instead they suggested a method to increase frame stiff-
ness by using cross members of partly closed and partly open section
members. Closed sections towards the ends and open section in the
middle of-the cross members were. used. Such composite section cross
members were used in a ladder frame and they found that the value of
the torsional stiffness was five times greater than with open section
cross members. High stresses, were found at the sections where open
and closed sections of a cross member joined. Zaks and Belokurov
8]investigated another composite shape of cross members in a ladder
frame with Z-section side members. The middle part of the cross member
.
was closed in this case. This resulted in a soft joint.

Zaks, Zaharov and Belokurov [ 9] reported that, in the case of short

beams of open sections, Vlasov's [1] of zero shear at the
assumption
mid plane was not valid. In short cross member beams the effect of
mid plane shear must be included in the calculations. Zaks and
[10] further investigated
Zaharov the effect of mid plane shear in
short open section beams and observed that the end conditions of a beam
produced more effects on displacements than stresses. Neglecting dis-
placements due to mid plane shear in the case of short beams led to
a big difference between experimental and. theoretical values. The
ratio-of beam length to beam depth was kept very small (i. e. only-0.5)
in their analysis.

1.1.2 The application theory in joint-analysis.

of plate

Zaks [ii]
employed the equations of classical plate theory to analyse
 4

stresses in an automobile chassis joint consisting of a channel

section side member and a symmetrical I-section cross member.
He made a comparison of the results of this analysis with the
results calculated using Vlasov's thin walled beam theory. Both
results were then compared with experimental values for various
locations of cross member web on the side member. He found that
in every case results predicted by plate theory were more accurate
in the region of the joint than those obtained by Vlasov's theory.
Both methods showed a good comparison in regions away from the joint
which were not affected by the localised affects of the joint. Zaks
concluded that the plate theory gave more accurate results than
Vlasov's theory if stress distributions were required in the region
of the joint. Both theoretical approaches employed by him gave
proportional values in terms of cross member flange end warping force
intensity, The value of warping force intensity in turn depends on the
amount of warping inhibiton which he measured experimentally. Then
he applied plate theory for a symmetric I-section cross member where
flange warping loads were antisymmetrically distributed on the side member
and there was no contribution from cross member web. It will be shown
in the present work that plate theory can be applied for a channel
section cross member to estimate stresses in the side member as well
as the cross member which Zaks has ignored because warping force in-
tensity was not estimated analytically. It seems that further use of
plate theory in the joint analysis for a channel section cross member
has not been undertaken because of unsymmetry of loading and distri-
bution of the cross member web warping load. Kobrin, Kilimnik and
[12] developed by Zaks [111
Titov used the equations of plate theory
to investigate the state of stress of chassis frame side member walls
at the points of load transfer. They also restricted themselves to the
analysis of a symmetric problem and obtained good agreement with
experimentally measured values.

1.1.3. Joint analysis using simple models

Takahashi [13] invesitgated the torsional of truck frames by
strength
using simple idealised models. He observed that when a channel section
cross member was attached to an equal channel section side member
and were joined flange to flange, the whole load is transferred through
flanges only and the webs do not share in the transfer. In order to
idealise this mode of load'transfer he idealised a channel section by
a rectangular frame consisting of four rods joined together. He
demonstrated the load transfer path by using paper models and then
used the. rectangular frame model of a joint to represent this effect.
In the case of full depth cross members which are attached to the flanges
and web, maximum warping inhibition of the cross member takes place.
It is not clear how he ignored the warping force of the cross member
web. It is only possible if it is allowed to warp freely. It is
accepted that the zero warping line on a channel section flange does
not lie in the middle but Takahashi has assumed that it is in the
middle. He has successfully shown that the problem of a chassis frame
with full depth channel section cross members could be analysed in a
simple manner by using a rod frame model of the actual frame. Simple
expressions for torsional, bending and axial rigidities of the rod
were employed to obtain accurate stresses and displacements. He used
 5'

a relatively new method as "Vibration

known Method" with a hand
held extensometer. to measure strains on various parts of the frame.
He obtained a good agrement between theoretical results and measured
values.

1.2. THE APPLICATION OF FINITE ELEMENT TECHNIQUES IN CHASSIS FRAME

ANALYSIS
With the development of computer storage facilities a large variety
of general purpose programs using finite element techniques have
been developed for the analysis of structural problems. Chassis
frames are analysed using these programs and some of the work done
in this field is reviewed below.

1.2.1. Analysis without consideration of restrained torsion effects

Ali, Hedges and Mills [14] type

analysed a chassis structure
consisting of open and closed section beams employing finite element
techniques. A stiffness matrix for a beam element was derived and
a displacement method was used. All frame beams were idealised into
beam elements connected at nodal points. No allowance was made for
restrained torsion of open section beams. Upper and lower limits for,
displacements of various parts of the structure were established and
comparison was made with measured values. This work was extended to
in the structure by Hedges, Norville Gurdogan [15].
compute stresses and
They found that in some cases stress values were over-estimated by
50% as compared to measured values. They concluded that for an
accurate analysis beam section properties must be estimated correctly.
All, Hedges and Mills [16] [14] to-
modified the stiffness matrix of
include inertia terms and applied the modified stiffness. to
investigate in -matrix
[14].
the dynamic response of the frame used From-,
these references it is clear that accurate stress estimation can only
be made if restrained torsion of open sections, is considered in a
realistic manner in finite element idealisations.

1.2.2. Estimation of the degree of warping restraint

Tidbury, Marshall and Roach 171 used a simple analytical approach to

estimate degree of warping inhibition of an open section cross member
by. a channel section side member in the joint of a ladder frame. They
assumed that only the side member web provided a resistance to the
warping of the cross member if the joint fixing was confined to the web.
The side member web was split into independent strip beams. The beams
were further assumed to be simply supported at-the flange/web corner
when a single force due to the warping of cross member flange act on
a strip beam. Linear variations of flange warping force on each strip
beam were approximated by point loads. Displacement was calculated. at
the load point and the ratio of the strip displacement to the normal
force was assumed constant for all the beams. This constant was used
to determine an effective torsion constant of the cross member.
Effective torsion constant for the side member of the frame was calculated
by treating the portion of its web between two cross members as a plate
under torsion separated from its flange. The cross section of the web
was constrained to remain plane at sectio^s where the cross members were
attached to it. The values of the effective torsion constants
of the cross members and the
 6

side members were used in the modified Erz formula to obtain the overall
torsional stiffness of the frame. The value of the frame torsional
stiffness compared well with the measured Awudu [18]applied
values.
the strip beam approach on isolated joints to'check its validity.
He obtained a good correlation between the strip beam and experi-
mental values in respect of percentage warping for large sections but
for sections with small flanges he found that their torque
experimental
deflection curves fell outside the free warping curves. The strip
beam approach was the first effort, to the best of the author's
knowledge, to estimate the degree of warping inhibition in a joint.
Obviously a very simplified approach cannot fulfil the rigorous
theoretical considerations but in certain practical applications it
provided an accurate estimation of overall frame torsional stiffness.
It has its limitations and does not work when the cross member web is
attached parallel to the longitudinal axis of the side-member [19].
Tidbury [19] has derived for the angle of twist
a general expression of
an open section beam in terms of end bimoments when the degree of
partial warping inhibition at its two ends was not equal. He has also
pointed out many practical cases where certain forces act on a chassis
frame to introduce bimoment stresses in the side member besides
twisting ' the frame. Azodi [20]has investigated the'effect of a
longitudinal force on a channel section side member to which an actual
chassis frame would be subjected due to the braking'force. He found
that in such a case bimoment stresses induced in the side frame were
high'as compared to the bending stresses. Megson Alade [21]
and
estimated the warping inhibition of a cross member by the side member
using finite element approach. Plate and membrane elements were used
in the idealisation of a joint between a channel side member
section
and a channel section cross member. The side member web was assumed
to be a rectangular plate simply supported along all the edges. Wagner
torsion bending theory was modified to include the effect of partial
warping inhibition in a joint. It was assumed that partial warping
displacement of the beam end was proportional to its free warping.
The validity of this assumption was verified by finite analysis
element
of the joint. Theoretical and experimental values for stresses and
angle of twist on the cross members of isolated joints were compared.
Good agreement was obtained for stresses
except when the value of the
cross member warping restraint factor than 0.44 [22].
was smaller
"Warping inhibition of the side member due to the cross member was
estimated by applying the moment couple [23] and
method of effective
values of torsion constants of the side and cross members were used in
an automatic structural program to estimate the overall torsional stiff-
ness of a ladder frame. Theoretical and experimentally measured values
compared very well. It has been shown in the present work that the
cross member warping restraint factors [17] [211 are directly
of and
related to each other as in an elastic system displacement will be
always be proportional to the applied force. The linearity of warping
displacement and warping force has always been assumed in all the
previously mentioned researches in [21]. The
except
values [211
of cross member warping restraint factors were found by finite
element method and must be accurate in comparison to the strip beam
approach which is a very simplified approach. It is not clear from
[22], how the side member warping restraint factors were evaluated.
The automatic structural program used for determining the torsional
stiffness of the frame required effective torsion constants of cross
 7

members and side members of the frame. as an input to the computer.

[22] is in the analysis of ladder frames. and graphs
a valuable work
showing values of cross member (Channel and I-sections) warping
restraint factors are very useful for practical purposes.

1.2.3. Stiffness matrix including restrained torsion terms

Ivanov [24] derived a stiffness matrix for a beam element with four
degrees of freedom including warping. He compared the finite element
results of displacements and stresses with analytical results using
Vlasov's theory and observed only 1% difference. He applied this
stiffness matrix to analyse an actual truck frame. In the finite
idealisation of the frame, beam elements were placed along
-element
shear centre lines with nodes at the intersection of shear centre
The frame joints were
axes of the side members and cross members.
assumed rigid and he obtained good agreement between the theoretical
and measured values. It is not certain whether the joints- used in
the actual frame behaved as rigid joints. Ivanov
[25] [24] to trailer frames
used the stiffness matrix of analyse semi
for torsional and bending loads applied separately. The frame
consisted of channel section cross members which were placed
unsymmetrically on the webs of I-section side members. Because of
the eccentricity of'shear'centre axes and also between the shear
centres and centres of gravity of the side members and cross members
it was not possible to calculate end loads from the computed values
at nodal points. Further simplifying assumptions had to be made to
overcome this problem and stresses did not compare well with the
measured values as compared to the displacements. Flexibility of
the joint was not accounted for in this analysis. Romanov, Slomka
and Sablocki [26] included additonal terms in a general stiff-
also
ness matrix to account for the effect torsion in the
of restrained
analysis of commercial vehicle chassis frames of open section profiles.
Beam elements were used in the finite element idealisation of the
frame and a displacement method was employed for analysis. It was
assumed initially that the frame joints were rigid and nodes coincided
with the intersection of the shear centre axes of side members and
cross members. It was further assumed that the beam cross section
was deformable. They found that real torsional stiffness of each
node was the important factor in the accurate estimation of overall
frame torsional stiffness. An coefficient was used in the
elasticity
analysis at each nodal point to account for the flexibility of the
joint. These coefficients were experimentally measured from tests on
isolated joints. Beeman [27] included the elasticity of nodal points
in the finite element analysis in a slightly different form. He used
triangular elements in the nodal areas with 5. degrees of freedom per
node thus 15 degrees of freedom for every element. He employed
displacement compatibility conditions at the nodal points. The effective
lengths of the cross members were used in the analysis. He. has men-
tioned that in some practical tests warping restraint factors of
greater than one were obtained.

The difficulties involved in the stress analysis of chassis frames

by finite element techniques using beam elements are due to the follow-
ing points:
 8

(a) Beam elements are to be placed along the shear centre

axes of the beams with nodal points\at the intersection
of shear centre axes. In cases where cross members are
not placed symmetrically on the side member web it cannot
be achieved and as a result of this bimoment equilibrium
of the joint is not maintained.

(b) Because of the arrangement suggested in point (a),

eccentricity exists between shear centre axes and'centroidal
axes. of the beams except for I-section beams. Because
of this eccentricity bending stresses are not predicted
accurately. It is reported because of this discrepancy
that some results are obtained accurately with respect to
shear centres axes as reference axes while in some other
cases centroidal and mid plane axes could lead to better
estimates.

(c) Elasticity of the nodal points require experimental tests

to determine the degree of warping inhibition in a joint.

1.3 OTHER APPROACHES

Kobrin [281
and Gorbonos carried out a statistical analysis to
determine the optimum values of design parameters of side rails in
truck frames. The analysis was based on the information collected
from the existing design of truckframes used by various major
industrial plants of the world. The design criteria for the side
rails was found to be bending due to payloads and the channel section
was the usual choice. They found the optimum ratio between the
flange and web height to be around 0.3. Torsional stiffness and
minimum weight requirements were the major factors-besides payloads
that controlled the-values of the section modulus. The authors
intended to carry out a similar analysis of the existing designs
of truck frames to establish optimum values for the cross members and
their joints with the side members. Such an analysis will be, very
useful in the design of chassis frames.

Lasevich, Skolnikov Podlegaeva [291 a method for the

and proposed
selection of optimum sections for commercial vehicle chassis frames.
They used the maximum allocable direct stress and the minimum weight
as the criteria in the selection of optimum sections. The maximum
total stress in a section was the algebraic sum of all effects
including the bimoment stress and this relation was used to determine
appropriate dimensions of the section concerned. They found since all
the stress contributing factors were functions of the section geometry
except bimoment which was also a function of frame geometry, so the
algebraic equation could not be directly solved. They used an iterative
calculation procedure on computer for solving the equation and compared
various sections on the basis of minimum weight. They found even
neglecting bimoment stresses in the first approximation led to eroneous
results.

1.4 APPROACH SELECTED FOR THE PRESENT ANALYSIS

It has been shown by the literature survey that the problem of stress
 9

analysis of chassis frame joints needed further investigation.

As the method of attachment of the two members varied from joint
to joint it was not possible to develop an analytical model which
could take care of all the designs. Since a ladder type frame
consisting of channel section side members and cross members with
welded joints presented a fundamental problem, it was decided to
investigate the problem of stress estimation in welded joints. It
was intended that such an analysis would be useful in the design of
chassis frames as it would provide an information regarding stress
distributions in various parts of the joints.

As the'purpose of the present investigation was to determine stress

distributions in the frame joints so it was necessary to use an
analytical approach which could give good estimation of stresses
preferably without experimental assistance. It has been shown by
Zaks [111 that the plate theory gave better estimation of stress
distributions in joints consisting of a symmetric I-section attached
to the web of a channel'section, than Vlasov's theory of thin walled
beams. It has been also reported, in many papers that for'unsymmetric
joints the condition of bimoment equilibrium was violated. The plate
theory does not possess sucha limitation and can be applied to
asymmetric positioning of the cross member in a joint. Vlasov's
theory of thin walled beams is-based on the assumption of non-deforma-
bility of the-section shape'and the absence of shear strains from the
middle-surface of the'section. - Since the deformations in the vicinity
of ajoint are affected by localised effects so it is doubtful that the
assumptions would be valid in the joint. Because of these considerations,
it was decided to apply the plate theory to analyse a welded joint
between two channel sections in a general manner.

It was also decided to estimate the degree of warping inhibition in a

welded joint analytically so that the-quantitative estimation of
stresses/displacements in such a joint was possible.

As it is necessary to verify the results of any proposed analytic

approach, it was decided to use photoelastic method of experimental
stress analysis for such a verification.
 10

CHAPTER 2 THEORETICAL INVESTIGATION

2.1. BRIEF DESCRIPTION

I
The state of stress in a joint of a chassis frame of ladder type
construction largely depends upon warping inhibition. High values
of localised stresses develop in the region of a joint which can
weaken the joint severely if bracket mounting holes are present in
the same area. It is important to locate areas of high stresses
in the joint for the overall safety of a chassis frame.

The problem of a joint consisting of channel section members with

crossmember attached to the web of the side member is theoretically
analysed in this chapter. The theoretical-model of a channel section
sidemember is assumed to be made by joining three flat plates. It
is subjected to the external forces which result from the warping of
a channel section cross member due to applied torque at its free
end. The resulting stresses/displacements in the side member are
analysed by employing the equations of classical plate theory.
The warping restraint factor for the cross member has been evaluated
from the present analysis. In order to obtain stress/displacement
distribution on any part of the side member, a computer programme
was written using the results of the present analysis and results
were computed for some cases. The programme is in'Basic language
and can be run on HP9830A. model.

2.1.1. Sequence of calculations

The sequence of the mathematical formulation of the problem and

its analytical treatment is described below.

1) The general equations of classical plate theory have been

employed to express results of plate bending and plate stretching
problems in terms of initial parameters of a plate. The forces/
displacements and moments etc., are assumed to be single trigono-.
metric Fourier series forms to satisfy simply supported boundary
conditions along the two parallel edges of plate, i. e. xl =0 and
and x= 2L in Fig. (2.1a). a channel section side member has
considered -Since
been to be made up by joining three Plates along their
edges, its analysis becomes easy if the resulting equations of plate
bending and plate stretching problems are presented in such forms
that can be applied directly to the junction of plates. It is
advantageous to express the resulting, equations of plate problems in
terms of the initial parameters of a plate defined along one edge
of the plate. Then stresses/forces/moments on any part of the plate
can be obtained if these initial parameters are given specified
values, subject to the boundary conditions along the edge at which they
are defined originally. The equations for plate bending and plate
stretching problems are expressed in Appendices I and II in terms of
these parameters. Zaks [11] has used a similar approach to express
 11

'0 X,

x2 x3

T
Fig. 2. la A rectangular plate referred to a Cartesian
co-ordinate system

`1'2.4

ly,
u

i X3 rn,i
mil -C-
mu
mu. ,.
i T,
i m22

'Aa
Fig. 2. lb Forces and moments on plate element subject
to bending loading.
XI

nzz
x x3

)1»
' nn

p n=3,
Fig. 2. lc Force components on a plate element subject to
stretching.
 12

the resulting equations in terms of the initial parameters of

a plate. But his results are only applicable to a plate problem
if the external loads are antisymmetrically applied to a plate. The
form of equations (AI. 23) and (AII. 19) of Appendices I and II is
such that they can be directly applied to a plate problem for any
arbitrary loading. These-equations express the Fourier coefficients
as shown in the Appendices and they do not include
of various series
any term for external forces. Any external load acting on a plate
has to be expanded in a single trigonometric Fourier series. The
form of this series is determined from the relev ant form of the
internal component of the force/moment acting in the plate. The
Fourier of this series is included in equations (AI. 23)
coefficient
and (AII. 19) to satisfy boundary conditions of the problem. Once
the values of Fourier coefficients are obtained then they are used
in the respective Fourier series to obtain the values of the desired
quantities.

ii) In section. 2.2, the side member web of a joint has been
considered as a rectangular flat plate with all edges simply
supported. This plate is. subjected to the cross member end warping
forces. The effect of cross member flange warping forces and web
warping forces has been analysed separately. The warping forces
by distributed forces as shown in fig. (2.2a). Distri-
are represented
bution, of these forces on'the side member web is given by equations
(2.2), (2.3) and (2.18).

Simply boundary conditions are expressed by equations (2.4)

supported
and (2.5) along the four edges of the side member web. Since the
web will bend due to the external warping forces, all the equations
of the plate bending problem of Appendix I are applicable to this
problem. Boundary conditions along xl =0 and xj = 2L expressed by
equations (2.4) are automatically satisfied as the assumed form of
u3(xl, x2) as given by equation (AI. 10) ensuresthis. The conditions
expressed by equations (2.5) are to be satisfied along the
other two parallel edges of the plate i. e. along x2"= 0 and x2 = 2H.
Since the initial parameters=as, employed in Appendix I are defined
along the edge x2 = 0, so, the equations (2.5) along x2 =0 for the
problem under consideration will result in equation (2.7). This means
that out of four initial parameters, there are only two non zero
parameters for this problem which are to be evaluated from the
(2.5) at x2 = 2H. The external warping forces must also
equations
be taken into account while satisfying the boundary conditions. As
warping forces and resultant vertical shear forces v2 are acting
identically on the plate so the warping forces must have an identical
form of Fourier series as V2. Equations(2.10), (2.11) and (2.23)
represent Fourier series forms of warping forces of the top flange,
bottom flange and web of the cross member respectively. Their res-
Fourier by equations (2.12) and (2.24)"
pective coefficients are expressed
The contribution of these external forces is considered by adding
respective terms in the equations (AI. 23) while applying them to the
boundary conditions (2.5) at x2 = 2H. The resulting form of these.
boundary conditions after necessary substitutions is given by equa-
tions (2.9) for the flange warping forces, and equations (2.22) for
the web warping forces.
 13

10

i
R-
L

LL

Fig. 2.2a Side iember web subject to the cross member

end warping forces

/H
/
Fig. 2.2b Side member web subject to the cross member
web warping forces.
 14

The unknown parameters are calculated from equations (2.9) and

(2.22). Once these parameters are known then their values are
used to calculate force/displacement/moment at any part of the
side member web by equations (AI. 23) and respective Fourier series.
The combined effect of flange and web warping forces can be. evaluated
by super position.

iii) A channel section-is treated as a joint between three rect-

flat in Section 2.3 as-shown in fig. (2.3a). Since
angular plates-,
warping loads will not only produce bending but also stretching of
the plate elements of a channel section side member, so this problem
is analysed by combining the bending and stretching problems. Eight
initial parameters are specified for each plate, four of-which
are used in the bending problem and the rest are used in stretching
problem. So in general, there are twentyfour parameters for a
channdl section. In fact, only eight of them are independent.

The relationships between the initial parameters of the web and the
top flange are governed by-boundary, conditions along the common
boundary AA as shown in fig(2.3a). Equations (2.27) represent these
relations while equations (2.28) represent relations between the
initial parameters of the bottom flange and the web of the channel
Zaks_. [ll] identical relationships. By using these
section. used
relations it is possible to express forces/displacements/moments etc.
on any part of the channel section in terms of initial parameters
of the top flange or bottom flange or web only. Therefore the
whole problem of determining forces/displacements/moments etc.
in a channel section side member is reduced to the determination of
suitable values of these eight parameters from the boundary condi-
tions.

iv) The effect of cross member warping loads on the side member
is analysed in Section 2.5 for the -joint -shown in fig. (2.3b1 The
flange warping loads are considered in a similar way as described
earlier in step(ii)whiletheweb warping loads are replaced by
various point loads which are applied to-the side member web so that
the total warping effect remains 'approximately unchanged. The
number of point loads for each particular, joint is determined by
computing their effect on-stresses and displacements in the side
member and increasing the-number of point loads until the difference
between successive results is less than 1%.

Starting from stress free boundary along the edge of the

conditions
top flange, it is inevitable that some o£ its initial parameters
will be zero as expressed by equations (2.33) through equations (2.32).
If the joint is symmetrical' i. e. T2 in fig. (2.4a). The rest of the
-T1=
unknown initial parameters can be determined from the boundary
conditions along the horizontal axis of symmetry x x, as expressed
by equations (2.30) and. (2.31). These equations are expressed in
terms of the unknown parameters of the top flange by equations (2.35)
which include appropriate terms for external loading. Solution of
these equations provides the values of the unknown parameters which
are used in conjunction with the equations of Appendices I and II,
to evaluate stresses/displacements at any point of the side member.
 15

Section wall thickness =h

Fig. 2.3a A channel section referred to the cartesiän co-Qrdinate

systems

Fig 2.3b Joint under investigation

 16

In the case of an asymmetrical joint i. e. T1# T2in fig. (2.4a)

the unknown initial parameters of the top flange are determined
from the boundary conditions along the free edge of the bottom
flange which are given by equations (2.43), (2.44) and (2.45).
These equations have to be written in terms of the initial para-
meters of the top flange by using relations given by equations (2.46).
The resulting equations are solved for the unknown, parameters. The
values of these parameters are used in the equations of Appendices
I and II to determine stress or displacement at any point of the
side member.

v) Section 2.6 deals with'the effect of the couple which is

transmitted through the, -cross member end' to the side member web, on
the stress/displacement distribution in the side member. The magni-
tude of the couple forces is determined from the applied torque onthe
free end of the cross member as expressed by equation (2.48). This
problem has been solved for a general case where T1 # T2 in fig. (2.4b)
The unknown initial parameters of the top flange remain the same as
in step(iv). They are determined from the boundary conditions along
the free edge of the bottom flange in an identical manner as descri-
bed in step(iv). Only the external loading terms change their values
which are calculated by using equation (2.53).

2.2.,. BENDING;, OF-A SIDE MEMBER WEB DUE TO CROSS MEMBER WARPING LOADS

Consider a joint of a channel section cross member attached to the

web of a channel-section side member. The cross member web is placed
vertical to the longitudinal axis of the side member. In this case the
warping of the cross member end is restrained by the side member web
only. A simple theoretical solution can be developed by idealising
the side member web as a flat plate simply supported along all the
four edges. Such simplification has been used in the estimation of
partial warping inhibition in joints 17,21]. This problem can be
reduced to the calculation of the deflection of such a plate subject
to the cross member end warping loads.

The mid plane of a simply supported plate referred to xlx2x3 cartesian

axes is shown in fig. (2.2a). The distribution of the cross member
end warping loads is assumed proportional to the free warping distri-
bution (or law of-ysectorial areas). This assumption is logical and
has been used directly in 11,17].
These external loads can be represented by distributed
forces placed along the cross member end section.

A complete solution of this plate bending problem can be obtained by

super position of two solutions obtained by solving separately for
cross member flange warping loads and cross member web warping loads.

2.2.1. Cross member flange warping loads

The maximum force is acting at the flange end i. e. at

xi =L+S

p=z

kS
0
(2.1)

where kois the intensity of force.

 17

Fig. 2.4a A channel section side member subject to a channel

section cross member end warping forces.

Fig 2.4b A channel section side member subject to a couple

loading.
 18

Distribution of the external forces is given by the following

equation:
O<x1 <(L-R)
o
for
P(xi) and (L+S)< xi<2L (2.2)

ko(xi -L ) for (L-R) < xl <(L+S)

Let pi(xl) and pii(xl) be the forces/unit length in the top and bottom
flanges of the cross member respectively.

Also pi(xi) PII(xl) (2.3)

Simply supported boundary conditions along the edges of the

plate can be written in the following form.

At xi =0 and xl = 2L

u3(xl, x2) =0)

(2.4)
Ml l (xl, x2) =0)

At x2 =0 and x2 = 2H
u3(xl, x2) =0)
(2.5)
m22(xl, x2) =0)

Results developed in I can be directly

Appendix applied to the
under consideration. Equation (AI. 10) satisfies the
problem
boundary conditions given by equation (2.4). Equations (AI. 10)
and (AI. 15) should satisfy the conditions given in equations (2.5).
On substitution in equations (2.5), the following conditions are
obtained

At x2 =0 and x2 = 2H
Wm(x2) 0)
(2.6)

M2m(x2) =0

Equations (2.6)
should include the respective loading terms for
cross member flange warping forces. Conditions at x2= 0 imply
that following initial parameters are zero.
0
W=0)
(2.7)
0)
M2 0)

The other two conditions at x2 = 2H give the'following equations:

 19

9+3
+ aD 32 V+P c-. - ý23ý.
am Im
aa a3 a2
mmmm

.. n
+pIIt P-2- al -0 (2.8)
3 a2
am m

[TD2 11

+ g.D3 6
'+
1! + aDg
0
V2 + PIM bä 11
+ aD3

2+ 0 -. ý (2.9)
+ PIIm aD3 _.

where

D1, D2, D3 and D4are the values of the hyperbolic functions

"1.02.03 and 04 determined at x2= 2H

Di D2 D3 and D4 x2-(2H-T1)
, ,

D" D" D" and D" of it it if It to XT

1' 2' 34 2ý 1

PIm and PIIm are Fourier coefficients of the external force

expansions of the following form:

PI PIm(x2) sin amxl (2.10)

L
m=1

pII PIIm(x2) sin amx1 (2.11)

M-1

' Series expansions. (2.10) and (2.11) are obtained from the fact
that the external forces are acting parallel to the resultant shearing
force vz and hence the expansion should have the same form. Coeffi-
PIm in the following
cients and PIIm can be evaluated way:
(L+S)
1Jp
P -P IIm sin amxidxl
Im L
(L-R)

(L+S)
1k (L-x ) sin ax dx
Lo1m11
(L-R)
 I

20

ko Is ina (L+S) -sins (L-R) I- (L-R)

= S cos 'a
am mmm
La
m
-R cos am(L-R) (2.12)

Equations (2.8). and (2.9). represent a system of two simultaneous

equations involving the only two non zero unknown initial para-0
9 V2 Once the values
mete V2s of the problem which are and . of 6
and are found, it is-easy. to determine the deflection through
equations (AI. 23) and (AI. 10) at any point of the plate. Equation
(AI. 10) has to include the external loading terms and is expressed
in the following form:

(i) When the point lies between oiý x24T1

o0 ý- j2 1V2
C« C ý+. (2.13)
u3(xlßx2) =mLl aý31 + sin amxi"'
a3 a
in in

(ii) When the point lies between T1<x2 (Tl+co)

. -<
CO
0
[b. ±31V2+C[±2-$31pImsinax2
u (x, x) _I _2+aý31+ßt2_
312 am in
m=l a3 a2 a3 a2m
(2.14)
(iii)When the point lies between (Tl+'co)<x1, <2H

u3(xlsx2) =G
m=l I[2 m.
+ ag3J0 + ['
a3 a2
mmm
V2+ & [(
a3
--
a2
)-.

L3 ) ']
-12 - P sin ax (2.15)
Im m1
a3 a2
mm
where
42
and are the values of the hyperbolic functions 02 and 03
respectively calculated at x2=(x2-T1)

2 and 3 x2=(x2-T1-c0)

2: 2.2 Cross member web warping loads

Since all equations developed in Appendix I have been expressed in

single trigonometric Fourier series along xl direction, so the
results can be directly applied to such problems where the external
forces can be expanded along x1. The cross member web warping forces are
varying along xi 1cording to fig. (2.2a), hence it is necessary to interchange
the direction of co-ordinate axes as shown in Fig. (2.2b). Now the results
of Appendix I can be directly applied for this case.
 21

The maximum force is exerted at the ends of the cross member

web and is given by

=2 koco (2.16)
PC
0
where ko is the intensity of force along the cross member web.
Relationship between ko and k* can be expressed as follows:

2 ko 0= 0R

2R
k* k (2.17)
or o co o
Distribution'of warping forces, on the plate is given by:

0 for 05x15T2 and (T2 +c )<x1-2H

p*(x1Y =*o (2.18)
ko(xl-T2-2°) for T2 : x1. (T2 +c )

Boundary for the simply supported edges of the plate

conditions
can be expressed in the following form:

At x1 =O and. x1 = 2H

u3(x1, x2) = mll(xl, x2) =0 (2.19)

and at x2 =0 and x2 = 2L

u3(x1, x2) m22(xl, x2) =0 (2.20)

The solution of this problem can be obtained in a similar

manner as alreadydescribed for the cross member flange warping
loads. Following resulting equations are obtained by satisfying
the boundary conditions (2.20).

W.
0)
(2.21)
d)
M2 =0)

b1+ [a312-F3 lp2

aF3 100 + + P* I-i=0 (2.22a)
L am JJ m LL
. a22 a3 a2

äF2+ [bF
+ SF3] 00 +I 2+ (2.22b)
71 aF3] 2+ p* aF3 J=0
Lmm

where

F2 and F3 are the ; values of the hyperbolic functions 42 and ¢3

respectively are determined at x2=2L

F2 and F3 x2L+R
 22

P* is the Fourier coefficient of the following equation: '

p J= Pm (x2) sin amxi (2.23)

m
The value of Pm(x2) can be evaluated from equation (2.23) in
the following way:
(T2+co)
ni P* sin
HJ amxldxl
T2

(T2+c
o)
1
=H ko(xl- T2-co)sin amxidxl
T2

or
[Hc 2a
PM ko sin cosa (T2 +-
bmm2 a

a2C0)(
CO COS( os
am(T2 +2-So )- (2.24)

Equations (%. 22a v.nd b) are solved for the unknown initial
parameters 0 and Their values are substituted to obtain the
following equations for the deflections of any point on the plate.

i) If the point lies between 0-<x25(L-R)

°C [3]0 öo Fý3
u3(xl, x2) aI + - V2 sin amxi (2.25)
m=1 m a2
mm,

ii) If the point lies between (L-R)<x2., <2L

" -2+ 0 ý-
u3(xl, x2) = G. aý3 0+ 30
am V
mil ag a2
mm

ý-
+E -P sin a" x (2.26)
2mm1
aM3 Cc!
"
where

F2 and F3 are the values of hyperbolic functions 41 and 42

calculated at x2 = x2-(L-R).

Deflection the plate

of (i. e. side member web) can be determined
at any point by simple algebraic addition of the two results of sections
2.2.1. and 2.2.2. This analysis does not calculate the stress
distribution in the plate.
 23

2.3. APPLICATION OF PLATE THEORY TO THIN WALLED BEAMS OF

OPEN SECTION

It has been shown in section 2.2 that side member web deflections
can be calculated when it is subjected. to the cross member warping
loads. Flanges of the side member were not considered in the
previous analysis. Only bending of the side member web was con-
sidered, while in the actual case, the side member can deform in
other planes as well. Deformation of the complete section can
be estimated by employing the equations of plate bending and
plate stretching problems together.

Stress, strain. and, displacement distributions of a side member

- .
of open section subject to warping loads can be analysed using the
equations of plate theory. It should be possible to consider such
an open section made by joining various plates along their edges.

Side members made of channel sections are very common in commercial

vehicle chassis frames. So the analytical treatment is developed
for a channel section side member. A channel section shown in fig.
(2.3a) can be considered to be made from three flat plates of
dimensions indicated. They 'are joined along their edges to form
a channel section. Angle of inclination between two adjacent plates
is taken as 900. Co-ordinate systems chosen for each plate are shown
in fig. (2.3a). '. Mid plane of each plate the reference
provides
for the co-ordinate system.

Equations and (AII. 19) can be applied

(AI. 23) to any plate for an
arbitrary external loading. Stresses/displacements and moments of any
plate can be determined if its eight initial parameters are known.
These parameters are the same as used in Appendices I& II. Hence
it can be seen that for each plate a system of eight simultaneous
equations with eight unknowns, (inital parameters) is obtained. -
It
is possible to find out the relationships between the initial para-
meters of the adjoining plates by satisfying the conditions along
the common boundary between them. In this way, the number of un-
knowns does,. not. increase in the, system and adequate number of equa-
tions are available to determine the unknowns.
_
V2
Let U. V, N22, N12, W, 0, M2 and bÜ týe jnitja1 paöa1eteandrs pf Platel
which are defined at x2 = 0. 22
,, 2,12, ,,
are the initial parameters of, Plate 2 at 2= 0. Following relations
are obtained from the common boundary conditions.

U um (at x2 = F1) )

V =W (n n) )
* )
N22 =V (n n)
2m
)
N12 N ( ýý n) )
12m
)
*
W = -V (n n) )
m
 24

0 0m (at x2 = Fl) ) (2.27)

*)
H2 = M2m (-n ") )

v2 = _N22 (n))

The right side of equations (2.27) can be determined from the

equations (AI. 23) and equations (AII. 19)on substitution of the
initial Plate 1. Hence equations (2.27) provide the
parameters of
relationships between the initial parameters of Plate 2 with the
initial parameters of Plate 1.

Similarly relations between the initial parameters of Plate'3 and

Plate 2 are obtained from the following equations.
** x2=
U .. = Um (at ` 2H) )
** )
V= Wm ( 11 11 ))

N22 V2m ( 11 n))

)
**
N12"'-'= N12m ( ; -11- 11-)) (2.28)
*)
w=- vm ( 11 11 ))

** 11
)
o=0( "))

m)
M2 = M2M ( It 11 ))

** ))
V2 ( 11 11 ))
-N22m

** ** ** ** **** ** **
Where U, V, N22, N12, W, 0, M2 and V2are the initial parameters of
Plate 3 defined at **
2 = 0.
ifF

Now on simple substitution of equations (2.27) in equations (2.28),

"relations between the parameters of Plate 3 and Plate 1 can be
obtained. Hence the problem of a channel section can be analysed by
determining the eight unknown parameters.

Beams of other open sections such as top hat and Z-sections can be
treated in a similar way. It is necessary that the end boundary
conditions of-the beams are hinge type, otherwise the results of
Appendices I and II are not applicable.

2.4. ANALYSIS OF A JOINT

The effect of a twisting moment applied on the free end of the cross
distribution in the joint in
member cn the stress side member of a shown
fig. (2.3b) has been analysed by using the approach suggested in section
2.3. The external load produces warping of the cross member section
thus exerting warping forces on the side member web. It also produces
a moment in the plane of the side member web. The stress/displacement
 25

distributions in the side member will depend upon the following:.

i) Cross member warping forces

ii) Moment acting in the plane of the side member web.

Separate analytical solutions are-developed-for these two loadings
in the preceding sections and their combined effect on the side
member is obtained by super-position of these solutions..

2.5. SIDE MEMBER SUBJECT TO CROSS MEMBER WARPING LOADS

Consider the
case of a joint in which a channel section cross
member is placed with its web vertical to the axis of a channel
section side member. Also the cross member is attached to the
inside of the side member web and the zero warping point on its
flange is located symmetrically along the length of sidejnember,
as shown in fig. 2 ,3b. The cross member has equal flanges.

When the'axis passing through the shear centres of a side member

and a cross member are intersecting each other, the joint will be
referred to as a symmetric joint in the further work.

2.5.1. Analysis of a symmetric joint

According to fig 2.3bthecondition for intersection of axes through

the shear centres of the respective members is fulfilled if:

Tj ° T2 (2.29)

Equation (2.29) that implies

cross member end warping forces are
anti-symmetric about the centre line ix- of the side member web.
Hence the resulting displacements of the side member will be also
anti-symmetric about RR which is an axis of symmetry. For this
reason, analysis of the half portion of side member consisting of
a flange and half of the web is sufficient. The top half of the
side member is considered in this analysis.

Boundary conditions at the line RR can be expressed in the following

way from the conditions of anti-symmetry.

" At X2_= H

üi -0)
)
n22 "' 0)
(2.30)
113 °0)

m22 "0)

Equations(2.30) lead to the following conditions through equations

(AI I. 9) (AI I-. 16) (AI. 10) and (AI. 15).
, ,
 26

At x2 -H

U=0)
m)
N =. 0 )
2m
(2.31)
Wm '=) 0

M2m =0)

Boundary conditions at the free edge of the flange are expressed

below.

At x2=-0

n22 =0 )
)
n12 0 )
) (2.32)
022 =0 )
)
V2 0

Equations (2.32) lead to the following conditions.

000 M2
N22 = N12 = m V2 =0 (2.33)

Initial parameters of the web can be expressed by substituting

equations (2.33), (AI. 23)and(AII. 19) in equations (2.27) by the
following equations

U= [A1 bamA4 JU- [aA2 + bamA3 (2.34a)

+ W

[A 0
V - a' JW ++ aA3J0 (2.34b)
ýýxxm
N22= [A2- &A3]W 8 (2.34c)
am +[öamA3

[fA2 U- [fa
N12 + famAý AIJ V (2.34d)
M
[aA2 [A1- bamA4 3V (2.34e)
W= - bamA3] U-

0 bamA2- + EA1 Jo 0M (2.34f)

aa2A3]W + aamA
[&LmAL]W [TI'A2
142 - + SAJ (2.34g)
V2 [famA4] [fA2
U - famA]V (2.34h)

where

Al, A2, A3 and A4 are the values of hyperbolic functions X1, ¢2,43 and
04 respectively, calculated at x2 = F1 = F2 . Equations (2.34),
 27

(AI. 23)and(AII. 19) are then substituted in the boundary conditions

given by equations (2.31) and the following equations result. after
rearranging the terms:
B2+ 'j 0

[(B1 + bamB4) (A1 + bamA4) + (a

mJ
cB3) (fA2 + °famA3) IU

- [(B1 + baýB4) (aA2 + bamA3)+(

m
+ cB3) (faA 4)1 V

[(aB2 'j0
-+ bamB3)(A1 - aamA4)+(cB4)(nA2 - 6A3)a2{W
J

- [(a112
+ ba&B3)(
2+
aA3)+(cB4)(öamA4)!
JJ
0O =0 (2.35a)
m

]u 0

-
[(faB)(Ai
[(m*"2
+ bamA4)+(aB2 +bämB3) (fA2 + famA3)

+ bamA3)+(aB2 +bamB3)(famA4
0
V

0
+ (fB2 - famB3) (A1 - aanA4)+(B1 bamB4) (nA2
- - 6A3)a2 JW

]00
+ [(fB2
- famB3) (äA + aA3)+(B1
m
- bamB4) (dam
,,
) 0 (2.35b)

3 B3)(famA4)J 'j 0 1(Bi

(B1- aamB4)(aA2 - bamA3) U -aamB4)
am m

3- ä2) 1V0
(Al -bamA4)+ F( (fA2 - famA3) LB
+[(+ aB3)

L) JW+
(BamA2- aamA3)-(F (6amA4) ) (4i1+
. aB aamA4)
ý(Mg+
mm3
ä4) r B3
(nA2 B1 ý. B3P`
+( + 5A3)1 8 +-"g - PIm+ -0 (2.35c).,
1a a2 J3 21 m
mm am am

'j 0 E&1BL+)
damBq) (aA2, bam 3)-(2 aB3) (famA4)] U-
m+
äB2+ ]V- [nB2
(Al- bam 4)+(. aB3) (fA2 - faýA3) +6B3)
m-
[(nB2
(ba k2 aa2A )-(B "aa B4)(damA4) W- + SB3)
l
m2m31m l

o (' 3j
(Al + aamA4)+(B1 + aamB4) (nA2 + SA3)] PIn
+'+ aB
m
]
+ I[2 Pm
+ aBg =0 (2.35d)
M'*
 28

where
B1, B2, B3 and B4 are the values of hyperbolic functions 41,42,3 and
$grespectively are evaluated at x2 =H

B2 and`B3 are the values of'hyperbolic functions ¢2 and 43 respect-

ively evaluated at xZ (H - T1). Also B2 and B3 are evaluated
w. r. t. the positions of point loads across the side member web such
that x2 = the distance of point load from the, centre line RR of the
web.

The Fourier coefficient P1m of the cross member top flange warping
force is evaluated by equation (2.12) and the Fourier coefficient Pm
of the cross member web warping force is evaluated as explained
below.

Let the web warping effect of a cross member be, represented by a

suitable number of point loads. The magnitude of each point load
equals the-sum of distributed forces it has replaced on the portion
of the end cross section. Each point load is located in the middle
of the portion of end cross section upon which it has replaced the
warping effect. If p'is a point load, it can be expanded in a Fourier
series over the side member web as below

P' sin mxl (2.36)

m1Pm
and (L-R+y)
Lim 1 (2.37)
P ,p sins x
m Y-}o L 2y m1
(L-R-y)

where Y is a small, dimension. -.

Deflection of a simply supported plate due to a point load has been

evaluated in [30] in a similar way. Solution of the integral of
equation (2.37) results in the following

P'sin ý Lim,
,_ m(L-R) sinamY (2.38)
Pm
Y-ýo
aL
my

Limit within brackets of equation (2.38) can be evaluated by expanding 10

(sinamy) in power series as:

Lim sins y Lim 1 (ya')3 (ya )5

m]=Y. Y (Yam) - ..ým +B-..
*O .
Y 3! 5..

= am (2.39)

Hence equation (2.38) becomes

p'sin m(L-R) (2.40)

PM =
mL
 29

The magnitude load p, is following

of a point evaluated'by-the
relation:
a2
P" ko sds =o (a2 (2.41)
2 - aj).
al

Where aia2 is the portion of end section of a cross member whose

warping effect is replaced by a point load p'.

Equations (2.35) represent a system of four simultaneous linear

involving four Ö.
equations non zero unknowns which are W and
Their values can be evaluated by solving equations (2.35). Then by
_
substituting the values of these initial parameters in equations (AI. 23)
and (AII. 19), Fourier coefficients Wm, 8m, M2m, V2m', UM, VR, N22m and
N12m are evaluated. The values of these Fourier coefficients are
substituted in the respective series to obtain the values of various
forces/displacements and moments etc. Values of W, e, M and V for
2<H external 2m 2m
Ti of the side member web must include the forces
and are given by the following equations:

W [bam' +aý [ý 4 IM2

as 041 W+ 3] e-
mmaa
mm
±3 131V2
+ +2 ý ]Pý +`'`
-2 D (2: 42a)
am m am am a3m am
m

6m e
[bam W+ [41+ aam 41 +. [ä +q 31M2
2 -aa2mý 31
m

(2.42b)
f m
IV2 C in
]PIm-lt
cLa*
am )Pm

M2m = Isam4lW- [02-`+431* [ý1+

+ aam41M2

+
+ag3IV2'+ [ [bä +a431 Pm (2.42c)
+a43]PIm+4
mmm

V Ina2 ý2 -da2.. ý3ýW ýba ý2 2+31M2

+Idä ýq]e + -aa
2m mmmmm

+ [ei-aam441ý2+[$1-as44jPIm+G4l-aam4]Pm (2.42d)

ee
ýi'42"43 and Ware the values of. hyperbolic functions 41,42'+3 and 4'z+
respectively evaluated at x2 = (ý. 2- Ti) and 41'2' 43' and 4' are
evaluated for point loads at x2 = di (distance of the point load from
the centre of the side member web).
 30

Evaluation of forces/stresses/moments. or displacements on any part

of the side member is done by using the values 8f önitialoparaöeters0
which are valid for that p4rticular i. U, N2? N12, W, $, M2
part, e
V2 *V,
and for the flange and u, V, N22, N129 W, 0, M2 and V2for the
web.
"y

2.5.2. Analysis of an asymmetric joint

Fig(2.4a) represents an asymmetric joint as the cross member web is

not located symmetrically' on the side member web i. e. T1 # T2. In this
case the entire side member section has to be considered in the
analysis.

Boundary conditions at the free edge of the top flange are the same
as those used in a symmetric joint and-are given by-equations (2.32)
and (2633). 0Four non zero initial parameters of the top flange are
e, ', W and 0. They have to be determined from boundary conditions
at"the free edge of the bottom flange which are expressed below:

At x2 J2 )

n22 =0
)
n12* =0) (2.43)

m2 2.0)
)
V2 =0)

Equations (2.43) result in the following equations through equations

(AII. 16), (AI. 15) and (AI. 17)

At X10
2= F2

N22m N12ý "°.= M2m = V2m 0 (2.44)"

.

Boundary conditions (2.44) can be written in extended form by

substituting equations (AI. 23) and (AII. 19) in them. They become

[fam fJ2-famJ3]V [Ji-bam 0 (2.45a)

41 U+[ +
4]N21+LaJ2+bam'AN12
** ** ** *st
rfJ2 + faM U- [farJ4]V [aJ2- baMJ3]N22+. [J1+bam34]N12 0 (2.45b)
+

[ÖamJ41W- [nJ2+s ý6 [ äJ2+ JV2 (2.45c)

+ýJ1 +äaß 4]M2+ aJ3 =0
m
[na2mJ2 6(12
- +[damJ4I0 +[bamJ2 - aal J3JM2'

+ EJ1 - aam 4 IV2 =0 (2.45d)

 31

Where

J1, J2, J3and 74 are the values of hyperbolic functions X1,42,43 and 44
respectively evaluated at x2 = F2.

Equations (2.45) represent a system of four simultaneous linear

equations involving eight unknown parameters which cannot be evaluated
directly from this system. However, the number of unknowns can be
reduced to four of these parameters which are expressed in terms of
non zero initial parameters of the top flange. These relations can be
obtained by substituting equations (2.34) in (2.28) and allowing for
the contribution of external warping forces on the side member web.
The final form of such relations after some -rearrangement is
expressed below:

Ü= [(G1 + bamG4) (A1+bamA4) +1 (äG2+

cG3) (fA2 +famA3)] u
m
[(G1 0
+bamG4) (aA2 + bamA3)+. (äG2+ cG3) (famA4)j V
m
-[(aG2 +bamG3) (A1 -aamAq)+(cG4) (nA2-SA3)a2] W

a0-
-[(aG2 +bamG3) (äA2+aA3)+(cG4) (öamA4)] (2.46a)
m

[(G1 G2)(fa
V= -aamG)(aA2 -bam 3) -ß(G3- Aq)] U
am am m
G3 -2) (fA2-fam V
-[(G1- as G4) (Al - bamA4)+ - 3)1
am am

ä4)(öamAi)] 0
+[ (ä2+aG3)(bamA2-
aa2A3) - (ý W
mm

+[(aG2+ aG3) (A1+ aamAO +( 4) (nA2 + SA3))a0

mm
"+` {1-I]
p (2.46b)
a a-
mmmm

tam (t1G3- 5G3) (aA2- Ü

N= ba*A3)-(G1- aaýGq) (famA4)]

[a2(lG2-SG3) (Ai
- -bamA4)+(Gl- aamG4) (fA2 - faI A3)J V

0
+[iaamG4)(bamA2- aamA3)+ibamG2-aamG A4)I W
3)('a

0
+[(öamG4) (A1+ aaA4)-(bamC2- aamG3) (nA2 +dA 3)]6
m

+[(G1- aamG4) - (G1 - aamG4)] PIm

+F [G1 1Pm (2.46c)

- aam
 32

N12 = -L(fG2+ fa G3) (A1 + ba A4)+(Gl+ ba G4) (fA2 + fä Ag)J U

mmmm

0
- L(fG2+ famG3) (aA2+ bamA3)+ (G1+ bamG4)'ifamA4)J V

0
(A1- aaýA4)- (aG2 - bamG3) (nA2-öA3)a 2
2]W
-ýfaýG4)

" 0
-[(fa G4) (äA2+ aA3)- (aG2- baG 3) (Sct 4) 0 (2.46d)
mmI

W= -[(bamG3-aG2) (A1+ baI A4)+(cG4) (fA2+ faýA3)l u

0
+[(ba G3- aG2) (aA2. + bamA3)+(cG4) (f mA4)
m]V
0
-[(G1 -ba Gq) (Ai-aamA4)+ (dG2- cG3) (nA2- 6A3)am2] W
m am
00
10
-t(G1 - bamG4) (äA2+ aA3) + (äG2- cG3) Oct m 4) (2.46e)
mm

6* [(bamG2 (aA2 (fam

_ - aa2mG3) -bam A3)+(ßa4) 4)] U
m
24)(fA2- 0
-r(bamG2-aamG3) (A1 -bamA4)- ( famA3)] V

0
+[(G1+ aamG4) (bamA2 -aamA3) '- &(ä + G3) (aas W
4)],
m
+[(G1+ aattG4) (A1+ aamA4)+ C(ä2+ G3) (nA2 '+ SA3)]8
* ,ý* m
I PIm }p (2.46f)
a -G l
a4 a4 m
mmm

t(öan G4) (aA2- 2+

M2 = bamA3)-( aG3) (fay U
4)1
m
0
-[(aamc4) (Al - bamA4)+(aG2+ aG3) (fA2 v
-fam 3)1
m
0
-InG2 +6G3) (bam 2-a mA3)-(Gl+aam 4) (Sam 4)1 w

O
r(nnG2+ 1G2) (Ai
- + aamA4) + (G1+ aam 4) (nA2+iA3))0
* **
[(äG2+ (aG2+ äG3 )]
- aG3) PIm
mm
10
ý+ Pm
aG3 (2.46g)
m
 33

0
V2 [(famGt) (A1+ bamA4) + (aG2+ bamG3) (fA3+ famA3)' U
-

+ C(faI G4) (aA2+ bamA3) +(aG2 + bamG3) (f cc A4)] V0

[(fG2- j O
famG3) (A1- aamA4)+ (G1-baMG4) (nA2 -SA3) ct
[(fG2- (2.46h)
- fOmG3) (äA2+ aA3) + (G1- bamGLi) (SamA4)100
m

Where
G1, G2, G3 and G4 are values of hyperbolic functions 41,42,43 and 44
x2
respectively calculated at = F2 .

Also
G1, G2, G4
G3, and are evaluated at x 2= 2H - T1
** if it It x2 = T2
G1, G2, G3 and G4

While
Gi, G2, G3and G4 are evaluated for each particular point load at x2=
distance of the point load from the corner of bottom flange and web.

Equations (2.45) after substituting equations (2.46) in them, are

solved forofour non zero initial parameters of the top flange, which
W 8. Their in
are and values are substituted the relevant equations
of Appendices I and II to obtain any particular distribution on the
top flange. For the web, initial parameters are evaluated from
equations (2.34) and they are used in equations of Appendices I and II
to determine a required value at any point on the web. Values of W ,6
m
M2m and V2m are modified to include the effect of external warping
forces on the web, and are to be calculated from the following equations:

ä4 M2+ ý2_ ý3
W, 1-aa 44 ti' +- +a43 - p2
m
m m a3 a2
in m
e3(_2 [i
+ ýý2 p +i 2ý3 (2.47a)
-, in
aa Im 3 a2
a3 a2 mm
m mm

l* l ý2
0 ba ý2 -aa243 w+ ý1+ as e4 p- + ý3 M2
mmm m am

Vý4 ý4 +4
-F ým _ P_C P (2.47b)
am am Im am m ,

[tc1] [41+
M2m= - n2 + S43] e- + as44]2
LI

+a+ a43 1V2+ (ä 2+ a03)

mm
- (äý2 + aý3)jPIm

bý2
+F ++3P (2.47c)
am m
 34

[nay DCIMý4] 0+ tbct

V2m _ 42- dam 4i31W+ 2- aas 3] M2

+[$1 -aame4]V2+ L( 1- aame4)-(1 - aame4)3 PIm

[ý1 (2.47d)
+L a e4"lpm

Where
44
ý1, ý2' 3 and are values of hyperbolic functions 0142,03 and ¢4
respectively, evaluated at x2 = x2 - T1.
04
01,02.03 and are evaluated at x= x, - (T1+ cA)

and '

X194243 and are evaluated for every individual point load at

x2 = (x2 - distance of the point load from the top flange and web
corner) such that the difference on the right side is always a
positive quantity.

Equations (2.47) are to be used 'for evaluation of 113',_9, m22 and

V2 on any part of"the web subject to the following conditions:

1. For o -5x2 <, T1

"1 = $2 ° 43 = ý4 =0'
** ** ** **
ý1
=ý2 =ý3 =ý4 =0
ý1 ý2 ý3 ý4
= = 0

2. For T1<x2.5(T1+ ca)

** ** ** **
41=0203°$4 =0

Equations (2.46) are used for evaluating the values of the initial
,
parameters of bottom flange and subsequently they are used in
equations of Appendices I and II to determine different distributions
on the bottom flange.

2.6 SIDE MEMBER SUBJECT TO A COUPLE LOADING

Torque applied on the free end of a cross member . in an isolated

'joint will produce a moment loading which acts in the plane of side
member web. The effect of this moment is considered by applying a
couple in the plane of side member web as shown in fig. (2.4b). It is
assumed that the couple forces are acting in the flanges of the cross
member end only and their magnitude is given by the following relation

Fxco= T (2.48)

where T is the applied torque and F is a force acting over a

 35

length (R + S) of the The force per unit length

side member web.
corresponding to this force is given by:

F
F(xl) (2,49a)
R+S)

or
"T
F(xi) a co(R+S) (2.49b)

Any force which acts in the plane of the web will result. in
stretching of the web. Hence the effect of such an external
loading can be included in equations derived in Appendix II.

It is considered that the cross member is not located symmetrically

on the side member web, i. e. T1 # T2 in general. A conplete section
of the side member has to be considered in this analysis.

Boundary conditions on the free edge of the top flange are given by
equations (2.32) and (2.33) while equations (2.43) and (2.44) give
the boundary conditions at the free edge of the bottom flange. The
former conditions are expressed by equations (2.45) in an extended
form through the use of equations (AII. 19c and d) and (AI. 23c and d).
Initial parameters of the bottom flange depend upon the initial para-
meters of the web and the external loading. These relations are
expressed by the following equations.
U raG2
G1+ bamG41 U- + bam 3lV -rcG41N22+[G2+ cG3IN12

(dc**
+ CG3) 12 (2.50a)
3JJF
11
l am am m

0- {ý ä4 22
1 G333
V. LG1 -aamG41 W+[.. I. aG3 1ýL+ E[. V2
m mJ ma
(2.50b)
N22 = [namC2 6am2
G3]W
+[6amG4]e..
+[bamG2- aamG3]M2

[G1 V2,
+ - as G4l (2.50c)
m

N12 [f G2 + fay G3]U* [famG4 ýV [aG2 bam631N22

- + -

+1G1 + bamG41 N12 - [(G1 + bamG4) - (G1 + bamG4 )] Fm

(2.50d)
w [bamG3- [G1 ýG2-
lm
aGýÜ + - bamG4]V + cG3IN22

+jcGq]N12 [(cGq) (CG4)] Fm (2.50e)

- -
J
 36

[baG2 W+ [G1 9 ä2
00- _ - aa2G3] + aamG4J -& + G3'M2
mLm

V2 (2.50f)
-
[&R4
M22 [&& G4 ]W- [nG2+SG3]6
= +[Gl + aalG4]M2

V2 (2.50g)
++ aG3
a

V2 1V [G1-
=-I fam 4]U + [fG2- famG3 + bamG4]N22

[aG2
+ + bamG3]N12- [(aG2 + bamG3) - (aG2 + bamG3 )l. ý

(2.50h)

where

Fm is the Fourier coefficient of, the following series:

F'(xi) _I Fm cosam 1 (2.51)

m

FF is evaluated from the following relation:

(L+S)
1J
FL F (x1) cosamxldxl (2.52)
(L-R)

The final form of equation (2.52) after its solution becomes:

[ 2T 1 S-R R+S (2.53)

Fm coscxm(L + sinam(2
coamL R+S) 2)

Relationships between the, initial parameters of bottom flange

and top flange are obtained by substituting equations (2.34) in
equations (2.50). The resulting equations are substituted in boundary
conditions (2.45) to obtain a system of four simultaneous linear
equations involving four non zero initial parameters of the top
flange. This system is solved for these unknowns which are then used
in equations of Appendices I and II to determine stress/displacement
or other distributions on the top flange of the side member. Similar
distributions are obtained on any part of the web by using the initial
parameters of the web which are evaluated from equations (2.34). Since
external forces are acting on the web, so some of the Fourier coeffi-
cients of equations (AI. 23) and equations (AI. 19) are modified to
include their effects. These modified relations are given below:
 37

Um [$1 [aý2 [041

+b mý41U - +b mq 31 V- *2+[da C43] N12
m+
1id + cog) -( 1 Fm (2.54a)
- c*3)
m m

[-3I
m0 3- ao2] +[ 1-b m 4] + c4 N22
Lb
C(04)
+L04JN1'2 (04)' F (2.54b)
-

ýf ]U
N22m mý4 +[2f m3]" +[i-b m44]N22

[ace ]N12- [(a$2+

+ + bamý3 b mý3) - (a$2+ b m$3)J m (2.54c)

[f2 [f [02-'
N12mý + fam431U - mý41 V+ b m43]
2`2
..
+L1 +baýe41N12 -[4l+ b m44) -ße2 + bamý4 )j (2.54d)
m

Equations (2.54) are to be used subject to the following conditions:

1. For 0, x25T1

0102$3=$40
** ** ** **
0102$3=040

2. For Tl<x2: (T1 + co)

** ** ** **
01 02=X3=$4 0

'Distributions of various quantities the bottom flange can be

on
obtained by using the values of initial parameters of the bottom
f lange in the relevant equations of Appendices I and II. Initial
parameters of the bottom flange can be evaluated from equations (2.50).

2.7 COMPUTATION OF ANALYTICAL SOLUTIONS

The analytical approach developed in the previous sections was applied

on the analysis of isolated joints and ladder frame subjected to
torsion. Since the approach involved the solution of simultaneous
equations to determine the initial parameters of the problem under
consideration, so it was necessary to use a computer to facilitate
calculation work. The nature of the calculation work involved did
not require a large computer or even sophisticated subroutines. A
small desk model computer H. P. 9830A was available in the School and
was used for the calculation work. It only accepted programs in
Basic language.
 ýS

Two separate programs based on the following methods were written:

(a) Plate bending solution derived in section '(2.2), to

calculate deflection of the side member web treated as a
rectangular plate simply supported along all its edges.

(b) Application of the plate theory on a channel section side

member subjected to (i) A channel section cross member
forces (Section 2.5) and (ii) cross member couple
warping
loading (Section 2.6).

The main based on (b) and was developed to analyse the

program was
following joint configurations.

i) A channel section cross member attached to a channel section

side member with the cross member web vertical to the longi-
tudinal axis of the side member. This included symmetric or
asymmetric location of the cross member in the joint.

ii) An I-section cross member attached to a channel section side

member with the same restrictions as described in (i).

Flow charts of both programs are included in Appendix III. Output

obtained from these programs was in terms of the intensity factor
of the cross member warping forces. All the values outputted were
for a unit torque applied to the free end of the beam representing
a part of the cross member in a joint. The displacements associated
with any point of the side member were multiplied by Young's Modulus
of the material before being outputted so that the values could be
directly used in the calculation of the cross member warping restraint
constants. Theoretical distributions of stress/displacements of all
the isolated joints and the ladder frame were obtained using the main
program up to the first 16 terms of the respective Fourier series. In
some cases where the numbers became very large during the execution of
the main program a lesser number of terms of the series were evaluated.

The values of the side member web normal displacements U3 and bending
moments M22 at a section below the cross member web were computed
from both of the programs fQr Joint No. III. Their distributions along
that section are shown in Figs (2.5a) and (2.5b). These distributions
were plotted to compare the approximation of the side member web as a
simply supported rectangular plate.

2.8 THEORETICAL ESTIMATION OF CROSS MEMBER WARPING RESTRAINT FACTORS

The magnitude of stress/displacement at any point of the side member

was dependent on the value of the cross member warping force intensity
factor in the analytical solution. So it was necessary to determine it
either experimentally or from theoretical considerations. It has been
mentioned in Chapter I that most of the research workers relied on
[17]
experimental measurements while only Tidbury, Marshall and Roach
Alade C21]employed
and Megson and analytical approaches to assess the
degree of warping inhibition in a joint. Once its value is
obtained then it is easy to estimate the magnitude of cross member

/
 39

warping force intensity factor.

The author wanted to make the analytical method developed in the

present work a self sufficient one for the estimation of stresses
in the joints. This could only, be achieved if the cross member
warping restraint factor for a 'joint configuration under consideration
was estimated from the present analytical work. The relationship
between the stri beam constant "C"[17] and the cross member warping
"V F21 as follows:
constant was derived
1

1 +ý (2.55)
utanh(p )]

The average value of'C for a particular joint configuration was defined
from the cross member end displacements and the stresses in the
foliwoing way:

*E fwds
Cav _ (2.56)
f v, ds

Cav was evaluated from the values of bending displacements of the side
member web under the cross member end profile by the following relation:

[E k0 f 3ds]/1I Lo f 1
Cav pds (2.57)
t
The values of the integral terms of equation (2.57) were estimated
by measuring areas under the respective The value of tav
curves.
obtained from equation (2.57) was used in equation (2.55) to estimate K.
Then the value of the cross member end warping force intensity factor
per, unit-applied torque "ko"was evaluated from the following equations.
T

Op = (1-K)a, (of Ref. 211 with original notation)

sinh p(L-0 `(of Ref-[31] with original

T2AE
coshuL notation)

The bending displacements of the side member web along the upper half
of the cross member profile (as the joints were symmetrical) was
plotted for each of the joints. These distributions are shown in
Figs"(2.6), (2.7), (2.8) (2.9). The values tav, K and ko/T were
and of
found using the strip beam theory [17], finite [21],
element analysis
plate bending solution and plate theory as used in the present work.
The values of the constants obtained from each method are given in
Table I.
 40

o Simply supported plate in bending only.

/Soo " Complete Channel Section (subject to warping forces)

\
goo `
o

\\ T
X

600 ý.
Jý
F.

/
Q Soo -
f\
N

/
o
o to 20 30 40 So
.D/STANCE AROM TivE FcAivGE/wsca Ca4NER iN MM. 00

Fig . 2.5a Theoretical distribution of normal displacements on section CC

of'the side member web. (Joint III)

t 0 Simply in bending
supported plate only.
e Complete Channel Section (subject to warping forces)

Co

l;
4s
x
ýN I

-3-D \

0 30 40 50
io 10
d /STANCE AROM ras FLA&GE/1E9 CORNER /4/ /"I". OF

/" S
Fig. 2.5b 'Theoretical dis ritýuti RI3 f the bending moment m22
on section CC. Joint
 41.

350

iT:: Zoo
ng only.
ct to

ct to

ý. /S °
x

Z 10,

H
So
Ä

-So

-/o 0
1

-/So

-200

-SZSo ý

Fig. 2.6 Theoretical distribution of the side member web normal

displacement along cross member end profile (Joint I)
 42

300

200
_

ISo
x

too

W
z 50

-So i bending

z (subject t
-/00

(subj ect
ISO
)uple)

-200

Fig 2.7- Theoretical distribution of the side member web normal displacement
along cross member end profile (Joint II)

t 1S-o F

200

o-Simply supported plate in bending only.

/so s-Complete (subject to
channel section
warping forces)
/oe x-Complete channel section (subject to
warping forces and couple)
x
rn So

O
10 20 30 40 SO 60
W .
D/. $ rCf FROM .6 iN P)" . No

- Sa
J

-/So

moo

Fig. 2.7b Theoretical distribution of the side

member web normal
displacement along cross (Joint IV)
member end profile
 43

o- Simply, supported plate in bending

only.
O-Complete channel section (subject
to warping forces)
a-Complete channel section (subject
to warping forces and couple)

o 30 40 60 70
/O Zo

Dis r4iv[E FROM 19ý .w mm .-

.

-Zoo

400

-Coo

-goo

Fig. 2.8 Theoretical distribution of the side member web normal

-iooo
displacement along cross member end profile (Joint III)
 44

1200 f

/ý

loco

800 /1 I
/

600
ti 1/ I
ý-.
I
x 400

I
I
WI
1
1

QO
IS 3o hS 60 7S

D/STANCE ýPOM A/ IN

o-Complete channel section (suhject to

-boa warping forces)

e-Complete channel section (subject to warping

forces and a couple)
_goo

1-0

- 10001

Fig. 2.9 Theoretical distribution of the side member web normal displace-
ment along cross member (Ladder frame inner joint).
end profile
 45

IH M M N VN
.4 V-4
ö UI Oý N N
.
cd ". a O O O O O

0
1 a+ N N O
cn 't
OA 4j 0 O 0 Oý

r+ aa
ß+Qýi o N
co ln o UI
U cd .o
4CU N. ý,
N u1 N
N

C)
te
cd ; IH ý 00 . a 00
r1 e 00 N %0

o O o O
0
,wo
4-3
or. N M
W 1-4 0O Cl
92. Ici
O r O O- - ,O O
C7
Co N M M Gý
Hpý
P4 "ý Oý ýC N N

N ýO Co
ýT Co
H O M ýY Cý
2IH --t 00 N 0
%. '" f
."
q O O O O O

P4 N
N %IO N N
e-4 10
WO
O O O O O
a1ý+

U,
U,
a FTr
U
ci 1 1 1 1
U
H
-1 N
N M
ý, O C'l O
o .4IH
O O O O O
O
E-4
"0 C% LI %0 %D
9
N. 4
C, C% 00
1- CO N r
x
0 0 o O
ca o
"ý N
N
N t, N
yý ýN r. 4

U M
CI -t CI

1J

4.4 O

0 Z H H O 0)
1"4 H
º7 10 C)
 46

CHAPTER 3 'EXPERIMENTAL INVESTIGATIONS

3.1 INTRODUCTION

The main purpose of the present experimental investigation was to

determine stress distributions in the joints of chassis frames.
Joints made of channel section members with cross membezs attached
to the web of the side member, were to be used in this investiga-
tion so that results could be compared with theoretical analysis.

A general survey of the experimental methods available for stress/

strain measurements showed that electrical strain gauging was a
commonly used technique. Many researchers investigating in the
area of vehicle chassis frames had employed dial gauges for displace-
ment measurements and electrical strain gauges for stress/strain
[13]
measurements. Takahashi employed a "vibration method" using
a hand held extensometer for measuring strains on a truck chassis
frame. He obtained very good results by this technique.

Since it was expected that localised stresses could develop in the

region of a joint, an experimental-method which could detect such
regions easily was considered more useful in the present case. It
was also intended to measure the magnitude of localised stresses
accurately. Various methods of stress/strain measurements could only
measure average values of these quantities over a certain gauge length.
This results in measuring lower than the actual values of localised
stresses. These inaccuracies depend upon the gauge length of the gauge
used and gradient of stress/strain. Electrical strain gauges and
other methods employing the use of an extensometer were not considered
suitable in the present experimental work. The photoelastic technique
was chosen as it would be suitable for qualitative and quantitative
analysis very effectively. -

Some simple photoelastic models were made and investigated to justify

the usefulness of the photoelastic technique in analysing thin walled
beams of open section. Once the initial experiments demonstrated
good results, channel-sections were cast in aluminium moulds and various
isolated joints and a model of a simple chassis frame were constructed.
Preparation of reliable photoelastic models by casting requires a
, in Appendix
specialised technique and the details are given IV for
future guidance. Photoelastic investigation of all the models was
carried out by using a reflection polariscope with a white light source.
Coloured and black and white photographs were taken of some of the
models'to show distribution of fringes.

3.2. METHOD OF PHOTOELASTIC ANALYSIS EMPLOYED

Photoelastic analysis of a thin walled structure such as a joint between

a side member and a cross member by a transmission polariscope is not
feasible. This is because the light transmitted through any plane
surface of a photoelastic model could be obstructed by some other surface
 47

of the same model, e. g. both of the flanges of a channel section

would lie in the way of a light ray. ' By mounting birefringement
coatings on the surface of a structure and analysing the photo-
elastic effect by a reflection polariscope, one could solve the above-
mentioned-difficulty. Use of birefringent coatings is quite common in
the analysis of complicated structures and the effect of bracket holes
in a side member of a truck frame was studied [32] by this
using method.

The birefringen. t coating technique has some limitations which effect the
quantitative analysis although qualitative results could be obtained
easily. These limitations' are:

(a) A birefringent coating tends to reinforce the strength of the

actual structure. The reinforcing effect depends on the coating
thickness [33].

(b) The coating material should be optically very sensitive so that

comparatively small surface strains in a metallic structure
could produce reasonable amount of birefringence in the coatings.

(c) Residual birefrigence could result during the cementing of

coatings on a structure.

(d) Strain distortions could result due to difference of the elastic

and adhesive [34].
properties of'plastic, structural-: material

The second limitation was the most important because of the experience
in [35].
of many other researchers and particular of the author's own
These considerations led to the choice of using the method successfully
in [35]. This involved the use of a reflection for
employed polariscope
analysing photoelastic models after putting a thin reflective layer on
one surface. The advantages are twofold. Firstly, the amount of bi-
refringence is increased by directly loading a photoelastic model and
secondly the loading rig becomes light and simple to construct. The use
of reflection techniques also doubles the birefringence effect. The
present experimental work confirmed this choice.

3.3. PHOTOELASTIC MODELS

Photoelastic models used in the present work were developed systematically

in stages. The details`of photoelastic model preparations are described
in Appendix IV. The sequence of development is described below.

(a) Channel sections of dimensions shown in Table JI were fabricated

from CT200 plastic sheets. Three pieces were cut to size and were
joined by Araldite adhesive to form a channel section of the required
dimensions. They proved successful in simple experiments but the
joints were not very reliable and it was not possible to obtain
sections of relatively large dimensions specially lengths, as
explained in Appendix IV. So this technique of making channel
sections was abandoned.
 48

TABLE II. FABRICATED' CHANNEL SECTIONS

Section Section Dimension (mm) Length (mra)

FI 68 x 32 x3 305
FII 36 x 38 x3 323
FIII 37 x 30 x3 297

(b) It was decided to cast channel sections. They were cast in

aluminium moulds by pouring a mixture of CT200 Araldite and
HY905'Hardener'in them, under temperature controlled_conditions.
Curing, and annealing of the cast sections were carried out in
an electric oven as described in Appendix IV. Dimensions of
the cast channel sections are given in Table III. The
aluminium moulds are shown in Plate 3.1 along with. some of the
cast sections. The sections produced by this method were almost
free from residual birefringence which is a necessary require-
ment for an accurate photoelastic analysis.

TABLE III DIMENSIONS OF CAST CHANNEL SECTIONS

Section Section Dimension -(mm) i Length '(mm)

CI 42 x 40(max. )x 3.8 320

CII 64 x 60(max. )x 3.8 545

CIII 99 x 52(max. )x 3.8 495

(c) An aluminium mould was designed to cast an Araldite model of a

complete joint between two channel sections. Plate 3.3 shows
various parts of this mould and its assembled view is shown in
Plate 3.4-Castings could be made in the mould but the cast model
could not be released easily from the mould surface. The mould
release agent which was used in casting individual channel
sections did not give satisfactory results. Even a small force
employed in releasing the model resulted in its breakage as
Araldite is a brittle material. This method of casting complete
joints was abandoned as a suitable mould release agent could not
be found. 'A broken model is shown in Plate 3.4.

(d) Photoelastic models of four isolated joints, in which the channel

section representing a part of a cross member was directly
attached to the web of a channel section representing a part of
a side member, were made by. using a special temperature cured
Araldite adhesive. Its constituents were AY105 Epoxy Resin and
IIY951 Hardener. Joints of sufficient strength were obtained by
this adhesive. One model of a joint in which the cross member
was attached to the side member web with Araldite brackets and
bolts was made. Bracket holes at sections of zero warping were
drilled at the cross member flanges and web. The cross member
 49

end was free to warp because of a clearance left between it and

the side member web. All the joints investigated experimentally in
the present work are shownin Plate 3.2, and the dimensions are given
in Table IV.

(e) A model of a three bay ladder type chassis frame was made by
joining cast sections with'the same adhesive as used for making
isolated joints. Residual stresses developed in the frame during
its assembly had to be reduced by subjecting the whole frame
to a heat curing cycle in a large oven. Usually it is not possible
to remove residual birefringence completely. By this method residual
stresses on the side members were removed but small permanent deforma-
tions were obtained in cross members due to sagging. This produced
residual birefringence in the middle portion of the cross members but
their ends and side members were suitable for photoelastic investi-
gation. Dimensions of the frame are shown in Fig. (3.1).

TABLE IV DIMENSIONS OF ISOLATED JOINTS

Side member Cross member

Section Section
Joint Description Length Length
Dimension Dimension
(mm) ()
(mm) (mm)

I Adhesive 64 x 59 x 3.8 382 , 42 x 33 x 3.8 255

Joint

II " 64 x 51 x 3.8 266 37 x 30 x3 297

III " 99 x 39 x4 482 64 x'45 x 3.8 500

IV 64 x 51 x 3.8 266 42 x 27.5 x 3.8 281

V Bolted 99 x 42 x4 445 64 x 39, x 3.8 455

Assembly

3.4. CALIBRATION OF CT200 ARALDITE PLASTIC

'It was essential to determine the values of Elastic constants of the

photoelastic material used for making models as the values were not
obtainable for the material used in the present work. Quantitative
results from photoelastic measurements can be obtained if the stress-
optic or strain-optic characteristics of the photoelastic material
are known.

3.4.1. Determination of Elastic Constants

A tension specimen of dimensions shown in Fig. 3.2 was cut from one of
the cast sections of CT200 Araldite. One electrical strain gauge
was stuck on each surface at the middle part of the specimen. These
gauges were mounted in such a way that one would indicate longitudinal
strain while the other would measure transverse strain in the specimen.
Dummy gauges were also mounted for temperature compensation on another
CT200 Araldite piece.
 50

to I\
}ýY
l

vV
Q
V

OI

a)
F v ý ý \
ý,
w
F o

ý ýZy b
cu
V Qä
o °ý
r rý
Q
w
0

0
", -1

W s'`\
t4
lý A
ýK
(PIG
M
Ný _J,
t4l 00

H
F
tu %
O
ü
.ý
her

11 E
zM
"Y
y
x

O
 51

Loads were applied to the specimen and strain readings were recorded
at different values of loads. Measurements were also made when the
specimen was unloaded. Average values of both sets of readings were
used in the calculation of the elastic- constants. Fig. (3.2) shows
variation of load and strains in the longitudinal and transverse
directionsof the test specimen. The following values of the various
material constants were calculated.

0 Young's Modulus "E" = 22.15 x 106N/m2

b) Poisson Ratio "v" = 0.405

c) Shear Modulus of "G" 7.882 x 106N/m2

2E1+y)=
elsticity

3.4.2. Determination of Material Fringe Value "a

Stress-optic characteristics of CT200 model material were investigated

from two types Hendry [36]suggested the use of
of calibration models.
a circular disc loaded in axial compression instead of a tension or
compression test piece as true axial load cannot be easily obtained.
The author preferred to use both, a circular disc and a tension piece,
to obtain a comparatively reliable value of the material fringe value.
One surface of each specimen was made reflective by spraying silver
paint. They were loaded accordingly and fringe order measurements were
made by a reflection polariscope at different values of load. The
fringe order was measured at the centre of the disc and at the middle
of the tension piece. The separation line between red and blue colours
was taken as a reference for counting the order of fringes at a parti-
cular point on the calibration models. Fractional fringe orders were re-
Compensation [37]. These measurements
corded by the Tardy technique
were recorded for increasing values of'load and when the load was
decreasing. Average values of the two sets were considered in the
calculation of the material fringe value.

A calibration curve between the applied load and the fringe order was
drawn in each case. The value of the material fringe constant was
obtained from the curve and it was used to calculate the material
.fringe from following
value the relations.

a) For a tension piece:

2P
Material fringe value "f "

(b) For a circular disc in axial compression:

Material fringe value obtained from fringe order at the centre

of calibration disc
16P
1
1a 7rDN

Figs (3.3), (3.4) show the dimensions of the circular disc and tension
The values f.
specimen along with calibration curves respectively. of
 52

b 9sl '1 ý
a. a

f,

O I" g

h i ö
q
I ý,
"1 u

ý I "ý
1 ul

W
SN
0

41

Its 111
O
0 O ", 4
`

ö
k
to Q Sol
l
Q41
n w b

W
IU 11 11

?
V W
20 h I 2 v

%4 I co
14
ö" ý2Q d

ö N
ti
o
I I v
W
-I
0
1
0
.41
%n Cd
1 "-4

Cd
I U
o N
00 00 ap 0
CD ýD N
e} M

(a. w
- ca/wca o/ r E) ivy /V/rsy1 S 76'N1Gn119No7
'
a-ýww/ww o/x1 -J Ni /vi ' '1S ýSäýý/15Nbä1
9
 53

lop

$o ös 14
nmw
KNCLs
=4
MM.
W
Co
Or
v
4

P" .
vý
40

0
W
V 20

0
0 /0 Z0 30 40 So 60
104. D P /N NEWTONS ---º

Fig 3.3 Calibration using a tension piece.

2#01

H
W

" V,
W
q mm
2

eý

4C
Q
Ö

W
u

0 40 80 120 160 Zoo

L oAD P IN NEW TONS

Fig. 3.4 Calibration a compression disc.

using
 54

obtained from the two calibration models agreed well with each
other and the following material fringe value was used in all
further calculations.

fa 11.4 N/mm/fringe.
=

3.5 PHOTOELASTIC INVESTIGATION OF A COMPLETE WARPING INHIBITION CASE

Fabricated channel section F. II was used to observe the photoelastic

effect of stresses produced by warping inhibition. One end of
section F. II was bonded to a thick mild steel plate with Araldite
adhesive to obtain a cantilever model of a channel section. A
reflective layer, made up by joining a silvered paper on a piece of
transparent double sided adhesive tape, was bonded to the inside
surface of the section.

The problem of application of a pure torque at the free end of a

cantilever was carefully studied. The drilling of holes in a photo-
elastic model usually results in residual stress patterra
distort .
which the stress patterns produced by external loads. The holes
also act as stress raisers thus changing stress patterrs in the
surrounding regions. It was decided to avoid any method of load
application which would require drilling of holes in the rrodel. Fork
ended mild steel pins shown in Plate 3.2. were made and the gap
between two legs was kept slightly more than the wall thickness of the
channel section. The pins were attached to the flanges and web of
the section at the points of theoretical zero warping.

A light aluminium circular disc was mounted on these pins at the free
end of the section. A couple was applied on the disc by applying
equal and opposite forces through a dead weight, pulley and string
arrangement.

The coloured stress pattern was observed through a reflection polari-

scope. Fringes due to warping inhibition stresses were very dominant
on the flanges at the built-in end and their number reduced sharply
as the distance from this end increased. A black zero order fringe
appeared on the flange along the line of zero warping. The fringe
'distribution on the model seemed to be in good agreement with theore-
tical stress distribution hiotig the edge. This model was kept as a
demonstration model and actual measurements were not made on it.

Another model of a built-in beam was made by using a cast channel

section and was loaded as described above. Fringe orders were
measured along the free edge of the flange. The distribution of
fringes and dimensionsof the beam are shown in Fig. (3.5).

3.6 TESTING OF ISOLATED JOINTS

All the five joints described insection 3.3 were analysed photo-
elastically in the following, manner:

3.6.1. Method of support and loading

The three point method of support at two ends was employed to support
 55

of

1ý

0
"P4
41
V
C)

C)
ii

w
0
to
00
b
C)
C)

4-1
d

w
C)

u
w
r4
o ''
Ici

tÄý
w
W0

ti ü
0
'I1

t
c4
DO
04
00oäo
1n °ý
Cc cý eý
ýan
isSýöý 'wo)
O/ X 59 SSýÖ'1S 1ý3ý/ f
Z-
 56

the part of each joint representing the side member. Fork-ended

pins were mounted with temperature cured Araldite at the two ends
of a side member. These pins were located at the points of zero
warping on the flanges and web of the section. The part representing
the side member was supported in two end plates vertically through
the pins. The pins could move freely in the holes provided in the',,
thus providing practically no warping inhibition to the
end plates
side member section of a joint.

Torque at the free end of the cross member section was applied
through a wooden bar which was supported on three pins as described
in Section 3.5. bar-was-used instead of a circular disc
A wooden
to obtain a longer moment arm without increasing the overall
weight of the arrangement.

3.6.2. Photoelastic investigation

The method of obtaining a reflective surface described in Section 3.5

showed that the fringe pattern was affected by the orientation of the
double sided adhesive tape. Spraying aluminium paint on the surface of a
model provided a more uniform and better reflective surface'than the
adhesive tape.

A fine square grid of-5mm spacing was drawn on the portion of side
in the region of the joint. Measurements of iso-
member section
chromatic fringe order and isoclinic parameters were made at various
the by a reflection polariscope. [383. The
points of grid presence of
residual birefringence affected the stress patterns on the flanges.
The position of isochromatic and isoclinic fringes was also sketched
on squared paper separately so as to have a check on the accuracy
of direct measurements made at individual points on the model. Iso-
fringes half isoclinic
chromatic of the whole order and order, and
fringes of 00,150,300,450,600, and 750 were sketched. Measurements
were made at various values of clockwise and anti-clockwise torques
applied to the cross member free end. Fringe orders were also recorded
at the free end of the cross member flanges.

Since light in the whole analysis, all iso-

. a white source was employed
chromatic fringes were coloured except for the zero order-fringe which
was black. All isoclinics were black. Some pictures were taken of
the stress patterns on the joints and are presented in this work.

Attempts were made to obtain another set of isochromatic fringe measure-

ments by using an oblique incidence attachment with the reflection
polariscope. They proved unsuccessful because scattered reflection of
light from the top surface of the photoelastic model made colour
distinction almost impossible. Oblique incidence measurements were
required for the calculation of individual principal stress values at
any point. Plate 3.5 shows the method of support and loading of a
photoelastic model of an isolated joint employed in the present experi-
mental work.

3.7 TESTING OF A LADDER FRAME MODEL

ACT200 Araldite photoelastic model of a ladder frame was subjected to

 57

torsion to investigate stress distribution in the region of its

joints. The loading and testing of the frame is described below.

3.7.1. Method of support and loading

As the frame had two axes of symmetry it was decided to support

the frame at four theoretical points of contraflexure as shown in Fig.
(3.6). Two brackets were specially designed to support the frame at
sections lying in the middle of the side members. The brackets
provided a three point support at these sections as shown in Fig.
(3.7). Drilling of holes in the side members was avoided by this
arrangement. The whole frame was suspended by long rods attached
to the support brackets at approximately the shear centre of the
section. The frame was supported on a cylindrical bar lying along
its longitudinal axis of symmetry.

Four brackets were designed similar to the support brackets for

loading the frame in torsion. They provided a method of applying
loads through the shear centre of the side member section without
producing local twisting of the section. They were attached to the
side member section at three points of zero warping without drilling
holes in the section. This design of bracket provided more flexibility
in regard to the choice of distance between two brackets on the same
side member. A twisted view of the frame is shown in Plate 3.6.
A similar loading in [22].
method of was used

3.7.2. Photoelastic investigation

Square grids of 5 mm spacing were drawn in black pencil on the

flanges and webs of the side members of the frame in the region of
the joints. Some residual birefringence was present on the free
edges of the side member flanges. It was due to cutting of the flange
edges with a band saw. The nature of residual birefringence was
studied with a reflection polariscope. It was noticed that the com-
pressive stresses produced as a result of cutting had produced this
effect. The measurement of fringe order was made to estimate the
magnitude of residual birefringence at various points on the model.

The stress pattern in the region of the joints was observed through
the reflection polariscope"after twisting the frame by the application
of shear loads in the brackets. Due to the presence of residual bi-
refringence on the side member flanges, the resulting stress patterns
were affected. The portion on which compressive stresses due to the
applied loads were produced, the combined fringe order increased due
to the presence of the compressive residual birefringence. But at
other parts where the loading and the residual effects were of opposite
nature, the combined fringe order reduced. The loading effect was
much more dominant on the side member flanges well away from their
free edges as residual birefringence was negligible in those regions.
The part of the side member web in the region of any joint did not show
fringes of significant order due to the applied loading. The fringe
pattern in that region was not a regular one and was largely affected
by the irregular flow of adhesive used in the construction of the frame.
The stress patterns on the side member flanges provided a better infor-
mation of stress distribution. The measurement of isochromatic and
 58

DEFLECTED FRAME 70 T0? S/ON

\AUE

\ý
;ý", ý
\`UNDEFL
EG TED FRAME

1, Z, 3 ANS 4 ARE TtiE PO/YTS OF SLPPD. PT

Fig. 3.6'". Support points on the ladder frame.

8 RACRE7'
W--

VVr. SCREINS

CNA/YNE L r °"
_s

Fig. 3.7 Method of attachment of brackets.

 59

isoclinic fringes was made when the frame was subjected to a

torque of 17.28 x 103N. mm. Similar measurements were made on
the portions of the. cross member flanges near the joints. The
fringe patterns were also observed when the direction of the
applied torque was reversed.

3.8. PHOTOGRAPHIC RECORD OF FRINGE PATTERNS

One of the main advantages of the photoelastic method lies in its

ability to provide a visual representation of the stress/strain
distributions in a model. Isochromatic and isoclinic fringe patterns
of a stressed photoelastic model provide such information. It is
easy to keep a permanent record of fringe patterns using the. photo-
graphic technique.

The fringe patterns obtained by a transmission polariscope are easy

to photograph as enough light is transmitted through the photoelastic
models during their photoelastic analysis. But the use of a reflection
polariscope for such analysis does not provide light of sufficient
intensity after being reflected from the reflective surface of the
photoelastic model. Most of the incident light is lost due to
scattering after reflection. This complicates the photographic
procedure. This difficulty was experienced in the photography of
fringe patterns in the present work. It was also experienced that
the reflection of the incident light from the top surface of every
photoelastic model proved a major problem in obtaining good photo-
graphic results.

The entire photoelastic analysis was done using a white light

projection lamp fitted in the reflection polariscope producing
:
A few colour
coloured fringe patterns. slides of these fringe patterns
were made. Plate Nos. (3.7) to (3.14) show colour prints made from
these slides and show isochromatic fringe patterns (i. e. magnitude
fringes) and combined fringe patterns (isochromatics and isoclinics)
as labelled on the photographs. Plate Nos. (3.9), (3.10), (3.13) and
(3.14) were taken with the lamp source provided in the reflection
polariscope while Plate Nos (3.7), (3.8), (3.11) and (3.12) were taken
with an electronic flash gun used as a light source.

Black and white photographs of various fringe patterns were taken

using a red filter with the camera. They are included in the next
chapter. -
 60

!, 1: 1t I. ý111piiIliII!, I I
: 1 '1l0II1cl : lll, t'[Itil c'! IýII)II0'I `ý('('I cfl
.

I'I; 3.. Ili, Ir-dl. lI&'U 101II

lI&. ý'IR)Luýýýatiý 1 01
 61

Plate 3.3. Parts of the Joint Cast int; mould

A
0

Mate 3,, 4m Aluminium moutu Emd in(-)doi UI th rýsý ; uiul

 62

ýI'rst! I)¶ of tin iso1: 1tk, (I jý, ' I

1'ý
_. , ý-
ý. -
--

Pi :ti3. () 'I'II, ý,iý! ý 1, i, I,, (,, iý, 11 11;1

 63

Plate 3.7 Whole order isochrömatic fringe pattern

on the flanges (Joint I)

Plate 3.8 Mixed pattern containing 00 isoclinics on

the flanges (Joint I)
 64

Plate 3.9 Whole order isochromatic fringe pattern

on the flanges (Joint III)

Plate 3.10 Whole order isochromatic fringe pattern on

the flanges (Joint III)
 65

Plate 3.11 Mixed fringe pattern on the flanges

(Joint III)

Plate 3.12 Mixed fringe pattern on the web (Joint III)

 66

Plate 3.13 Whole order isochromatic pattern on the

flanges (Ladder frame joint)

Plate 3.14 Fringe on the flanges (Ladder

pattern
frame joint).
 67

CHAPTER 4 DISCUSSION OF RESULTS

4.1 BEHAVIOUR OF CT200 MODEL MATERIAL

CT200 Araldite was used as a photoelastic model material in

the present linear and elastic behaviour for the
work and showed
load on various models. This trend is shown
maximum requirements
in Fig (3.2)which indicates this material is suitable as a model
for the analysis of stresses and strains in structures
material
in the elastic range. Fig (3.3)shows its linear optical-stress
but its material fringe value is high which indicates
property
its low sensitivity. The number of fringes at the highly stressed
regions on the side member of. the joint remained less than two.
This can be seen from all the photographs presented in this work.
However on the cross member flange near the joint, fringe orders of
three were obtained.

4.2. INTERPRETATION OF FRINGE PATTERNS

It is important to identify isochromatics (magnitude fringes

representing lines of constant maximum shear stress/strain) and
isoclinics (directional fringes identifying lines of constant
inclination of principle stress/strain) in a fringe pattern.
Isochromatic fringe patterns are obtained with circularly polarised
light and usually contain a zero order isochromatic fringe of black
colour in coloured as well as black and white fringe patterns. Iso-
clinic fringes (always of black colour) superimposed by isochromatic
fringes appear in a fringe pattern obtained with . plane polarised
light. If the stresses induced in the model are small then the
pattern can contain only isoclinics otherwise both types of fringes
appear. Various fringe patterns obtained during the present work are
described in the light of the above-mentioned introduction as follows.

4.2.1. Coloured fringe patterns

The distribution of isochromatic fringes on the flanges of the side

in the Joint I is in Plate (3.7).
member and the cross member shown
The region of maximum fringe order on the side member flange is clearly
identified by a blue fringe next to the flange/web corner. The blue
fringe is also present on the free edge of the flange in the upper
half portion only. The presence of the residual birefringence of
compressive nature on the free edge has caused an increase of the
fringe order in the upper portion while it has reduced the fringe
order in the lower portion. This effect was observed on all the
models near the edges. The presence of a black fringe in the middle
portion of the upper half indicates that the maximum shear stress
becomes zero but attains maximum values on the edges. The fringe
pattern shows that warping inhibition causes only localised effects
on the side member. The isochromatic fringes are distributed along
the cross member length and their maxmimum concentration is visible in
 68

the joint near the free edge of the flange which shows that the
maximum stress occurs there. It has been noticed from the various
photographs taken with flash gun that the zero order black iso-*
chromatics on the cross member flanges have not appeard distinctly
as were observed with the actual light source during analysis.
A white band has appeared in its place because of their inadequate
contrast in the coloured pattern, which approximately coincides
with the zero warping line on the flange. It is also seen that
the fringes go on disappearing rapidly along the free edge of the
flange as the distance increases from the joint. This clearly
indicates that the warping inhibition stresses, along the cross member
decrease rapidly as the distance from the joint increases. Some
distortion of the fringe pattern is visible near the joint. It
was observed that irregular fringe patterns were produced along the
boundaries of the joints due to curing of the Araldite adhesive.
This local effect produced residual birefringence in the joint. Iso-
chromatic fringes with a better colour contrast are visible for Joint
I in a mixed fringe pattern shown in Plate (3.8). The black fringes
surrounding the coloured fringes on the side member identify the
position of 00/900 isoclinics depending on the orientation of the
polarising axes of the reflection polariscope. The region of the
maximum fringe order on the side member flange can be clearly
identified which occurs near the flange/web corner in line approxim-
ately with the white. band on the cross member flange. The white
band as described above is the position of the zero order fringe
identifying the position of the theoretical zero warping section.

The distribution of the isochromatic fringes on the members. of Joint

III is shown in Plates (3.9) and (3.10)for equal clockwise and anti-
clockwise torques applied respectively at the end of the cross
member . The anti-symmetry of the two fringe patterns with respect
to the joint agrees with the theoretical considerations observed for
a symmetric joint. The region of the maximum fringe order is again
near the flange/web corner on the side member -flange. The black
fringes are visible on the side member and the cross member showing
regions of zero maximum shear stress. The reflection of the light
source from the outer surface of the model has affected the visibility
of the fringes on the cross member. The region of the maximum fringe
order on the side member flange of this joint is better identified in
Plate (3.11)which represents a mixed fringe pattern of the joint.
This region is in line with the white band on the cross member as
for joint I but is slightly away from the flange/web corner. This
shows that the location of the maximum fringe order region on the
side member flange is affected by the dimensions of the members
meeting in a joint.

A photograph of the mixed fringe pattern on the side member web in

Joint III is shown in Plate (3.12). An irregular fringe pattern exists
which is difficult to interpret. The main reason for this irregular
pattern is due to the presence of the residual birefringence as a
result of adhesive curing. It was observed by varying applied torque
on the joint that a very small birefringence affect was produced on
the side member web as compared to the side member flanges. This
indicated that the flanges experienced more stresses than the web
of the side member. Roach has [39lthat by far the
also reported
 69

greatest number of failures which occurred on frame side members

took place in the top or compression flanX. The distortion of the
fringe pattern in the side member web occurred in every joint., as the
fringes, due to load, were small so the photoelastic analysis was
carried on the flanges only.

The isochromatic fringe patterns obtained on the inner joint of the

ladder frame are shown in Plates (3.13), (3.14). The zero order
black fringe on the cross member flange coinciding with the line
of theoretical zero warping is clearly shown in Plate (3.13). The
fringes are regularly distributed on the cross member as observed
in the fringe patterns of isolated joints which indicates that
warping inhibition is the main factor. This is also demonstrated
from the fringe patterns on the side member flanges. Some distortion
of the pattern is attributed to the residual birefringence caused in
the frame during assembly which was later partially relieved by
annealing. The region of the maximum fringe order on the side member
flangesas seen in Plate(3.9)showsa fair agreement with that in Plate
(3.14). The distribution of fringes on the side member flange towards
the left end was affected by the support bracket. It was observed
that fringes appeared in the sections near the loading and support
brackets because of the tightening of the clamping screws. It was
difficult to adjust the tightening force on the various screws
uniformly but care was taken not to exert more force on the plastic
model. The effect of using these special brackets was certainly
better than attaching brackets through bolt holes as was experienced
with Joint V.

4.2.2. Black and white fringe patterns

The development of the isochromatic fringe pattern in the-Joint I due

to applied torque is shown in Plates (4.1a) and (4.1b). The latter
pattern was photographed at a higher. torque than the former. A compari-
son of the two shows that the fringe encircling the maximum stressed
area near the side member flange/web corner has expanded slightly
while an additional fringe has appeared on the free edge of the cross
member flange. This demonstrates that the warping inhibition stresses
increase very rapidly in the cross member flange as compared to the
side member flange. Plates (4.2a)and (4.2b) represent the mixed
fringe patterns on the Joint I showing the position of 00/900 and
45°isoclinic fringes on the side member flange. A comparison of the
two indicates that the principal stresses change their directions
very rapidly near the flange/web corner on the side member flange.
The direction of the principal stresses has changed through 45°in a
distance of less than 10 mm.near the corner but this effect decreases
towards the middle of the flange. The inner black fringe of order one
in Plate (4.2a) encloses the region of maximum fringe order.

The positions of the whole order and half order fringes on Joint II
are shown in Plates (4.3a) and (4.3b) respectively. The region of
the maximum fringe order on the side member flange is shown in the
mixed pattern Plate (4.3c). The channel section used to represent
a portion of the cross member in the Joint I was made from a CT200
plate. During cementing of the strips to form a channel section, one
of the flange strips moved to distort the section. The effect of this
distortion has slightly distorted the isoclinic patterns obtained on
 w
70

the two flanges of the side member shown in Plates (4.4), (4.5)
and (4.6). A comparison of Plates (4.5) and (4.6) shows the
anti-symmetry of the respective isoclinic fringe patterns on the,
side member flanges as expected in a symmetric joint between two
channel sections. These patterns do not show isochromatic fringes
on the side member flanges as they were recorded at a small applied
torque. The small movement of the isoclinic fringes of various
parameters near the flange/web corner of the side member clearly
indicate the transitional state of principal stresses in this
region.

The mixed fringe patterns on the side member web in the joints of
the ladder frame shown in Plates (4.8a) and (4.8b) indicate
irregular patterns as experienced in the isolated Joint III. It
appears if the additional amount of adhesive is somehow removed
during its curing stage then a better fringe pattern representation
may result. This would affect the joint strength considerably.
However, the effect of the residual birefringence caused by the
curing of the adhesive would still distort the fringe pattern in
the region of the joint. Plates (4.9a), (4.9b) and (4.9c) show the
patterns of-whole, order and half order isochromatic fringes on the
joints of the ladder frame during testing at various loads. It is
clear from these patterns. that the region of the maximum fringe order
on the side member flange exists nearer to the flange/web corner
than the free edge of the flange. The distribution of the fringes
around the region of the maximum fringe order on these joints resembles
the corresponding distribution on the isolated joints. The cross
member flanges also show identical fringe distribution to the cross
member flange of an isolated joint.

Small distortions present in the fringe patterns were largely due to

the presence of residual birefringence mostly along the free edges
of the flanges which were taken into consideration during actual
measurements. Because of the presence of the residual birefringence
it was not possible to obtain direct measurements from the photographic
record. An irregular residual birefringence in a model can further
the photoelastic [38]. The photographic record
complicate analysis
of the fringe patterns on the various joints clearly identify the
" regions of maximum fringe order qualitatively. It has been shown
that on a side member this region is present on the flanges near the
flange/web corner. Its position is also identified to coincide
approximately with the line of theoretical zero warping on the
cross member flange. A comparison betwean the experimental stress
values and the theoretical ones is made at the end of this chapter.

Joint V which represented a bolted joint did not show enough bire-
fringence on the side member or cross member for the applied torque. This
shows that when the cross member end in the joint is not restrained
to warp then very small stresses develop in the joint. It was observed
that high birefringence resulted around bracket holes because of bolt
tightening force. Such effects must be occurring in the actual joints
but are ignored.
 71

4.3 CROSS MEMBER PARTIAL WARPING RESTRAINT FACTORS

Since the degree of warping inhibition plays an important role

in the theoretical estimation of stresses in various members of
a joint, an approximate method for its calculation was used in
the present work. calculation was based on the assumption
*This
that the values of C and K remain constant around the cross member
end profile in a joint. The values of these constants were calcula-
ted for comparison purposes from various available methods and are
shown in Table. I.

The comparison indicates that the strip beam theory estimates

the values of these constants near the free warping condition
of the cross member in these joints. If these values are used in the
estimation of stresses in the side member or cross member of a joint
then the resulting stresses will be very small as compared to the
practical case. This can be readily checked for any joint by,
multiplying the computed stress value with a ratio of
L To
Iko/T I by strip beam] I by theory]
plate
To
However, the values of K'and obtained by-the other three-methods
show a good agreement between them. In most cases the values com-
puted by the Plate theory estimates stresses up to 9% higher than
by the finite [21]
estimated element method which shows a good
comparison.

4.4. COMPARISON OF EXPERIMENTAL AND THEORETICAL STRESS VALUES

The isochromatic fringes directly give the values of the maximum

shear stress while the isoclinic fringes determine the direction of
the principal stresses in a photoelastic model. In order to deter-
mine the individual values of the principal stresses at a point, an
experimental method known as "Oblique Incidence Method" is usually
employed to obtain another set of measurements involving principal
stresses. Then individual principal stresses are found from the two
measurements. The oblique incidence technique could not be used
successfully in the present work as reflections from the model surface
made the distinction of colours unreliable. The analytical method
known as-"Shear Difference Method" which is based on the numerical
integration of the stress equilibrium equations was tried to separate
the principal stresses. It is a very cumbersome method and can lead
to incorrect results if'the input data is not estimated accurately.
The isoclinic fringes being thick black bands could not be estimated
with a fair accuracy on small models without a photometer. The
presence ofýresidual birefringence on the edges further made this
estimation difficult and so this approach was abandoned.

The failure criteria of a material subject to a biaxial state of

stress depends on the value of the maximum shear stress rather than
the maximum direct stress in the material. The stresses occurring
in the joints, consisting of thin walled beams, are not uniaxial but
combined and so maximum shear stress becomes the most important stress
regarding their failure. Considering the importance of the maximum
shear stresses in the joints it was decided to make a comparison
between their theoretical and experimental values at the critical sections.
 72

The isochromatic fringes provided their values directly. The

fringe order at a point lying on a stress free boundary directly
determines the value of the tangential principal stress as the
normal stress is zero. The edges of. the side member and cross
member flanges being free from stresses made it easy to compare
direct stresses along free edges. The values of the cartesian
shear stress component T12 along the critical section on the side
member flanges were also compared with theoretical values. '

The maximum direct' ' stress for the complete warping inhibition
condition at the built-in end of a channel section beam shows a
good agreement with theory as shown in Fig (3.5). The experimental
value is smaller than the theoretical value by less than 52. The
experimental stress decreases more rapidly than the theorefical
stress as the distance from the fixed end increases.

Figs. (2.5a) and (2.5b) show that the flange/web corners do not behave
as simply supported edges but they transmit moment m22 and also undergo
bending displacement u3.

It was found that the couple loading produced stresses/displacements

of the opposite nature than the cross member warping forces in the
side member thus reducing the combined effect. But the magnitude of
stress/displacements produced due to the couple loading was found to
be very small in all the joints analysed in this work. Their contri-
bution was less than 2% on the Joints I, II and IV, but up to 10% on
Joint III. It was also found that the affect of the couple loading
on the distribution of stresses/displacements in the side member was
larger on the web than the flanges of the side member in a joint. he
effect of the couple loading was considered in the evaluation of av
and K in Joint III and the ladder frame joints but was ignored in the
comparison of stresses on the flanges of the side members.

4.4.1. Isolated joints

i) The comparison of the maximum shear stress values on the side

member flange shows that the experimental values lie between
the estimated partial warping and zero warping values. in the
critical region as shown in Figs (4.1) to (4.4), they do not
indicate a regular trend near the free edge of the flange.
The experimental fringe values in this region were very small
and were affected by the presence of residual birefringence.
The location of the critical area predicted by theory agrees
very well with the experimental results and the maximum estimated
stress value is within 6% to 13% of the experimental value
except for Joint II.

ii) Figures (4.5) to (4.8) show that the comparison of the shear
stress T12 on the side member flange also follow the same
pattern as observed for the maximum shear stress on the same
section.

iii) The experimental distribution of the direct stress on the

free edge of the side member flange indicates the same trend as
 73

as predicted by the plate theory This is shown in Fig. (4.9).

.. .
to (4.12). The experimental values lie close to the partial
warping curve for joints III and IV while Joint II does not
compare well. The Joint I also shows closer comparison with
partial warping values than zero warping values. The theore-
tical estimation of the maximum stress region agrees very well
with the experimental one. The length of the side member
influenced by local effects of a joint can be approxi-
portion
mately determined with reference to the position of the cross
member flange zero warping line. This length extends on both
sides of this reference line equal to the distance of the sum of
the widths of the cross member and the side member flanges.

iv) The direct stress along the free edge of the crass member
flange in Joints I and III shows a better correlation with
the partially warping case near the joint than with free warping
zero warping conditions as shown in Figs. (4.13)and (4.15)
or
But Fig. (4.14) indicates that the values of direct stress
in Joint II are closer to the zero warping curve.

The comparison of the distribution of various experimental stresses

the the cross member portions of the joints I, III
on side member and
and IV described above clearly indicates that the plate theory can
estimate their theoretical distributions very accurately. However,
the values of the stresses can be determined within close approxi-
mations by considering the partial warping of the cross member. The
regions of the maximum stress on the side member are accurately deter-
mined by the plate theory. Joint II has not shown a good agreement
with theoretical values. It seems that the permanent distortion of
the section caused during assembly of the section is responsible for
these errors. Also the joint has a small value of K so the actual
in [22]. But
stresses approach zero warping condition as reported
the other joints do not confirm this trend although their K values
are slightly higher than Joint II.

4.4.2. Ladder frame joints

The values measured by the

photoelastic method in the joints
of stresses
of the ladder frame required some comparison with the estimated values.
It has been shown in the isolated joints that for known values of the
applied torque on the free end of the channel section representing a
portion of the cross member in a joint, it was possible to estimate
values of stresses analytically.

In the case of the ladder frame testing, the total torque applied to
the whole frame was known but the distribution of the external torque
in individual members of the frame was not known. It is difficult
to the share of the external torque in every member of the frame.
assess
As the cross members were attached to the side member web, so the
cross member offers a very small restraint to the warping of the side
In such joints, the warping inhibition of the cross member
member.
to in joint [11]. As first
mainly contributes the stresses a a
approximation in solving the underlying problem of torque distribution,
it was assumed that the external torque applied to the ladder frame was
distributed entirely in the cross members of the frame. It was further
 74

assumed that the torque was shared by all the cross members of the
frame proportional to their flange widths as their web heights were
equal. The resulting values of torque were used to estimate stresses
in the inner joints of the frame.

It was interesting to see that the experimental values of the maxi-

mum shear stress on the critical section of the side member flanges
in two identically loaded joints of the frame showed a good compari-
son with the theoretical values. The agreement for the location of
the maximum stress region was even better which confirmed the assump-
tion that the cross member warping inhibition forces were the main
contributing factor. Regarding the assumption about the distribution
of the external torque in various members of the frame, it proved
satisfactory for the joints in which the cross members were only
attached to the side member webs and the ratio of their web heights being 0.65.

The comparison of the tangential stress on the free edge of the

cross members showed that the estimation of partial warping inhibition
in the joint predicted stress values which compared better than the
zero warping or the free warping conditions. Small differences in the
measured values of stresses on the two identically loaded joints in
the frame were observed. These could be due to the minor difference
in the end conditions in the support brackets.
 75

It'
ý'I: iE 4.! ý. h& I Iv
i`Il SI
 76

i,, ý... ýiý. il i, iýýi

 77

air

(h) Half order 1 rin,; cs

o') (>rdcr 1"111". <'s wit ... ý,

Fringe feit tern:; uii till

 78

l7
tu

(1)) 1 ill!
i; iOC C'1i
130

(c) 145' !; (, (" !iH. ("c;

I' ate 4.4. Isoclinic f he ! (,; H'1n!

, riný; c ýmt Lcrns on iý r" ('! ()i lit !! )
 79

SI CI11lic3

.I,,

Plate 4.5. Isec,, inie f riugý the ?eft

p.-ii I( ti:: un 1 (.!,, inn I, )
 80

3ý I3°i OC in 1ý'r'

isocI ii1_Cs

\, I)Cti n* c

Plate 4.6 Isoclinic fringe patterns flange

on the right
(Joint II)
 Plate 4.7 A mixed fringe the flanges (Joint
pattern on III)

Plate 4.8 ?fixed fringe patterns on the side member weh

(Ladder Frame Joints)
 i

ýýýi

Plate 4.9 lsochromatic fringe patterns on the ladder frame

joints
 83.

APPLIED 70, PQVE = 25Co N. mm.

_ ? Eýn ý inn o....

fi

N
10

Ic

Ný ip

0 10 Z0 30 40 So

.P/STANCE BROM r1le ýQEE ED4E OF FLAN647 /N 771%rl, ---ý!

Fig. 4.1 The distribution of maximum shear stress on the side member
flange at Section CC. (Joint I)
 84

APPL/E. D TORQ&e _ /2 O N" ý? »"

j' ZERO WARP/NG

o PARTIAL WARP/NG

" EX PER, MENTAL

IOO

9a

so

70

co

N
O

So
qf

40

ti

Oc 30

2o

/0

0
0 10 20 30 40 50

1)/5TANCE FQOM TN, FREE EDGE OF FLANGE IN 7WP7

.i

Fig. 4.2. The distribution of maximum shear stress on the side member
flange at Section CC. (Joint II)
 85

APPL/tI' TORQUE = 2Soo N. "'""

+ ZERO WARP/, VG
o PART/AL wARO/NC.
"E XPER/MENTAL
im

J3;

121
.

/O;

9c

Itli
ý 6c
i
ýF

¢5

30

/S

0
D 10 20 30 37

DISTANC. E FROM THE FREE EDGE OF FLANGE /1/ 711M-

.

Fig. 4.3. The distribution of maximum shear stress on the side member
flange at Section CC. (Joint III)
 86

APPLIED 7-0R6J UE c 2500 N . nrn

. .
" ZEkPO WARO/NG
PAQ -r/.
41- WARP/Nr.

EX PER/MENTAL

ISo

as

124

.+ /o5

ö
96

j 75

ti

13- 60
ý

Ot
Z
¢S
y

30

/5

0
o 10 20 30 40 50
. .

DIS -ANCE F4POM THE Fh?EE EDGE Or' FLANGE

. /N mr». --ýý

Fig. 4.4. The distribution of maximum shear stress on the side member
flange at Section CC. (Joint IV)
 87

+ ZE-PO WARPING

0 PARTIAL /NA, 'P/nNG

" EXPER/MENTAL

/So APPL/Ep TrRaLE

2560 v, »m.

135

/20

/o

C70 1
1

75 "
"

i
n
ýO 60
0

45

30

15

0
oZ
10 20 30 40 So
o

J)iSTANCE AROM THE EDGE OF FLANGE 9V ?11rrl.

-cQEE

Fig. 4.5 The distribution of the shear stress on the side member flange
at Section CC. (Joint I)
 88

APP. Z TO, QQUE /Z 90

, N" mm.
f ZERO ! /ARP/NG
D PART/4L LV RP/NG
I0

9a

$c

iG

60

NO

50
k

40

H
H
30
1
ti
V)

20

U)

I0

o
0 /O 20 30 40 SO

p/STANCE FQOM THE Fe--4 E,D4E OF FLflivat iN -

m2? --
,

Fig. 4.6 The distribution of the shear stress on the side member
flange at Section CC. (Joint II)
 89
ýF

APPL/f. D TORQUE= 2500 N mm

iý
ýý ZERO «ARp1ivC,
Uý +
RW R771AL WARPINr.

/50 ExPER/MEN TAL

/35
4.
ý
/-i-
\
ý/
/20 /
/
Ti

/o5

/
90
/
/e
/
/"
75 I.

60 /
/
/
/
45 /
e
/
30 e
1

/5

D
0 io 20 30
.
P/STANCE FQoM THE FREE EDGE OF FLANGE iN rnm. -º
.

Fig. 4.7 The distribution of the shear stress on the side member
flange at Section CC. (Joint III)
 90

APOLIED 7"agouE = 2500 N"rnm.

f ZEAPO W4RP/Ak

0 PAPPT'/AL /N/1 RR/NG

ý
aý

N
O

b
H
W

.pis rANcE FRo, 7 7/E FREE EDGE of FL. 4Nre iN YYiº11' -"

Fig. 4.8 The distribution of the shear stress on the side member
flange at Section CC. (Joint IV)
 91

0
H

4J
i;
0
ti

Q3ý o'
o\
\ Qi
CA 3. g
O
" ýC
\ iJ
vex u
na tu v
h\ Co
u
o0 eu

"
00

0'
w
®\
v
19
O\ v
O
b
i
7
0
-i-.
1.ý
0
. -

O
h
pý 0
ýe
41
u
N
F
o 10

0)
4-1
Aj
W
0
A
0
"r1
iJ
F

ä N
o . rl
v ý 'd
C1
W
H
v
e a
h

i W
OD

bO b 0
0ý
ao cý
i
 92

60

k
0 ~7

40

.ýF 4b 0
4 ýc
' ,.1
o u

41

"\
4OO 4)
tko
\ a
do cd
0\ 44
44
It

40

C)

/ C)
Y

- "
4ý

1 ý
0
O 10

` G)
r

O
m
"
O

ý W
p

ö v
\ '°
`\"
H
\
V
\ O

q "a
w

r h O n ') a lý lo ºý
lö 'V MN 1I
a11

"el/ O/ x) IV' /2 5s' LS 1: p3ö/Q'

. *--(
r-
 93

ý
IK3O
\ H
\ H

1i
\
\
Ö v

Its
I

Aj

41
WQý
ehýW
C3 to

to" "} -4
w
" $4

H " \ý

r
" b

O
---ý Cl)

CJ
"
N
ýe ýO

4.'

*" u
\. ai
\ 41 .d
u
"
44
S 0
0

$4
oO i+

O
H

y
r4
Ä

oD
P4

0 <1 aOopO00 O
qz 00 o to
ý`I n\N ýj pp ` c(nh
cý
1111

ýý
C1 x
ýuýuiý%N t .
 94

O
4
H
Li

INT 0
v
o
a
O

N Qi

0
ä a
v2 O Li
ý
Q u
\ cn
Li
\ t°
co W O

QN Q '`
w
a o "
N

O
0\ 'd

s
gi
GJ

y
Oý

"/ U,

o :J

/ý "ý
+ "

w0

1
$4
Ö

aý

e
CV

\ ºQ, GO

40 M ý
aý O- I
1 1
 95

NO

. vr ý/ Xss ßö'1 s1 J3? J/O

,Zruw/Ar z
 96

4 H
H

0
ti
O
C1
OA

44

w
o ,n
Q- ý
E
U)
En
0
ov
00
N

w
0
C
b
E C'
C!
GJ
zH

ou

Z +''
w ap

t'
end

ä
p
41
wN
0
F-

i 01

fit b
ti

w
Wo
u
0
H 41

oA "ý
N

aý
0

aD

0w

4ý'c'ui/ r) n.i
ion S53&1 S1 vgb'lq-
 97

Z-

o°°Ö eýýc O
A Z% V) 11113
le ' 4. M cX

('6 ýnissP1W0
°ýX% nor Ss3ýLS 1 äb/Q
_Ut/n'
 98

AooL/ED TOeQuE = 4ooo Al-mm

zERO W4 'P/NG
_+ -
PARTIAL wARPlov
-0-
LADDE4P FR4ME' J041Y7 :Q
-i-
FRAME' IC
LADDER JOINT
---s--

200

/
Ný
X50

.
N
ýO
ýý

/
/00
ti 1
w
H
o:
ti
h
75 ii
C
W
y

J 5o i

ss ýýý

01rr

0 20 30 40

D/S rANCE r, POlº7 rsNE "EPEE ED6f OF FL AIV60 " MM.

,

Fig. 4.16 The distribution of maximum shear stress on the side member
flange at Section CC. (Ladder frame joints)
 99

""ýiO
a
9
w

v
b

(1)
on

a)

Ei

O
U

C1
4i
i-ý

W
O
Cl

G1
G)
4!
i+
44

0
L 0
y
C1
14
h° N

ti

ö 4J
U

10
W
u
Z
0

h 0
'9

y
.. a
10
C)
H

r.

cc
(ýn iss3. ýdivo J)
ýZu'w; /N
of xý Nr SS3a'1s j
z
 100

CHAPTER 5 CONCLUSIONS AND FUTURE APPROACH

5.1. CONCLUSIONS

The present work has mainly shown that the plate theory can be
successfully applied to analyse the state of stress in a chassis
frame joint of a channel section cross member directly attached to
the web of a channel section side member. Previously the applica-
tion of the plate theory was restricted to the analysis of joints
consisting of only symmetric I-section cross members attached to the
web of the channel section side members. But now joints consisting
of asymmetric I-section or channel section cross members can be
effectively analysed. The contribution of the channel section cross
member web warping forces on the side member in a joint has been
included in the present analysis.

A close comparison between the experimental and the analytical

estimated values of the maximum shear stress at the critical section
of the side member flange clearly indicates that the consideration of
partial warping inhibition in welded joints is justified. The degree
of partial warping inhibition of the cross member in welded joints can
be estimated with a fair accuracy by the proposed analytical approach.
It has been shown that the average bending displacement of the side
member web can be related to the partial warping of the cross member
end in the joint. The relationship between the partial warping
beam [171
constants of the strip theory and the finite element method
t211 force
can be used to estimate the cross member warping exerted
to the side member web in a joint. The results have shown in most
cases that the prediction of stresses in various members of the joint
on the basis of the present approach gave better estimation than other
methods.

The effect of the cross member torque on the stress distribution in a

side member has been analysed by considering the warping forces and the
shear couple forces of the cross member separately by using the plate
theory. It was found that the contribution of the couple loading on
the stresses in the side member was very small as compared to the cross
member warping forces.

The other information provided by the present work can be described as

follows:

i) The photoelastic models as employed in this investigation can

be used to determine stresses in thin walled open section
beams. For problems in which a combined state of stress is
expected, photoelastic analysis will lead to the accurate
location and measurement of the critical stress.

ii)The critical section where the maximum shear stress attains its
peak value in the side member portion of a welded joint lies
on the flange in line with the zero warping line on the cross
 101

member flange. The critical region lies on this section

near the flange/web corner. The point at which the maximum
stress occurs can be located by the present analysis. The
distributions of other stress/displacement components on
the channel section side member can also be determined by
the present analysis.

iii) The length of the side member portion affected by the localised
effects of the joint extends on both sides of the joint
approximately equal to the sum of the widths of flanges of
the cross member and side member.

5.2 SUGGESTIONS FOR FUTURE APPROACH

i) A more sensitive photoelastic model material should be used

in future work to obtain better fringe patterns.

ii) The present mathematical model of the joint can be modified

to include the effect of other forces acting on the side member,
e. g. the longitudinal forces described in [19]. The photoelastic
model of the ladder frame used in the present investigation
can be used to determine the effect of such forces on the frame.

iii) The present analysis assumes that the cross member is attached
with its web vertical to the longitudinal axis of the side
member. It is possible to modify it to consider any orientation
of the cross member web. It should be modified to include top
hat section cross members. This can be achieved by adding terns
in the present analysis to represent the warping effect of a top
hat section. The warping effect of each part of the section can
be represented in a similar manner as shown for a channel section
in the present analysis.
 102

REFERENCES

1. VLASOV, V. S. "Thin walled elastic beams"

Israel Programme for Scientific translations,
Jerusalem 1961.

2. DOICH, I. "An approach to the calculation of an automobile

frame in torsion"
Avtom. Prom. No. 6,1961 (in Ruessian)

3. GELF'GAT, D. B. "Commercial Vehicle hassis frames"

and MASGIZ, 1959 (in Russian)
OSHOKOV,V. A.

4. HANKE, M. "Theory of the Monoplane joint (from Thia-walled

beams in vehicles")
Part II, Automobile No. 8,1959 (English transla-
tion of Czech original).

5. ZAKS, M. "Calculation of torsional stiffness of commercial

vehicle frames with flexible joints"
Autmobile Industry, No. 4,1964 (English transla-
tion of Russian original).

6. ZAKS, N. Nand "Investigation of soft the joint layout in a twisted

BELOKUROV, automobile frame"
V. N. Avtom. Prom. No. 5,1971 (in Russian)

7. ZAKS, M. N. and "Control of torsinnal stiffness of the automobile

BELOKUROV,V. frame"
Avtom. Prom No. 4,1972 (in Russian)

8. ZAKS, M. N. and "Bimoment equation in a soft joint of an automobile

BELOKUROV,V. frame"
Avtom. Prom. No. 5,1973 (in Russian)

9. ZAKS, M. N. and "Calculation of short open section thin walled

SAHAROV,A. A. beams in Automobile frames"
and Avtom. Prom. No. 6,1974 (in Russian)
BELOKUROV,V.

10. ZAKS, M. N. and "Investigation of the deformation of the chassis

ZAHAROV, A. A. frame members".
Avtom. Prom. No. 5,1975 (in Russian)

11. ZAKS, M. N. "Stress state in joints of a twisted automobile

frame"
NAMI Proceedings, No. 61,1963 (English translation
of Russian original)

12. KOBRIN, M. N. "Investigation of the stress state and durability

KILIMNIK, of (vehicle)frame side member walls at the point
L. Sh and of load transfer" Avtom. Prom No. 11,1969. (English
TITOV, A. A. translation of Russian original).

13. TAKAIIASHI, K. "A torsion strength analysis of truck frame using

open section members"
SAE No. 710595,1971.
 103

14. ALI, R. HEDGES. J. L "The application of finite element techniques

and MILLS, B. to the analysis of an automobile structure.
Static analysis of an automobile chassis frame. "
Proc. Inst. Mech. E. Vol 185,1970-1971.

15. HEDGES, J. L., "The application of finite element to the

NORVILLE, C. C. analysis of an automobile structure. Stress
& GURDOGAN, 0. analysis of an automobile chassis frame. "
Proc. Inst. Mech. E. Vol 185,1970-1971.

16. ALI, R., HEDGES, "Dynamic analysis of an automobile chassis

J. L. & MILLS, B. frame", Proc. I. Mech. E., Vä1.185,1970-71.

17. MARSHALL, P. H., "Torsional stiffness of commercial vehicle

ROACH A. A. and chassis frames. "
TIDBURY, G. H. F. I. S. I. T. A. XII Congtess, Barcelona, 1968.

18. AWUDU, G. "Warping inhibitions in the joints of vehicle

chassis frames. "
" A. S. A. E. Thesis, Cranfield 1968.

19. TIDBURY, G. H. i) "The torsional stiffness of an open section

thin walled beam in terms of bimQments and the
generalisation of partial warping inhibition
boundary " Motor & Veh. Conf. Belgrade,.
conditions.
1974. ii) Thin walled beam theory in commercial
vehicle design.
-Unpublished report. of School of Automotive
Studies, Cranfield, 1975.

f' 20. AZODI, K. D. "Effects of longitudinal loads on the stress

distribution of open sections. "
M. Sc. Thesis, Cranfield Institute of Technology,
1977.

2 21. MEGSON, T. H. G. "Structural analysis of ladder frames under

and ALADE, G. A. torsion"
Proc. Inst. Mech. E. Vol 190,1976.
"

? 22k. ALADE, G. A. "Structural analysis of ladder frames under

torsion". [
Ph. D. Thesis, University of Leeds,, 1974.

23.. MEGSON,T. H. G. "Extension the Wagner Torsion

of Bending Theory
to allow for several systems of loading. "
The Aeronautical Quarterly, Vol XXVI, August,
1975.

24. IVANOV, A. A. "Calculation of automobile frames using finite

element methods"
Avtom. Prom. No. 4,1973, (in Russian)

25. IVANOV, A. A. "Application of finite element method in the

calculation of semi-trailer frames. "
Avtom. Prom. No. 8,1974 (in Russian)
 104

"Calculation of lorry frames, taking into

iC 26. ROMANOW, F.,
SLOMKA, A and consideration the theory of restrained
ZABLOCKI, W. torsion. "
Politech. Wrochawskia, IKEM 1976 (in German)

27. BEERMANN, J. "Calculation of torsion of weak commercial

vehicle frames taking into consideration twist-
restraint nodes"
Automobile Industry, 1976 (in German)

/ 28. KOBRIN, M and "Determination of Optimum parameter values

GORBONOS, M. of truck frames by analysis of the existing
designs"
Israel Institute of Tech. Haifa, May 1976.

y 29. LASEVICH, L. G. "Selection of optimum sections of commercial

SKOLNIKOV, M. B. vehicle chassis frames"
Avtom. Prom No. 2,1975 (in Russian)
and PODLEGAEVA,
T. D.

? 30. TIMOSHENKO, S. "Theory of Plates and Shells"

WOINOWSKY- Edition 2, McGraw-Hill Book Co Inc., 1959.
KRIEGER, S.

? 31. MEGSON,T. G. H. "Linear analysis of thin walled elastic structurs"

Surrey University Press, 1974.

32. CARVER, G. C. "Truck chassis frame considerations in equipment

mounting"
S. A. E. No. 760291,1976

33. ZANDMAN, F., "Reinforcing effect of birefringent coatings. "

RENDER, S. S. & Exptl. Mech., Vol. 2., No. 2,1962.
RIEGNER, E. I.

34. DUFFY, 'J., and "Measurement of surface strain by means of Bonded

" LEE, T. C. Birefringent Strips".
Exptl. Mech. Vol. 1, No. 9,1961.

35. ALVI, M. S. I. "Photoelastic analysis of perforated (circular

holes) tubes under axial compression"
M. A. Sc. Thesis, University of Toronto, 1967.

36. HENDRY, A. W. "Photoelastic Analysis"

st. Edition, Pergamon Press 1966.

2 37. DALLY, J. W. and "Experimental stress analysis"

RILEY, W. F. McGraw-Hill Book Co, 1965.

38. Instruction manual for large field universal meter.

Bulletin PS-3043, BUDD, S. A. Comp. U. S. A.

39. ROACH, A. H. "Warping inhibition in commercial frames. "

i vehicle
A. S. A. E. Cranfield, 1966.
 105

40. "Process for Casting Araldite CT200 Sheets for Photoelastic

Purposes. " Instruction Sheet. Sharples Ltd.

1 "
41. "Materials for Photoelastic Coatings and Photoeslastic Midels.
Bulletin P-1120-1 Photolastic Inc. Pa. U. S. A.

42. "Notes on a technique for fabricating Araldite Models

from Araldite CT200 Sheets. "
constructed
Instruction Sheet, Sharples Ltd.

0
 106

APPENDIX I

GENERAL BENDING PROBLEM OF A RECTANGULAR PLATE

Consider a rectangular flat plate of uniform thickness h referred

to a cartesian co-ordinate system as shown in fig. (2.1a). The xix2-
plane coincides with the mid plane of the plate. The plate is
subjected to some forces which are perpendicular to its mid plane
only. The plate will bend due to the action of these forces.

The deflection of the middle surface of this plate can be determined

by taking into Kirchhoff's [30].
account assumptions

The expressions for moments "ý-" and vertical shearing forces

"qi" are given below in indicial notation.

+h/2 (i 2)
Moments -'iJ = Tij X3dx3 (AI. l)

-h/2
1 iJ ° 1: 2)

+h/2
Vertical Shearing forces J (AI. 2)
qi ridx3

-h/2

Displacement, strain and stress fields can be expressed by the

following equations

ui (X1, X2"X3) ui(X1, X2) - Xaa3(X1, X2)

1
(AI. 3)
u3 (x19X29X3) ° u3(xlsx2)

e.. = eil -x (Al 4)

13
3Ki. .

where

0
C.. (uo '+,. e
lj 1,3 J, i

K.. u3, ij
lj

T.. E 1
lj {(1_v)C?. + vsijekk)-{(1-v)K1j+v6ijKkk)x3] (A1.5)
(I--v2)
3.j

Substitution of equation(AI. 5)in equations(AI. 1)gives the following

expression for moments:

m. D[ (1-v)K. + va. Klck

ýj lj ij
C (1-v)u3, 3
-D ij + v5.. u3, kk (Al 6)
.
 107

It is easy to obtain the following expression for qi after

integrating the equilibrium equations in the absence of body forces.

(V2u3) (AI. 7)
q:i _D -

where D Eh3 is called the Bending Rigidity of the plate.

12 1-v )

The resultant vertical shearing forces become

vl Q1 p12,2
(AI. 8)
V2 42 m12,1

The positive signs for moments and vertical shearing forces are
taken according to fig. (2.1b).

The governing equation of the bending problem of a plate in the

absence of the external forces can be expressed as

Vtu3 0 (AI. 9)

This equation has to be solved for any particular problem and the
boundary conditions have to be satisfied. The external forces
have to be accounted for while satisfying the boundary conditions
of the problem.

RECTANGULAR PLATE WITH TWO SIMPLY SUPPORTED EDGES

The bending problem of a rectangular plate with two simply

supported parallel edges can be solved by assuming a solution of
the governing equation (AI. 9) in the form of a single trigonometric
Fourier series as
cc
(where L) (AI. 10)
u3(xi, x2) _I m(x2) sin a am
m=1

Equation (AI. 10) satisfies the boundary conditions along the two
simply supported edges of the plate as shown in fig. (2.1d. ). The
`boundary conditions along the two edges can be written as below:

1 Along the edge xi a0

mil= 0)

(AI. 11)
u3 0)

2 Along the edge xl = 2L

mil= 0)

U3 =0
 108

Equation (AI. 10) should now satisfy the governing equation (AI. 9),
which takes the following form on substitution of equation (AI. 10).

2_W
2Ma + am Wm a0 (AI. 12)
im
aT 22

Equation (AI. 12) is a fourth order ordinary homogeneous differential

equation whose general solution can be written in the following form.

W (x2) _ c1cosham 2+ c2x2coshamx2 + c3cinhamx2+ c,ýx2sinhamx2

or W (x2) = ci41+c20 + C4 04 (AI. 13)

3+c3¢. 2

where cosh aý 03 = x2cosh amx2 )

2
(AI. 14)
02 sinh am 04 ax2 sink amx2 )
2
)

cl, c2, c3 and cy are constants which have to be determined

from the four boundary conditions along the edges x2 =0 and
x2 = 2H of the plate.

Once Wm is known, its value can be substituted in the relev rnt

formulae to obtain the values of plate deflection, moments, vertical
shearing forces and resultant vertical shearing forces. If the
bending moment m22, the angle of twist along x2 i. e. O= 8u3/3x2
and the resultant vertical shearing force v2 are assumed as single
trigonometric Fourier series of the following forms:

m22 (x 12,x)_1 : Z2m(x2) sin amxl (AI. 15)

m=1

9(xl, x2) _i Om(X2) sinamX, (AI. 16)

m=1
i V2m(x2)sinamxl (AI.
V2(x1, x2)= 17)
m=1

The corresponding expressions for 022,9 and v2 can be obtained

in the following form on substitution (AI. 13) in the
of equation
relevant formulae
a

m22(xl, [2amc4
x2) _I -D +aß(1-v)c1] ý1 +a2(1-v)c2ý3
m=1

+ [2aMc2 + (AI. 18)

a2(1-v)c3102 +am(1-V)c444 sin amx1
cc
O(xl, x2) C(c2.
_1 + amcg)O1 + ania403+(c4+aý c1)02+a c2041sinax1
m=1 m

(AI. 19)
 109

a
(7-19N2) ýIiamD {ý(l+v)c2- aß(1-v)cýý1 -am(1-v)c4ý3
V2

[(1+v)c4-am(1-v)ci]42- }sin (amxl) (AI. 20)

am(1-v)c2c

Fourier M2m(x2), 0 (x2) V2m(x2)

The values of the coefficients and
(AI. 15) änd (AI. 18),
can be determined by equating equations
(AI. 16) and (AI. 19) and equations (AI. 17) and (AI. 20)
equations
respectively.

The equations for deflection, moments, resultant shearing forces

be in terms of four initial
and angle of twist can represented
parameters of the plate defined along the edge x2 0ý
0000
The initial parameters chosen are W, M2,0 and V2, and their values
can be obtained in the following form by substituting x2 -0 in
in the expressions for Wm(x2); (x2), om(x2) and V2m(x2)
x2m
W= )
ci
R2 2amc41 )
_D[am(1-v)cl+

o) (Al. 21)
0 c2 + amc3 )

Vol -D[a2 (1+v)c2 a3(1-v)c33 )

-

'Equations (AI. 21) can be solved for cl, c2, c3 and c4 in terms of
the initial parameters giving the following relations:
0
cl W)

C2 -t2 V2
2
(AI. 22)
(l+v)0
c +10)
3 2p 3D 2)
2a In )

am (1-v) °
c10_ M22W
4 2ahD

Now the Fourier coefficients W (x2), 0 (x2), '2m(x2) and V2m(x2)

can be expressed in terms of tle initial parameters by the'following
equations after substituting the values of the constants c1, c2, C3
and c4 from equations (AI. 22).

W(x2) [01
-aam041W +Cb. + ao3]O
m
0
JM2
_E mm + ý.
am ý02 (AI. 23a)
 110

00 Di
0 (x2) i-
Ebam02 - aal 0 3]W + + aam410
=

am-+031M2-t
'j°' 14 'j ° (AI. 23b)
J V2
m

00 [n$2+ a43Je
(x) [dam44]W
M _ -
2M-2
0
f [el +aam44JM2 +2V 2
(AI. 23c) ý

00 [damp
V2M(x2) [n4ý2- am43JW + 0

Eý1 (AI. 23d)

+[ bam42- aa2431M2 + -aam44JV2

where
1+v
a2

1-v
b =2

6(1-v2) (AI. 24)

""-'E

n Eham(3+v)
24(1+v)

Eh3am(1-v)

-24(1+v)

are constants.
 111

APPENDIX II

GENERAL STRETCHING PROBLEM OF A RECTANGULAR PLATE

Consider a rectangular plate as shown in fig. 2.1a being subjected

to forces which are acting in the plane of the plate. The plate
will deform in its own plane. This problem is known as a stretching
problem.

This problem can be formulated interms of displacements from the

basic equations of theory of plates which are summarised below in
indicial notation.

1- Equilibrium equations in Xi direction: :

ý;

nij, + si + gi =0 For i-1,2 (AII. 1)

j
j-1,2

where h
at 2
S.1 = T33

at-h
f

+h
gi °J Fidx3
h

Fig(2. lc)shows. the sign convention for in-plane forces on the plate.

2 Constitutive equations:

B [c? (1-V) 3 (AII. 2)

nib . +6 ii Ekk

where
E1-v) is known as Plate
B= Stretching Rigidity.

3 Strain displacement relations:

1+ (AII. 3)
eo uý
ij

where ü and e°. are displacements and strains of the mid plane
of the plate respectively.

4 Compatibility equations:

0 (AII. 4
e11922 + C22f 11 - 2C129 12 00
 112

Substituting (AII. 3) in equation (AII. 2) and then the

equation
resulting equation in equation (AII. 1), the following system of
coupled equations is obtained:

uls11 + (i-v) uls22 + u2 (AII. 5)

12 0
2 2(l+v) 30

(1-y) 00 (l+v) 0 (AII. 6)

u2 11 + u2922 + ul 12 0
22

These coupled equations have to be solved for any particular

problem subject to the prescribed boundary conditions. The
external forces have to be considered while satisfying the boundary
conditions.

-RECTANGULAR PLATE WITH HINGED SUPPORTS

For a rectangular plate with hinged supports at the two ends and
defined by the following conditions

Atxl=0 andxl=2L

T11 =0 (i. e. no axial force at the ends) )

(AII. 7)
=0 (i. e. twist angle is zero) )

These conditions lead to the following boundary conditions at

the two ends:

At x1 =0 and x1 = 2L

=0 and u2 0 (AII. 8)
1

The boundary conditions (AII. 8) are satisfied if the displacement

components ul and u2 are assumed of the following forms:

ul(x1, x2) _ Um(x2) cosamxl )

.
m=1, )
(AII. 9)

u2(x1, x2) -I Vm(x2) sinamx1 )

m-1

mA
where a= 2L
m

On substitution of equations (AII. 9) in equations (AII. 5) and

(AII. 6) following
equations are obtained.

32Un 3V
-2amUm + (1-v) +(l+v)am axm =0
2) (AII. 10)
4im
2 - (1-v)am Vm - (l+v)am =0
arX
22
 113

Solving the two equations (AII. 7) for Um results in the

following single equation.

4 Um ? Um
- 2a2 + a4 U no (AII. 11)
ax4 m ax2 mm
22

Equation (AII. 8) is an ordinary homogeneous differential equation

of fourth order and its solution is given below:

U(x2)= clcosham c2x2coshamx, + c3sinham c4 x2sinhamx2

2+ 2+

This solution can be written in the following form by substituting

relations (AI. 14).

UM(x2) X141 +" + C02 + 0404 (AII. 12)

243

The constants cj, c2, c3 and c4 are to be determined from the

boundary conditions along the other two edges of the plate. Then
the displacements/stresses/forces can be evaluated on the plate by
substitution of equations (AII. 9) in the relevant formulae.

It becomes easy to evaluate various results if the initial parameters

of the plate are specified along the edge x2 = 0. It is necessary
to express all the resulting equations in terms of the initial para-
meters. Fourier coefficient Vm(x2) can be expressed as
[c1+11+v ýa
Vm(x2) =I c2+c31c1 + c4] c443+c244 (AII. 13)
J+v, x 2+
mm
Force components n22 and n12 can be determined by substituting
equations (AII. 9,11 and 13) in equations (AII. 1) and the following
results are obtained:

2(1+ýý [am(1-v)c3
n22(xl, x2) By - c4]ý1 + am(1-v)c2ý3+
ý1([an(1-v)cl

2(1+v-)
42+ am(1-V)C404) sin amxl (AII. 14)
21

aac- (1-v) lm (1-v)c

(x
n 12 121m3
x) - B(1-v) +c- 1
'j +acm 4'3
. l+v 2J41 l+v 4Jý2

)cos
+ amc2$L am 1 (AII. 15)

Equations (AII. 14 (AII. 15 in

and can be expressed simple forms as

n22(x2, x2) =L N22m(x2) sin amxl

m=1
(AII. 16)
and n12(xl, I N12m(x2)
x2) cos amxl )
 114

The Fourier coefficients N22m and N12m can be evaluated by

comparing the coefficients of equations (AII. 16) and equations
(AII. 14,15) for N22 and N12 respectývely. Four Anitial parameters
o
V, N22
are prescribed at x=0 which are u, and 12. Their values
can be obtained as follows by substituting x2 =0 in the respective
Fourier coefficients:
0
U cý )

0 [(V-3
II c+2 C31 )
a (l+v)
m (AII. 17)
N22 B[ aU(1-v) cl - C4
(1+vv)

0 r j
N12 = B(1-v) lamc3- (1+v) C2 )

Constants cl, c2, c3 and c4 can be expressed in terms of initial

parameters as follows:
0
cl u

(1+v)2 aß(1+v) 0
C2
2 N_ v
2Eh 12 2
(1+v) (3-v) 0 0 (AII. 18)
C3 N12 -(1-v)
2Eha 2V
m

a (1+v) U (1+v) N22

c4 in -00
2Eh
2

Now Fourier coefficients Um(x2), Vm(x2), N22m(x2) and N12m(x2)

can be expressed in terms of the initial parameters through
equations (AII. 18) in the following forms:

[ci+bame4JU-10 [ae2+bame3]v 00 [ce4]N22

Um(x2) _ - -

d-2' (AII. 19a)

+[ +c 431N12
a
m

V'(x2) [-ae2+bamý3 ]U +[41-bamý] +[ä -cý31N22

m

+[ ]N12 (AII. 19b)

ceq

[f 0O [ý1-b
N22m(x2) amý41U +Efý2- famg3]V + m$41N22

[02+ 0
+ bamc3] N12 (AII. 19c)
 115

[f42+ 000 [famý4JV

N12m(x2) famý3]U - +ýa¢2- bamý3]N22
_

[ý1+ býý4I°N12 (AII 19d)

+ .

where
1-v )
a2

1+v
b 2) (AII. 20)
(1+v)2 )
c= 2Eh )ý.
(1+v)(3-v)
d
2Eh
Eham )
f=
2)

nll(xl, x2) can be expressed in the following form:

N11m(x2) sin amx (AII. 21)

nll(xl9x2) L
m=1

where Fourier coefficient Nllm(x2) can be evaluated in terms

of the initial parameters as given below:

3+v) 0
Nl lm(x2) Eh -[ a Uo +V N22J ý1 + ra-
[' V- N 121 ý2
m Eh 2Eh
2
bam amt bam-
+[2o- N 12ý $3+[ U+ 2 2h+
2 Eh 2 Eh

(AII. 22)
 116

APPENDIX III

FLOW CHARTS
 117

PLATE BENDING SOLUTION

INPUT

INITIALISE
U3 °0
M22' 0

IS
N >,17

NO

WARPING FORCE
TERMS

INITIAL PARA-
METERS OF
THE WEB

DISPLACEMENT
AND MOMENT
Ua. M

OUTPUT
U3 1122
9

END
 118

PLATE THEORY SOLUTION

INPUT

INITIALISATION

Yes
IS
N>- 17
OUTPUT
No
- --------- ~ --
FLANGE WARPING OR I.
J
TERMS COUPLE FORCE TERMS
END -------- ---

No WEB WARPING
I0=0 TERMS
Yes

INITIAL PARA-
METERS
OF THE SECTION

CALCULATION
OF STRESS
DISP. & MOMENTS

IS
9>T1 No

es

WARPING
COUPLE FORCE TERMS
CONTRIBUTION CONTRIBUTION
 119

APPENDIX IV

PREPARATION OF PHOTOELASTIC MODELS

Photoelastic models of channel sections were made from CT 200

Araldite by two different methods which are described below.

i) FABRICATED PHOTOELASTIC MODELS

Sheets of CT 200 Araldite were obtained from Sharples Photomechanic

Ltd. They were available in }inch thickness or more as desired.
The maximum size of the sheet obtainable from the above source was
approximately 16" x 12".

A channel section was made by cutting three strips (two for the flanges
and one for the web of a channel section) which were bonded together
with Araldite adhesive to form the required section. Later two
similarly fabricated channel sections were bonded together to obtain
a model öf a joint in which a channel section representing a part of
a cross member is attached to the web of another channel section
representing a part of a side member. The purpose of this exercise
was to studythe feasibility of the method of fabricating photoelastic
models needed in the present experimental work.

It was considered that the photoelastic models fabricated in this way

would save time and labour. But the experience showed that finishing
and smoothing of the straight edges of strips cut from a large sheet
was a laborious job. The dimensional accuracy of a section could only be
maintained if the edges of the strips were perfectly straight and smooth
before bonding with Araldite adhesive. Strips were cut by jig saw
and the edges were later smoothed by filing. During the bonding process
it was noticed that it was difficult layer
to ensure a uniform of
adhesive. in a_joint of such long length (approximately 16in. Some
.a
portion of the joint could be starved ofadhesive thus resulting in an
unreliable joint.

The models made by this fabrication method did not prove very reliable
as joints were broken during testing. Fabrication of channel sections
-of lengths larger than 16 inches was not possible as the supplied
sheet size did not permit this. These limitations led to the decision
to casting channel sections.

ii) CAST PHOTOELASTIC MODELS

The major portion of the experimental work was conducted on photo-

elastic models made from cast channel sections. The success of a photo-
elastic investigation mainly depends on the proper manufacture of good
photoelastic models which are free from residual birefringence. The
process of casting photoelastic models of even simple shapes requires
adequate facilities and sufficient experience in the field. It is
often preferred to use commercially available photoelastic sheets,
blocks or circular bars rather than casting the desired sections.
Sharples Ltd (U. K. ) and Photolastic Inc (U. S. A. ) have specialised
interests in the photoelastic field and they provide consultancy,
design and manufacturing facilities in this field. Unfortunately like
all other commercially orientated business concerns, the cost of such
 120

service is extremely high for academic institutions to bear.

.
Because of this consideration, the bare necessary facilities were
created in the laboratory to undertake the casting of photoelastic
models. After a series of tests and modifications it became possible
to cast photoelastic models of channel section beams with little
residual birefringence.

This exercise took more than a year before giving adequate results. Zhe
various steps involved in the casting process are described below.

ii. a) Choice of photoelastic model material

The choice of a proper model material is very important in photoelas-

tic work and the necessary characteristics for its selection depend
These characteristics described in [373.
oii the type of owork. are

A channel section fabricated from CT 200 Araldite sheet showed good

results in terms of isochromatic fringe distribution when stresses
due to complete warping inhibiton were studied with it. This material
was basically recommended for stress freezing technique. The availa-
ble facilities in The School of Automotive Studies laboratory were not
adequate for stress freezing. CT 200 exhibited a low stress optical
sensitivity in normal photoelastic work and frinpp orders between one
and two were obtained Ia reflection polariscope without distorting
the shape of a model. This material is quite brittle and all machining
operations require careful handling. Small residual birefringence was
noticed on machined edges of the models made from CT 200. Time edge
effect was practically very small over a period of sic months and can
easily be ignored. It did not exhibit appreciable creeping effects
under load over a period of 24 hours. The casting procedure requires
annealing in carefully controlled temperature conditions i. e. the temp-
erature drop should be 2°C/hour 1401.
not more than

Some other photoelastic materials with high stress optical sensitivity

and requiring simple casting methods are available from Photolastic
Inc. (U. S. A. )[_41]. These materials were considered very expensive
for use in the present work. CT 200 was the least expensive model
material for casting purposes and its constituents were available
from Ciba Geigy (U. K. ) in bulk quantities. The commercial names of
these constituents of CT 200 are:

(i) Araldite CT 200

(ii) -Hardener HT 901

The choice of CT 200 as a model material in the present experimental

work wasmainly made because of its low cost and quick procurement.

ii. b) Design of the moulds

A major requirement of the experimental work was to cast channel section

models of uniform wall thickness and this was considered in the design
of suitable moulds.

A very simple mould consisting of an outer U-shaped aluminium shell

 121

with a solid aluminium block was designed and manufactured. The

block when placed in the outer shell provided a uniform gap in
which a channel section of uniform thickness can be cast. The casting
surfaces of the mould were polished so as not to produce scratches
in the casting. One end of the mould was closed by an aluminium end
plate and two aluminium spacers were inserted along the mould flanges
Molten Araldite and Hardener mixture was poured in the cavity and
the mould was kept in vertical position. The leakage of the mixture
from the mould posed a serious problem. A large variety of sealing
agents such as Hermatite, Pyroma, clay, plaster of Paris and high
melting point grease etc. were tried to stop this leakage but in vain.
At a casting temperature of 130°C, the leakage was
not be controlled. Whatever mixture was left in the mould solidified
as desired, but releasing this casting from the mould posed another
difficulty. Draught angles were not used on the flange mould surfaces
as a uniform wall thickness was a main requirement. The photo-
elastic analysis of a model with a non-uniform wall thickness becomes
a very complicated problem. It was decided to modify the mould
design to overcome these (leakage and mould releasing) problems.

The outer shell was made by joining three aluminium plates with
screws so as to strip the shell during the releasing operatioti. "0"
rings were used to solvethe leakage problem and they were placed in
slots provided along all the mating surfaces of the mould. The solid
aluminium block was attached to a top cover plate which was screwed
on the side plates of the outer shell during assembly of the mould.
Two end plates with rubber gaskets were screwed to the mould ends.
The length of the mould was decided from the space available in the
electric oven. The aluminium moulds are shown in Plate 3. L These
moulds were successfully employed in casting many photoelastic models
of channel sections.

ii. c) Casting procedure

A thermostatically controlled electric oven (Gallenkamp Electric Oven)

was used for heating purposes during the casting process. The
controlled cooling rate for CT 200 Araldite during casting was
-recommended as 20C/hour. It was not possible to achieve this control
by manual operation which was available in the oven. A small electri-
cally operated mechanism consisting of an electric motor and a
reduction gear box was installed on the thermostat of the oven. This
mechanism provided one rotation of the drive shaft in 48 hours which
was further reduced to 1 rev. per 96 hours by installing small plastic
gears. This arrangement gave the desired automatic temperature control
of the thermostat.

The aluminium mould was always cleaned by Gramasol de-greasing agent

and the casting surfaces were polished before every casting. Silicon
mould release agent was applied to all casting surfaces and the unassem-
bled mould was pre-heated for 15 minutes at 110°C in the oven. All
the casting surfaces were polished again with a fine cloth so that the
excess mould release agent was removed. Then the mould was assembled
by screwing tightly against the "0 rings", and an end plate was
fitted along with a rubber gasket on one side only, leaving the other
 122

end open for pouring purposes. The assembled mould was then
kept in the oven at 140°C for at least half an hour before pouring
the plastic in it. This was to ensure that constant temperature was
maintained throughout the mould components.

Araldite CT 200 and the hardener HT 901 in proportions of 4: 1 by

weight were measured in quantities which would be sufficient to fill
the mould when mixed in liquid state. Each one of the constituents
was heated in a glass beaker to 140°C in the oven.

The glass beaker containing the hardener was always kept covered
during heating as very pungent vapours formed when it melted.
It took about 45 minutes on the average to obtain a homogeneous
solution of each. At this stage the Araldite solution was taken
out of the oven and stirred slowly until the temperature dropped
to about 130°C. The hardener solution was poured slowly into the
Araldite solution and the mixture was stirred constantly to keep
the mixture temperature around 130°C. The exothermic reaction
tended to increase the mixture temperature if the mixing was done
too quickly and with inadequate stirring. The mixture was kept for
about 20 minutes at 130°C until a clear bubble-free solution was
obtained. The solution was filtered through a cloth filter if some
impurities were found in it. The clear mixture was poured slowly
intc th_t preheated mould. Care was taken to ensure that as far as poss-
ible. air bubbles were not trapped inside. The second end plate was
screwed in position and the mould was transferred back to the oven
and the temperature was set to 126°C.

The mould was allowed to stay at this temperature for about six hours
after which the temperature control mechanism was switched on so that
the temperature would decrease at 2°C/hour rate to 70°C. The oven
and temperature control switch were switched off at this temperature
and the mould was allowed to cool to 40°C. It took about 36 hours
in all to reach the. stage at which the mould was ready for stripping
to ' release the photoelastic model from it.

ii. c) Curing and annealing of photoelastic models

The photoelastic models produced by

casting, as described in the
previous section, could not be used directly for photoelastic
analysis. Residual birefringence was visible in them when they were
examined by a transmission polariscope. They had to be further
cured and annealed to remove this residual birefringence.

Each of the cast models was placed in the electric oven on a flat
surface free from constraints. The temperature was raised to
134°C and the models were kept for about four hours at this constant
temperature. It was observed that the models, when cured at
temperatures higher than 138°C, deformed permanently. The best
results were obtained by controlling the curing temperature carefully
between 130°C to 135°C. The temperature of the oven was reduced
after four hours by switching on the temperature control mechanism
which brought the temperature to 75°C in about 36 hours. The oven
was then switched off and the models were allowed to attain room
temperature inside the oven. The models were examined once again
for residual birefringence with a transmission polariscope.

It was found that the models obtained after performing the above
curing and annealing cycle were almost free from residual bire-
fringence effects. A small amount of birefringence was noticed
in some models around tiny impurities which could not be filtered
out during the casting procedure. Care was taken during fabrication
of joints that spots showing residual birefringence did not occur
in the critical region of the joint so that the stress pattern did
not get distorted. It was also observed in almost all the models
that the free edges of the flanges showed some residual birefringence.

ii. d) Machining of CT 200 models

The rubber gasket used to stop leakage from the ends of the aluminium
moulds became imbedded in the plastic models. The two ends of every
cast section had to be cut off to remove portions of gasket material.
Also, to obtain sections of the desired flange width, the flanges
were cut. The cutting of flanges and ends of a cast section was done
on a band saw. The models were cut very carefully to avoid chipping
of the edges and the introduction of excessive machining stresses.

The edges of cast sections were later smoothed by filing and using
fine emery paper. A small residual birefringence remained on the
edges cut with a band saw. In some models chipping produced small
effects which changed the fringe distribution along an edge.

ii. e) Photoelastic models of isolated joints and a ladder frame

After successfully casting channel sections, it was decided to extend

this procedure to cast a model of a complete isolated joint consisting
of two channel sections attached to each other.

An aluminium mould was designed for casting a complete joint. Plates

(3.3) & (3.4) show this mould in unassembled and assembled form. This
mould was used to cast a model of a joint using the casting procedure
employed for casting individual channel sections. But releasing the
cast model from the mould caused a major problem. CT 200 plastic
being brittle could not stand even the small force used to release it.
The model broke during the releasing stage. Later, a few more attempts
were made to cast a complete joint but unfortunately every model broke
during the releasing stage. The failure to free the cast model in
one piece was apparently due to the mould release agent although it
was used satisfactorily in casting channel sections. Ultimately it
was decided to use a suitable adhesive for making models of isolated
joints and a ladder frame from cast channel sections.

A heat curing adhesive 1421 for bonding

was recommended purposes.
Its constituents were:

1. Araldite Epoxy Resin AY105

2. Araldite Hardener DIY 951.
 124

They were mixed in recommended proportions and the adhesive was

applied to the bonding surfaces. The joint was cured at 50°C for
twenty hours in the electric oven. Bonds of sufficient strength
were obtained after some experience. The models of isolated joints
made in this manner are shown in Plate No. (3.2). The cross member
in the isolated joints I, II and IV was attached on the outside of
the channel section web so that photoelastic measurements could be
made on the cross member near the joint.

A model of a ladder frame shown in Plate (3.6) was fabricated from

cast channel sections. The ends of cross members were supported and
secured carefully in place during the initial stages of curing the
adhesive. All the cross members were first attached to one side member
and the joints were cured. Then the free ends of the cross members
were positioned with respect to the second side member and joints
were made. Each section used in the ladder frame was inspected for
residual birefringence by a transmission polariscope. The sections
used had minimum residual effects. But when the portions of the
complete ladder frame were seen through a reflection polariscope,
a large number of areas of residual birefringence appeared especially
on the side members. This was probably due to the force used to
position the cross member ends in the second side member. The entire
ladder frame was then subjected to a stress annealing cycle in a large
electric oven with coarse thermostatic controls. This procedure helped
in reducing residual birefringence on the side members but in
resulted
deforming the cross members. The cross members sagged in the middle by
a small amount and residual birefringence appeared in the middle. The
effect of residual birefringence near the cross member ends was quite
small. It was expected that during the loading of the frame fringe
measurements in the region of the joint would not be affected.