Al-Hakeem Thesis 1991

CRANFIELD INSTITUTE OF TECHNOLOGY
COLLEGE OF AERONAUTICS
IMPACT CENTRE
PH. D. THESIS
Academic Year 1990-1
ALI HASHIM AL-HAKEEM
STRUCTURAL ANALYSIS OF
TRUCK CHASSIS FRAMES
UNDER LONGITUDINAL LOADS
CONSIDERING BIMOMENT EFFECTS
Supervisor JC Brown
July 1991
Ivp
7757
*Ji', 1 L4j
. 4-
pU\34fi2LGj1
---
In the name of God, most Gracious Most Merciful
"Of Knowledge it is only a little that is
communicated to you, (0 men! )"
Quran 17/85
This thesis is dedicated to
my son Hashim, my daughter Haneen

and my little boy Hussein
to whom the future belongs.
ACKNOWLEDGEMENTS
WITHOUT the help given by several benefactors this work could not have been done.
4 le
K.J11 g
),
. u IL11
in other words "great gifts are from great men".

Words are not enough to express my deepest gratitude to; -
- Mr. JC Brown, my supervisor whose excellent tuition, valuable advice, continuous

encouragement and productive guidance has a vital role for the accomplishment
of this research.
Mr. GH Tidbury, for his help, kind encouragement, timely suggestion and useful
-
discussions.
- Cranfield impact centre staff, for their support and friendly relationship.
- My wife Muna, for her patience and support through the darkest hours of this
work.
For those persons whose names are not mentioned, but in one way or another have
assisted me, I am much obliged.
What merits this work may have, spring from the meticulous help from the people
whom I have named. As for the demerits, responsibility for which rests on my
shoulders, may I ask for the user's indulgence.
o °'
- "perfection is exclusively a divine attribute'.

ABSTRACT
Thin walled beams warp under torsional and longitudinal loads. Warping restraint
produces high longitudinal stresses.
This is an analysis of the stress distribution in the side members of commercial
vehicle chassis frames under the effects of the previously little studied longitudinal
loads which may act on a truck chassis through spring hanger brackets. The structure
analysed is a model chassis frame consisting of channel section side members and
four cross members with different joint connections.
The developed theories are incorporated into a special purpose finite element program
which may be used in the preliminary stages of chassis frame design. Although the
program is only used for the longitudinal load case in this thesis, it is generally
applicable for other chassis load cases, including torsion, bending,. etc and
..
combination of these.
The theoretical results obtained from the program and the finite element analysis on
complete chassis frame models are validated against experiments performed on a

strain-gauged chassis frame model constructed with the same dimensions and
constructional details as the finite element models with the appropriate loading and
boundary conditions.
Suggestions for the optimum design and attachment positions for components such as
spring hanger bracket which may apply longitudinal loads to the side members of the
chassis frame are discussed from the point of view of longitudinal loadings.
i
CONTENTS
CHAPTER. PAGE
1. INTRODUCTION
1.1 GENERAL 1
1.2 REVIEW OF PREVIOUS WORK 3
1.3 RESEARCH CONTRIBUTION AND PRESENTATION 13
2. WARPING THEORY OF THIN-WALLED

OPEN SECTION BEAMS
2.1 INTRODUCTION 18
2.2 WARPING 18
2.3 SAINT VENANT THEORY (FREE WARPING) 20
2.4 AXIAL CONSTRAINT EFFECTS (RESTRAINED WARPING) 20
2.5 TORSION BENDING THEORY 21
2.6 BIMOMENT AND FLEXURAL TWIST 22
2.7 THE DIFFERENTIAL EQUATION FOR RATE OF TWIST
ALONG A THIN-WALLED BEAM 23
2.8 COMPLETE WARPING RESTRAINT 24
2.9 PARTIAL WARPING RESTRAINT 27
3. STIFFNESS MATRIX OF THIN-WALLED

OPEN SECTION BEAMS
3.1 GENERAL 37
3.2 FUNDAMENTAL CONSIDERATIONS 37
3.3 EQUILIBRIUM AND COMPATIBILITY 38
3.4 DERIVATION OF THE STIFFNESS MATRIX 39
3.5 TRANSFORMATION FOR STIFFNESS MATRIX 44
4. THIN-WALLED CLOSED-SECTION BEAMS

4.1 GENERAL 52
ii
4.2 WARPLESS CLOSED SECTIONS 52

4.3 COMPLETELY RESTRAINED WARPING 54
4.4 PARTIALLY RESTRAINED WARPING 57
4.5 THE STIFFNESS MATRIX 58
S. CHASSIS FRAME JOINTS

5.1 GENERAL 65
5.2.1 RIGID JOINTS ASSUMPTIONS 65
5.2.2 JOINT COMPATIBILITY 65
5.2.3 EQUILIBRIUM MATRIX 68
5.3.1 FLEXIBLE JOINT ASSUMPTIONS 70
5.3.2 RATE OF TWIST STIFFNESS 70
5.3.3 BENDING AND TORSION STIFFNESS
COEFFICIENTS 71
5.3.4 THE JOINT ELEMENT 71
6. THEORETICAL INVESTIGATION OF THE

EFFECT OF LONGITUDINAL LOADS
6.1 GENERAL 78
6.2 LONGITUDINAL LOADS APPLIED OUTSIDE
SECTION PROFILE 78
6.3 TOTAL STRESSES DUE TO LONGITUDINAL LOAD 80
7. FINITE ELEMENT ANALYSIS

7.1 GENERAL 84
7.2 FINITE ELEMENT SYSTEM 84
7.3 STRUCTURAL IDEALISATION OF A JOINT 85
7.3.1 CROSS-MEMBER IDEALISATION 85
7.3.2 SIDE-MEMBER IDEALISATION 85
7.4 FINITE ELEMENT IDEALISATION OF THE JOINTS 86
7.5 RESULTS FROM JOINT FINITE ELEMENT ANALYSIS 88
7.6 FINITE ELEMENT IDEALISATION OF A COMPLETE
CHASSIS FRAME 90
iii
8. THE "A. SAFE PROGRAM"

8.1 GENERAL 112
8.2 A. SAFE PROGRAM 112
8.3 THEORETICAL CHASSIS FRAME MODEL
UNDER LONGITUDINAL LOAD 115
9. EXPERMENTAL INVESTIGATIONS
9.1 GENERAL 124
9.2 MATERIAL PROPERTIES 124
9.3 CONSTRUCTION OF THE TEST MODEL 124
9.4 STRAIN MEASURMENT 125
9.5 SUPPORT AND LOADING CONDITIONS 126
9.6 TESTING PROCEDURE 127
10. CONCLUSION
10.1 GENERAL 145
10.2 SUMMARY 145
10.2.1 THEORETICAL INVESTIGATION 145
10.2.2 EXPERIMENTAL INVESTIGATION 146
10.3 DISCUSSION 147
10.3.1 THEORETICAL ANALYSIS 147
10.3.2 FINITE ELEMENT ANALYSIS 148
10.3.3 A. SAFE PROGRAM 149
10.3.4 GENERAL 149
10.3.5 THE GRAPHS SHAPE 149
10.4 PRACTICAL SUGGESTIONS 150
10.5 SUGGESTIONS FOR FURTHER RESEARCH 151
REFERENCES 153
BIBLIOGRAPHY 159
APPENDIX (A)
iv
LIST OF FIGURES
FIGURE PAGE
1.1 Typical commercial vehicle with a ladder chassis frame 16

1.2 A ladder chassis frame with variety of joint designs 17
2.1 Typical types of thin-walled open section beams 30
2.2 Channel section beam under pure torsion 31
2.3 Shear stress variation of an open section across the
thickness under torsion 31

2.4 Distribution of sectorial area 32
2.5 Warping inhibition in open section beams 33
2.6 Boundary conditions for the chassis frame cross members 33
2.7 Loads that introduce a bimoment into a channel section 34
2.8 The distortion of an I-section due to torsion 35
3.1 Mode of load transfer 48
3.2 Moment and bimoment equilibrium at a joint 48
3.3 Force and displacementcomponentsin local coordinates 49
3.4 Sign convetion for loads and displacements in a warped
channel-section beam element 49

3.5 Stiffness matrix for beam element with bimoment terms 50
3.6 Coordinate system for a grillage 51
3.7 A generalised coordinate for 3D frame 51
4.1 Generation of warping displacement of closed section 61
4.2 Warping of rectangular box section 61
4.3 General coordinate system of thin-walled tube 62
4.4 Displacement function of a box beam 62
4.5 Displacement function in the plane of the cross
section of a box beam 63

5.1 A joint defined by a (TSW) node with a horizontal web
channel cross member 73

5.2 A joint with a vertical web channel cross member 73
V
5.3 Displacements of nodal points at the ends of beams meeting at
a joint, used to build up the equilibrium matrix [H] 74

5.4 Loads corresponding to the displacement in fig. (5.3) 75
5.5 Distances of torsion axes and warping point from the centroid
axis for various joint configuration, see fig. (5.3) 76

5.6 Warping lines and warping axis of a channel section beam 77
5.7 Example of a joint element 77
6.1 Bimoment created due to longitudinal load 81
6.2 Principal sectorial area (ws) of the model used 81
7.1 Finite element idealisation of joints type 1&2 91
7.2 Finite element idealisation of joints type 3&4 92
7.3 Displacements due to bending flexibility 93
7.4 Displacements due to rate of twist flexibility 93
7.5 Joint deformations 94
7.6 Joint deformations 95
7.7 Stress distribution due to partial warping inhibition at joint No. 1 96
7.11 Warping displacements for joint No. 1 100
7.15 Ladder chassis frame used for F.E. analysis
and experimental tests 104

7.16 Finite element idealisation of model chassis frame under
longitudinal loads, case 105
-1-
-2-
-3-
vi
7.19 Deformation and stress distribution for case (1), when the
longitudinal loads are applied at position (2) 108
8.1 Flow chart of A. SAFE program 116

8.2 Flow charts of subroutines used in A. SAFE program 117
8.3 Flow charts shows the incorporation of A. SAFE program into
chassis design programme 118

8.4 Cases and positions of loading 119
8.5 Bimoment distribution due to longitudinal
load, (see fig. 8.5) 120
8.6 Moment case (1), (p) at position (2), (see fig. 8.4) 121
9.1 Stress-strain curves for a mild steel tensile test 130
9.2 Strain-gauge positions and stations 131
9.3 Stress distribution for load-case No. 1, when the load (p)
is applied at position (1) 132
9.9 Total stresses around selected cross sections of the side
member, see fig. (9.2), for loading-case No. 1, when the

load (p) is applied at position (2) 138
vii


10.1 A variety of typical spring hanger brackets 152
10.2 Alternative design and attachment of a spring hanger bracket 152
LIST OF TABLES
TABLE PAGE
2.1 Comparision of the sectorial properties 36

4.1 Summary of functions used in the analysis of open and
closed section beams 64

4.2 Summary of constants used in stiffness matrices of open
and closed section beams 64

7.1 Rate of twist stiffnesses of joints, obtained by F.E. analysis 111
LIST OF PLATES
PLATE PAGE
9.1 General arrangement of chassis test rig 141

9.2 Loading-case No. 1 142
viii
NOTATION
b cross section width

B bimoment of a node in open section member
Bp bimoment due to partial warping in open section member
Br bimoment due to complete warping inhibition in open section member
denotes closed section member
C
e distance from the shear centre to web of the member
E Young's modulus of member materials
G shear modulus of member materials
h half the web width of a channel section or the hight of a box section
[H] equilibrium matrix
J torsion constant of an open section
J, torsion constant of an open. section with partially warping restrained at one
end
Jr torsion constant of an open section with completely warping restrained at one
end
K warping restraint factor
[Kw] stiffness sub-matrix for the warping terms of an open section member
L length of a member
s cross section profile

t thickness
[T] transformation matrix
T total torque
Ty St. venant (or pure) torque
Tr torque due to bimoment (or flexural twist)
W warping displacement of an open section member
Wp warping displacement of an open section member partially warping restrained
at one end
Wr warping displacement of an open section member completely warping
restrained at one end
x distance along the member
ix
0. angle of twist of an open section member

0, angle of twist at the free end of an open section member
partially restrained against warping at the other end
denotes differentiation w. r.t x
cpx angle of twist of a closed section member

cps angle of twist at the free end of a closed section member
partially restrained against warping at the other end
v Poisson's ratio
T shear stress
ws principal sectorial area of a point
P characteristic length of an open section, p= GJ/EF
F warping constant
Or stress due to bimoment with complete warping inhibition at one end
ap stress due to bimoment with partial warping inhibition at one end
ßA stress due to axial load
ß,. stress due to moment (M1)
ßZ stress due to moment (Me)
US stress due to bimoment
ßT total stress
CHAPTER ONE
INTRODUCTION
1
1.1 GENERAL
Thin walled beams are used for many types of structures such as a ship, a
bridge, an aircraft, a space-ship, a motor-car and other structural design. They are
popular with vehicle designers since the manufacturing and economic possibilites are
greater. There is no clear distinction between thin and thick walled beams. It is
generally accepted that thin walled beam theory may be applied with reasonable
accuracy to sections for which the wall thickness is small compared with any cross
section dimension (50.1), which is itself small compared with the length of the beam.
The latter condition may not be as important as the first. The first condition is
important and if it is not satisfied the theory of thin walled beams may lead to
erroneous numerical results due to the break down of the approximating assumptions
used.
Under torsional load thin walled beams can be divided into two types. The first
are warpless sections (i. e. plane sections remain plane) such as squares, circles,
triangles and other regular polygons will not warp if the material thickness is
constant. The second type comprises warping sections (i. e. plane sections do not
remain plane) such as rectangular, channel, and I-sections.
In members subject to torsion or longitudinal loads, warping effects are more
significant in open sections than in closed sections. Thin walled beams with closed
or open section are used in the construction of commercial vehicle chassis frames
(see figure 1.1).
Commercial vehicles such as trucks, trailers and semi-trailers have chassis frames
which are of the ladder type. These are so termed because of the configuration of
their members. They generally consist of two side members arranged parallel to the
longitudinal axis of the chassis and several cross members placed laterally between
the side members. Thus, the axles, as well as the power plant, the driver's cab and
platform or other superstructures, are easy to attach. Whilst it must be stated that the
conventional ladder type frame is an inefficient structure for carrying bending and
torsion loads it remains true that for historical and economic reasons virtually every
commercial vehicle in the world is based on such a chassis. The demand for vehicles
2
with chassis frames continues to increase in the commercial vehicle industry where
the modem trend is towards lorries and articulated vehicles carrying large loads.
The side members of ladder chassis frames are usually made from open channel
or I-sections. An I-section is very efficient in providing bending stiffness but
manufacturers prefer channel section side members because of cost and ease of
construction. The cross members are often made from hollow rectangular, channel,
tophat or I-sections. Hollow rectangular sections give efficient torsional and bending
stiffness , but can lead to high overall frame torsional stiffness. The most flexible
design of frame would have open section cross members attached through end plates
to the side member webs. Cross members can have variable cross section, i. e, shaped
members to act as engine supports, cab mounts, also members whose depths are
reduced in centre span to miss transmission arrangements.
There is a great variety of design of joints between cross members and side
members (see figure 1.2), both as to joint configuration and the method of attachment
of cross member to side member. The joints can greatly affect the torsional stiffness
of the frame and cause high stress concentrations to develop. It is necessary that the
desired torsional stiffness of a frame should not produce very high stresses in the
joints which could cause them to fail. Welded joints are more difficult and hence
more expensive to fabricate. Bolts and rivets in joints, although being the easiest
methods of fixation, can cause stress concentration in the region of holes through
which they pass.
British chassis frames tend to be much more flexible in torsion than their
continental counter-parts which employ closed section cross members or extensive

use of gusseting to increase torsional stiffness. However, frames should be stiff
enough to ensure good vehicle road holding as well as flexible enough to supplement
the suspension system. thus ensuring that wheel to ground contact is maintained.
Chassis frames are an important structure that must resist various loads during
operation; vertical as well as longitudinal and torsional static and dynamic loads.
Although adequate durability under dynamic conditions is an important design
3
requirement, static loads can not be disregarded and should be analysed. It is

necessary to forecast fatigue life of a frame already in the construction process as
well as to maximize the value of the load-carrying capacity of the material used.
This, in turn necessitates more accurate estimation of stresses in those elements of
frames.
The conventional design of ladder chassis frames has hitherto been based on the
provision of side members of sufficient strength and stiffness to support the bending
load due to the payload carried by the vehicle. The cross members chosen are based
on the designers experience, while the effects of warping restraint, which is the major
factor for the cause of high stresses in the frame, are neglected completely or
regarded as secondary. The bimoment due to horizontal braking forces can cause large
stresses in the frame, which are not usually considered in the design of side members.
Analysis of the whole chassis frame by standard beam elements in finite element
programs leads to unacceptable approximation especially where the beam element has
uniform cross section. Thus, automatic structural programmes based on a hybrid
method of analysis, which combines finite element idealization of the joint areas with
analytically derived beam elements for the chassis frame members are required to
obtain a more reliable estimation of overall stiffness and stress distribution in chassis
frame members for various loading conditions.
1.2 REVIEW OF PREVIOUS WORK

The inhibition of the warping at joints by other elements in the frame, gives
rise to large longitudinal stresses, which have been the cause of many failures in
chassis frames. These effects have been analysed by many authors following the
publication of the major work on thin walled beams by Vlasov (1) where he
introduced the concept of the bimoment. His theory allows for the extreme cases of
complete warping inhibition or free warping at the joint. An introduction to Vlasov's
work is given by Zbirohowski-koscia (2).
Hanke (3) analysed a ladder frame joint consisting of a channel section cross
member symmetrically bolted to the web or flanges of the channel section side
4
member. The rate of twist of all members ending in the joint was assumed to be
equal and the condition of bimoment equilibrium at the joint was obtained. He
developed a differential equation using this concept and solved for certain boundary
conditions. Hanke was one of the first authors who suggested that joints do not
behave in a rigid manner and introduced the term 'induction factor' to estimate the
degree of warping restraint in them. If the side member completely restrains the
warping of the cross member, the joint is perfectly inductive and the factor has a
unity value. In the case of complete absence of an inductive connection, when the
cross member is allowed to warp freely, the induction factor is zero. However, the
true value for a real joint lies some where between the two. He carried out
experiments on several types of joints to determine the value of the 'induction factor'
by measuring rotation and stresses of the members for each constructed joint. He
obtained a high 'induction factor' when the cross member was bolted to the flanges
of the side member, while bolting to the web of the side member alone resulted in
a low 'induction factor'.
Zaks (4-5) investigated the warping effects in cross members welded to channel
section side members. The asymmetry of the connection of the cross member to the
side member with the intersection of the neutral axes offset was considered. He used
Vlasov (1) thin walled beam theory and introduced the term 'bonding factor' which
is equivalent to Hanke's (3) 'induction factor' to estimate the degree of warping
restraint by measuring stresses and rotations on members of individual joints. He used

these values to estimate stresses and deflection in the chassis frame. He also
introduced the concept of 'kinematic aspect' which implies that a rigid joint has
perfect kinematic coupling between members meeting in the joint.
Zaks also included the length of the member in the derivation of the bimoment
equation and pointed out that the 'bonding factor' could depend on the length of the
member if the beams are very short (very short beams are rarely used in practice).
Otherwise the 'bonding factor' is constant for any particular design. He compared
results obtained from plate theory (6) with those obtained from Vlasov's theory (1)
for an I-section cross member symmetrically attached to a channel section side
member. Both predicted values compared well with measured values in the regions
S
away from the joint where there are no localised effects of the joint. Plate theory
equations developed by Zaks have been used by Kobrin, Kilimnik and Titov (7) to
investigate the stresses in the walls of the chassis frame side member and good
agreement between the predicted and experimentally measured values were obtained.
Seitler (8) also carried out tests on a model of a welded chassis frame using
tubular cross members. He obtained about 80% lower stresses at the joints and he
found that the torsional stiffness of the whole frame was about three times greater
by using circular tubes rather the channel sections as cross members.
Cooke (9) demonstrated an iterative method using the strain energy theorems of
Castigliano to estimate chassis frame stiffness with uniform section members. He first
considered the rectangular outerframe, consisting of the two end cross members as a
basic outline frame and determined its torsional stiffness by strain energy methods.
Then the torque required on each internal cross member so that it would cause no
further displacements of the outline frame when placed in the deflected outline was
calculated. This torque was then converted to an external torque on the frame. The
new external torque was applied to the outline frame to find the new deflected shape,
the iterative process being repeated until no significant difference in the two values
of the torque was obtained. Cooke's method requires a careful prediction of the
deflection mode and, especially for a large number of cross members, can become
very laborious. It does not take into account the effect of actual warping restraint in
the joints.
Tidbury/Marshall/Roach (10-12) dealt with the problem of determining

theoretically the warping restraint factor for a chassis joint and the torsional stiffness
of a ladder frame, and of verifying the theoretical results experimentally. The
arrangement selected for the main analysis was a chassis/joint connecting channel
section cross members. In the analysis of the warping restraint factor, only the web
of the channel section side member was taken into consideration. This web was
divided into independent strip beams which were taken to be simply supported at the
flange/web corner. The strip beams were taken to resist the cross member warping
forces by bending. Linear variations of these forces on a strip beam were
6
approximated by point load. Bending theory was applied to the strip beams and
Wagner-Kappus torsion theory to the cross member.
It was assumed that the partial warping displacement of the cross member at its
joint end was proportional to its free warping displacement and that the partial
warping stress was proportional to its fully warping inhibited stresses. An average
value of the warping restraint factor was found and used for the cross member cross
section at its joint end. From the warping restraint factor the effective torsion constant
of the cross member was found.
The effective torsion constant for the side member was found by taking only the
web of the side member into consideration within a bay. The web was treated as
rectangular plates under torsion and restrained at both ends by the cross members. The
effective torsion constant of the cross member and the side member were used in a
modified Erz (13) formula to get overall torsional stiffness of the chassis, which
compared well with measured values. Also good agreement with experimentally
measured values for stresses in the side member were obtained when the value of the
partial warping restraint factor was smaller than (0.4). Similar work on the effect of
warping inhibition in joints on the torsional stiffness of chassis frame has been
reported by Awudu (14).
Megson/Alade/Nuttall (15-20) dealt with the theoretical determination of warping
restraint factor for a chassis joint and from it the torsional stiffness of a chassis under
torsion. Also they dealt with the theoretical determination of warping stresses and
displacement of the cross member under torsion and the verification of the theoretical
results by experimentally measured values.
The arrangement analysed was a chassis having channel or I-section members.

The torsion load was applied as equal and opposite couples at the ends of the chassis.
The couples were comprised of individual loads applied through the shear centres of
the side members to prevent local twisting. Two methods of idealisation i. e,

,
assumptions of infinite and finite stiffness in warping of all members of the chassis,
were used in the determination of the torsional stiffness of the chassis. For the infinite
7
solution, the Erz (13) formula was used to find the torsional stiffness. For the finite
stiffness solution the finite element method was used. Nodes were taken especially at
the joint between the side member and the cross member and at the points where the
member changes section. The beam elements were solved for bending, shear force and
torsion and the results were super-imposed to get resultant terms. Stiffness method
(displacement method) was used in an available computer program which solved for
a chassis comprised of beam elements. The effective torsion constant used to find the
torsional stiffness was found by considering the warping restraint factor of the cross
member on the side member. The warping restraint factor used to find the effective
torsion constant of the cross member, was found from the joint of a section of the
chassis.
In the joint, the side member was treated as a plate (web only considered) simply
supported at a channel or I-section cross member attached symmetrically. One end of

the cross member was considered partially restrained and the other (mid-span) free to
warp. Torque was applied at the free end. In the finite element solution the side
member was considered using plate elements in bending, and for the cross member
plate and membrane elements, in bending and stretching were used.
Wagner's torsion bending theory for axially constrained open sections was applied
to the cross members and side members. The assumption made was that the partially
restrained warping displacement of the cross member was directly proportional to its
completely free warping displacement. The warping restraint factor for the cross
member was obtained from the solution of Wagner's torsion theory and finite
elements. The moment couple method was used in Wagner's torsion theory to include
the effect of warping restraint of the cross member on the side member and from this
the effective torsion constant for the side member was found.
Experimental values for a range of torque, loads, dimensions for the chassis
member were obtained using strain gauges. The agreement between experimental and
theoretical results was found to be good, so was the correlation between infinite and
finite stiffness in warping assumption. Also the assumption that the partial warping
displacement was linearly proportional to the completely free warping displacement
8
was validated. The limitation of this method is that it is applicable only for
symmetrical attachment of the cross member to the side member. The assumption
mentioned immediately above may not be true for unsymmetrical attachments. Also
the local twisting of the side member is not considered.
Similar work but on the stress distribution in the vicinity of the connection of
a joint of a ladder frame subjected to torsion has been reported by Datoo (21). His
finite element analyses indicate a redistribution of the axial constraint stresses in the
vicinity of the connection which produces stress concentrations at the cross member
flange tips. He carried out torsion tests on glued perspex and welded steel joints to
verify those stresses. He produced guidance charts of stress concentration factors for
selected joints, using the finite element method.
Alvi (22-23) dealt with the problem of determining theoretically the stress
distribution in a chassis joint and of verifying the theoretical results by measured
values. The arrangement selected by him was a cross member attached to the web
of the side member. Two cases for the side member were taken. In the first, the side
member was a plate and in the second a channel section. The cross member was a
channel section. Free warping was taken for the side member ends and a torsion load
was applied at the (free) end of the cross member. The ends of the channel section
side member were taken to be simply supported and so were all the edges of the
plate.
Alvi applied Wagner's torsion theory for thin walled open sections to the cross
members and classical plate theory (6) to the side member. The channel section side
member was treated as three plates joined together, i. e the two flanges and the web.
Bending and stretching of the plates due to the warping and couple loads of the cross
member acting on it were included in the solution. For the cross member which is
under torsion, the partial warping stress developed at its joint end was taken to be
proportional to the total warping inhibition stress and the partial warping displacement
proportional to the free warping displacement. The proportionality was expressed in
terms of the warping restraint factor. An average warping restraint factor for the cross
member cross section at the joint was found using plate theory and Wagner's torsion
9
theory. From this thesis it is not clear what method was used to find the average
stress. Stresses due to partial warping restraint at the cross member joint end was
found from the warping restraint factor.
By using plate theory Alvi had overcome the problem of unsymmetrical

attachment of the cross member to side member, and he solved for both symmetrical
and unsymmetrical attachments. Fourier series form was used to express terms of
displacement, stress and the external force acting on the web. By this method it was
possible to find the stress and displacement at any point on the side member. Super-
position of the bending and stretching solution gave the final distribution of the stress
and displacement. The critical area where the shear stress attains its peak value was
found to be in line with the zero warping line of the cross member flange and near
the flange web comer of the side member. A warping restraint factor was calculated
it
and showed good agreement with those of the other two previous methods, i. e strip
beam theory and finite element method.
A photo-elastic technique was used to measure the values of stress and verify
the theoretical results. The theoretical results compared well with the experimentally
measured values and was considered more accurate than the other two methods of
joint analysis mentioned above. From Alvi's thesis it can be concluded that the region
of the joint contains high localised stresses, due to the warping restraint of the cross
member and can not be left out of an anaylsis of the side member.
Sharman (24) investigated the optimisation of ladder frames. A technique was
presented to maximise the torsional stiffness with minimum weight of chassis frame
having uniform closed or solid section with rigid joints. The members are analysed
in terms of their stiffness-weight ratios and the weight parameter was calculated for
families of peripheral and ladder frames. He developed a torsional stiffness theory
which considered ladder frames of uniform spacing as an assembly of peripheral

frames. He produced design charts for certain cases. Unfortunately the method did not
ensure compatibility of bending deformations of the cross members leading to

torsional stiffnesses which were in error by up to (5%) and more for cross members
very stiff in bending.
10
Optimising techniques are generally multi-constraint problems most suited to

computer analysis. Sharman in his paper considered just one aspect of the more
complicated problem which includes various parameters such as different geometries,
load, displacements, stress, warping restraints and other economic constraints which
have to be considered. Similar work which chose maximum direct stress and
minimum weight as their variable has been reported by Lasevich, Sholnikov and
Podlegaeva (25).
Later Sharman (26-27) considered the torsional problems of chassis frames. He

discussed the types of torsional load being applied to the chassis frame, and outlined
four main cases, those cases are asymmetric loading, twist ground plane, lateral
acceleration on the load during cornering and finally, severe manoeuvres. The chassis
structure is then considered and observations made about the effects of the different
types of loading on different types of frame. He considered the side members from
the point of view of optimum cross sectional shape. Other weight saving features are
then explored. Since his design is dependent on the allowable stresses in tension and
compression, these values are derived from a series of experiments in which the
dynamic strain in semi-trailer are recorded. Sharman (28) showed that closed members
are generally more efficient but emphasised that careful design of joints is necessary
to avoid high localised stresses. He carried out experiments on thin fabricated box
members in a tee joint to investigate the effect of joint flexibility in torsion. The
behaviour was also observed in finite element analyses of the joints. The application
of classical beam and torsional theory gave a result which was ten times the
experimental value, while his method which includes the joint flexibility as predicted
by finite element model of the localized region at the joint, gave an improved result
which was (26%) higher than the experimental value. He suggested that further
investigations should be made to find ground rules for defining the extent of the joint
area.
Sharman (29) has also investigated the problems of incorporating the effects of
cross sectional warping, offset shear centre and orientation of an element for structures
assembled from channel and I-sections. He incorporated the kinematics of a variety
of joint intersections and stiffening schemes. A transformation matrix was given to
11
account for non-coincidence of the shear centre and centroid, which enabled a solution
to be obtained for comparison with a number of experimental and analytical studies.
He has shown that the inclusion of the torsion warping terms for thin walled open
section beams into the standard stiffness for a uniform beam may be achieved by the
addition of the twist rates at each end as a degree of freedom. He concluded that the
torsional aspects of chassis frame design are complex and though approximate
analytical methods can be used in the early stage of design, computer analysis appears
to be necessary to ensure structural integrity under a variety of load conditions.
Tidbury (30-31) decribed matrix computer methods which may be applied to
analysis vehicle structures. They are displacement methods and force methods. He
used a force method to estimate the torsional stiffness of a rectangular frame and
derived an expression similar to the one proposed by Cook (9). His method requires
a careful selection of the unknown redundant forces in order to make the flexibility
matrix manageable when solved by the computer. A determinate system capable of
supporting the external load should be chosen and then calculating the loads induced
in all members.
Marshall (32) extended Tidbury's matrix force method for simple frames to
frames with five cross members. This demonstrated the disadvantage of the method
which has to be reprogrammed for each new type of structure analysed. Automatic
selection of redundanices in the force method has been developed by Robinson (33),
which in turn adjusts the flexibility matrix for a particular problem.
The displacement method which is based on the formulation of a stiffness matrix

is much easier. It is widely used in computer analysis by the finite element method.
Ali, Hedges and Mill (34) used a finite element method which is basically a
displacement method to analyse a chassis frame. A stiffness matrix based on beam
theory was used to predict the static deflection of the frame subject to torsion or
bending loads. Hedges, Noville and Gurdogen (35) extended this analysis to determine
the stress values in chassis frame members. The results compared well with the
measured values for bending loads but not for torsional loads due to imperfectly
12
idealised torsional properties of the cross members and joints. The effects of warping
restraint in the joint were not taken into consideration.
Triman (36) included the effects of warping torsion in a computer program which
uses the stiffness matrix method. The bimoment and the rate of twist are added as an
additional degree of freedom. His program was sucessfully tested by him to calculate
the internal forces including bimoment as well as nodal displacement including the
rate of twist in a cantilever beam, fixed-ended beam and continous beam. The
limitation of the program was that it was unable to calculate bimoment distribution
in the grid structures such as frames.
Lee (37) developed Triman's (36) work to allow for the bimoment equilibrium
at the joints. This was concerned with completely inhibited warping only.
Beermann (38-39) included the elasticity of nodal points in the finite element
analysis. He defined the node by more than one point. The (TSW) node incorporating
the torsion centre, the centroid and the warping point at the end of one of the beam
elements meeting at the joint are used for this definition. This theory and further
development by the present author will shown in chapter (5). The effective lengths
of the cross members were used in his analysis. He included the flexibility of the
joints obtained from a finite element idealization together with the compatibility of
their displacement.
A hybrid method of analysis is presented, which combines finite element

idealization of the joint areas with analytically derived beam elements for the cross
member and side member sections. The beam element includes warping/torsion force
displacement relationships. The flexibility of the joints is included together with the
compatibility of their displacements. He claimed that his method gives close

agreement with experimental results.
The concepts of 'zero warping lines' and the 'zero warping axis' in a cross section
have been introduced by Tidbury the editor of reference (39) for clarification of the
author's argument.
13
Al-Hakeem (40) dealt with the problem of longitudinal or torsional loads acting
on channel section side member through a spring hanger bracket and the longitudinal
stresses developed in the side member. His study has shown that the bimoments due
to the horizontal braking forces can cause large stresses in the frame which are not
usually considered in the design of the side members.
The arrangement selected was a bracket attached to the flanges of the channel
section side member equivalent to one bay of a chassis side member. Vlasov's (1)
bimoment theory was applied to the side member. The boundary conditions taken
for the side member were, total inhibition and free warping of the ends. The localised
effects due to the bracket were not taken into consideration. He also applied bending
theory to the side member to include and compare the stresses due to bending with
those of bimoment.
Fixed-fixed and fixed-free end conditions for the side member were taken in
bending. Direct stress was also included in the summing of the longitudinal stresses.
He obtained close agreement between theoretical and experimental results with both
loading cases. The conclusion was that the bimoment stresses due to longitudinal
loads were quite large when compared with the sum of bending and direct stress, and
added significantly to the accuracy of the results.
In view of the volume work, it has only been possible in this review to point
out some emphasis on chassis frame researches. Similarly, no claim can be made
that review is exhaustive and the reference list is in any way comprehensive or that
even all the significant contribution have been included. Apologies are therefore
offered to those individuals and organization whose work, although known to me, has
only been included in the bibliography.
1.3 RESEARCH CONTRIBUTION AND PRESENTATION

This project deals with the problem of longitudinal loads acting on the side
members of chassis frames through spring hanger brackets and the longitudinal
stresses developed in the side members due to warping restraint effects.
14
The members of a ladder chassis frame are regarded as having thin walled
sections. Hence, the establised Vlasov theory for thin walled open section beams,
when their cross sections are allowed to warp freely (St. venant torsion) or completely
restrained from warping (torsion bending theory), are presented in chapter (2).
Also the theory of torsion of thin walled open section beams, having cross
sections partially restrained from warping are presented in the same chapter.
Investigators have made various assumptions for determining the torsional stiffness
and eventually stresses in cross members. These analyses are now quite well
understood and documented, with the work of various authors (see section 1.2).
In chapter (3), a general elastic stiffness matrix is derived including warping

inhibition in thin walled open section beams. The rate of twist and bimoment are
used to modify the conventional (6) degrees of freedom beam structural analysis to
one which has (7) degrees of freedom for each node. The warping displacement at
the joints for the derivation of this stiffness matrix is considered as being completely
transferred to the other connected parts as far as the joints are concerned. Accordingly
a general transformation from local to global axes was derived to account for the new
load system.
The analysis of chassis frame members with closed section cross members is
also considered. In chapter (4) formulae for analysing thin walled box section beams
subjected to torsional loading and considering fully restrained warping is presented.

The method is extended to take only partial restraint of warping into account but does
not include the effects of cross sectional distortion. A stiffness matrix similar to the
one presented in chapter (3), but for thin walled closed section beams was developed
in this chapter.
Chapter (5) contains a developed equilibrium matrix for different types and
orientation of cross members beams meeting at the joint. This equilibrium matrix is
built on a rigid joint assumptions, but it allows for the axis offset of the members
meeting at the joint. Flexible joint assumptions are also presented, and the rate of
twist stiffness coefficient of a joint was defined.
15
A method developed for calculating the bimoment created due to the longitudinal
load was presented in chapter (6).
In chapter (7), a description is given of the finite element system used for this
research project and the structural and finite element idealisations of ladder chassis
frame joints having channel and closed section cross members. A finite element
idealisation of complete chassis frames under longitudinal loads were also presented.
The same chapter contains the description and the results of the finite element
analyses on ladder chassis frame joints.
In chapter (8) theories and methods, developed in previous chapters, were

incorporated into a special purpose finite element program using the direct stiffness
method. The program includes warping inhibition effects in thin walled beams. Both
assumptions for either rigid or flexible joints are used in the analysis by this program.
The program can be used as a design tool in the preliminary stages of chassis design.
The distribution of bimoments and of moments along the side members of a chassis
frame due to longitudinal loads for different loading cases is also presented in this
chapter.
A description of the longitudinal load tests performed on a ladder chassis frame

in the laboratory with different longitudinal load cases are presented in chapter (9).
The experimental rig used together with the measurement techniques adopted to
determine stresses in the side members are discussed in detail. The discussion of the
measured and corresponding finite element values with the theoretical results of
stresses obtained from the program developed in chapter (8) are also presented.
Finally, chapter (10) contains general conclusions, discussion and several

suggestions for the optimum design and attachment positions for components such
as spring hanger brackets to the side members of the chassis frame, from the point
of view of longitudinal loadings. Lines of research are suggested which follow on
from those presented in this thesis.
E
w
w
U,
Co
U,
tu
U
N
V
V
tu
41
.r.,
U
-. -/
ý--I
N
E
O
U
U
", i
r1
w
17
REAR
FRONT ENGINE CENTRE
CLOSING MOUNT BEARING No..4 No.5 REAR
ASING
a) Truck chassis
REAR CLOSING
-a -ý
mos. 4t5 REAR ENGINE

MOUNT
LtN IKt ötAKNlti II JI\1 I. LVJIilu
b) Cross member joints
Fig. (1.2) A ladder chassis frame with variety of joint designs

CHAPTER TWO
WARPING THEORY OF
THIN-WALLED OPEN
SECTION BEAMS
18
2.1 INTRODUCTION
Thin walled beams have been defined as structural elements whose three
dimensions all have a different order of magnitude. The wall thickness being small
compared with any cross section dimension which is itself small compared with the
length. Such a definition can be applied to sheet metal cold formed or coiled strip to
form various shapes such as a channel, I- or tophat sections as shown in figure (2.1).
It is assumed that the shape of the cross section is maintained constant. Most of the
beam structures may be classified as having either thick or thin walled sections. There
is no clearly defined border between sections which may be regarded as thin and
those which must be considered as thick. Some criterion is therefore required to
distinguish between a thick and a thin sections, as assumptions of thin walled theory
will decrease in validity the thicker a section becomes. It has been suggested by
Vlasov (1) that thin walled theory may be applied with reasonable accuracy to
sections for which the ratio: -
t.
`ß 0.1
h
where (trn) is the maximum thickness in the section, and (h) is the typical cross
sectional dimension.
Warping effects of thin walled beams can be as important as bending in

determination of stresses and displacements of these structural elements. The torsional
properties of thin walled beams can be markedly different from those calculated by
elementary methods and an appreciation of the theory is of considerable importance
in Automotive structural design. For this reason a summary of the theory is given in
this chapter with an emphasis on the important relationships for chassis frame
analysis.
2.2 WARPING
The cross section of a thin walled open section beam subject to pure torsion
will not remain plane, the displacements of the cross section in the axial direction
of the beam is called warping. Two types of warping displacement take place
19
simultaneously. The first type is the warping of the mid-plane of the cross section
as shown in figure (2.2) which is assumed constant across the wall thickness and is
known as primary warping. The second type is the warping of the section across its
wall thickness and this is known as secondary warping. Secondary warping and the
effects of restrained secondary warping are very much less than primary warping and
the effects of restrained primary warping, therefore secondary warping effects are
usually neglected. As far as this research is concerned, the primary warping will be
considered as the main effect to be studied.
Although warping of the cross section of thin walled beams occurs mainly in
torsion, it can also arise:-
i) from longitudinal loads, except when they act through special points on the cross
section and,
ii) from bending moments caused by pairs of normal loads acting in planes which
do not pass through the torsion centre as shown in figure (2.7).
Thus, the warping displacement of the mid-plane of the cross section of an open
beam is shown by Vlasov (1) to be of form: -
d8
W= - ws (2-1)
dx
Where (cos) is called sectorial area (or sectorial co-ordinate). This is twice the area
swept out by a generator rotating about the centre of twist (R) from the point of
zero warping in the cross section to any point (s) as shown in figure (2.4).
With respect to the physical properties of thin walled beams, additional sectional
properties based on the sectorial co-ordinate, which are called the sectorial properties
should be introduced. In the same way as for the conventional beam theory, those
sectorial properties are defined as shear centre, sectorial linear statical moment of a
section, second moment of sectorial area and principal second moment of sectorial
area from the principal sectorial co-ordinate. The relationship between the sectorial
and conventional properties of thin walled open section are expressed in table (2.1).
20
2.3 SAINT VENANT THEORY (FREE WARPING)

The assumptions on which the theory is based are similar to those for the torsion
of a closed section in that the cross section is assumed to remain undistorted in its
own plane after loading. In this case the beam is under pure torsion and does not
produce any longitudinal stresses even when there is an axial restraint (so long as this
does not restrain warping). The rate of twist of the beam is constant. The axial
displacement which is called warping displacement must not be prevented at any
section and the warping distribution of the cross section is identical throughout the
beam as shown in figure (2.2). Thus, the plane of the cross sections do not remain
plane.
To obtain the value of the shear stress in a section subject to St. venant torsion,
it is necessary to solve the Laplace equation. Shear stress varies across the thickness.
The distribution of shear stress across the thickness is shown in figure (2.3). The
expression for St. venant maximum shear stress distribution in thin walled open
section beam subject to unrestrained torsion is given by Megson (17) to be:-
= Ty (2-2)
mu -
Where (J) is St. venant torsional constant

(t) is the thickness
(T)is the applied torque in the St. venant case
St. venant torsion (T, ) is given as:-
dO
T, = GJ ------------------------ (2-3)
dx
2.4 AXIAL CONSTRAINT EFFECTS (RESTRAINED WARPING)

As pointed out in section (2.3), the cross sections of thin walled open beams
subjected to unrestrained (St. venant) torsion experience a free warping distribution.

21
In such a case the complete cross section suffers identical warping displacement
distribution along the longitudinal generator of the beam surface. If one end of the
beam is completely or partially prevented from warping as shown in figure (2.5),
the longitudinal generators of the beam surface are strained, the rate of twist along
the length of the beam is no longer constant.
2.5 TORSION BENDING THEORY

Consider a thin walled open section beam subjected to a torque at one end and
rigidly restrained from warping at the other end as shown in figure (2.5). The total
resistance is provided by a combination of the St. Venant shear stresses and the
resistance of the web and flanges of a channel or an I-section which are no longer
free to warp at the built-in end and this is responsible for the bending of the flanges
in their own planes. Therefore, the total torque is a sum of St. Venant torque and an
additional term called by Vlasov the flexural twist.
T= T7 + Tr (2-4)
Where (T) is St. Venant torque from the free end warping, but (d8/dx) is no
longer constant and (Tr) is the contribution from the warping restraint.
The validity of equation (2-4) has been shown in figure (2.8) for an I-section
beam where the first deformation is due to St. Venant torque while, the second due
to flexural twist. The relevant equations are shown by Vlasov (1) to be: -
d8
TJ=GJ -dx
-------------------------- (2-5)
d38
TrEr 3dx
Where (F) is the warping constant or principal sectorial moment of inertia and:-
r'= ws2 t ds
J «c8oa
22
2.6 BIMOMENT AND FLEXURAL TWIST

Vlasov (1), introduces the concept of a bimoment, (B), as shown in figure (2.7),
which is defined as the product of a pair of equal and opposite moments and the
distance between them. He defines the flexural twist (Tr) as the derivative of the
bimoment. The relevant equations are quoted from Vlasov (1).
dO
(os
dx
d20
Bý EI'
)= - dx ý
----------------------------- (2-6)
dB d30
ET
dx 3
Bws
6nxý=
r
From figure (2.8) it can be concluded that the flexural twist causes a bimoment
representing the warping forces acting at a sectorial area, and resembling of two
mutually balancing equal but opposite bending moments acting in two parallel planes.
Subsituting equations (2-5) into equation (2-4): -
d8 d38
T= GJ -Er dx3 --------------------------- (2-7)
Where (G7) is defined as torsional rigidity and,

(EI') sectorial rigidity
If the applied torsion load is distributed along the member, this becomes a fourth
order differential equation. The fourth order differential disappears if external torsion
is applied at the ends only.
For thin walled beams it should be noted that the sectorial rigidity is much
greater than the torsional rigidity.

23
2.7 THE DIFFERENTIAL EQUATION FOR RATE

OF TWIST ALONG A THIN-WALLED BEAM
In actual structures, the geometrical restraint along the longitudinal direction of
the member, such as that at the joints of a chassis frame, will cause additional
flexural twist, so the rate of twist (dO/dx) is not constant.
GJ
Rearranging equation (2-7) using p2 =
Er
Thus: -
d2 d8 d8 T
(dx )- }12 -Z -------------------- (2-8)
s GJ
To find the general solution for the rate of twist (d8/dx) in equation (2-8) we apply: -
i) Complementary function to solve L. H. S with auxiliary equation.
m2-pZ=0
m=±p
d8
Accordingly, (dx ), = Cd' + De"
d8
(dx )c-F= (C+D) Coshpx + (C-D) Sinhpx
Let (C+D) = C, and (C-D) = C2

,
dO
( )cg = C, Coshpx + C2 Sinhpx (2-9)
dx -----------------
ii) Particular Integral to solve R. H. S
dO
=K
dx
d20 d dO
dx2 dx dx
24
and,
d30 d2 (
dO
=2(ý=0
dx3
substiute in equation (2-8) becomes:-
T
0- u2 (K) _- u2
G7
T
K=
GJ
d8 T
( )pl
dx GJ
So the general solution for the rate of twist is: -
dO T
= Cl Coshpx + C2 Sinhpx + ----------------- (2-10)
dx GJ
Where (x) is measured from the built-in end (see figure 2.6), and (C, ) and (C2) are
constants determined from the boundary conditions.
As far as warping is concerned with the chassis frame subjected to any type of
loading such as, torsional or longitudinal load, equation (2-10) is the basic equation
to be executed with the appropriate boundary conditions for free and completely
inhibited warping conditions at the ends of the members, and also in the case of
partially inhibited warping.
2.8 COMPLETE WARPING RESTRAINT

For a uniform member with complete warping inhibition at both ends it can be
shown that, using symmetry, the correct result for torsional stiffness can be obtained
by considering it as two torsional cantilevers half of the length as shown in figure
(2.6).
25
Therefore, for an open section cantilever, the arbitrary constants (C, ) and (CZ)
in equation (2-10) can be obtained using the following boundary conditions.
i) at the warping inhibited end where x=0, W=O
d8
Thus: - =0
dx
T
C'
GJ
ü) at free end where x=L ßr =0

,
Thus: -
d20 =0
dx2
T
C2 = tanhpL
GJ
Subsituting these constants for (Cl) and (C2) in equation (2-10) becomes:-
d8 T Coshp(L-x)
(1- ) (2-11)
dx GJ CoshpL
in which (L) is the beam span.
The first term of equation (2-11) corresponds to the rate of twist in the
unconstrained (free warping) open section beam. The second term represents the
reduction in the rate of twist due to axial restraint.
After integrating equation (2-11) with the appropriate boundary condition (i, e
0=0 at the warping inhibited end, where x=0). The angle of twist (0) thus can be
obtained as:-
T Sinhu(x-L) - SinhpL
0= (x+ ) (2-12)
GJ pCosh}iL
26
At the free end where x=L, the angle of twist is: -
TL pL - tanhiL
gTP (2-13)
GJ pL
Thus, defining an effective torsion constant (Jr) for the member where warping is
completely inhibited at the end as:-
TL
7-
Jr ---------------- ------ --- (2-14)
G81
and subsituting for (8., ) from equation (2-13) get: -
pL
Jr 7( ) ----------------------------- (2-15)
pL - tanhpL
Subsituting equation (2-11) in equations (2-6) for a beam completely restrained from
warping at one end and subjected to a torque at the another, the bimoment (B), the
warping displacement (W) and the longitudinal normal stress variation along the axial
direction of the beam (x) can be written as:-
Tws Coshµ(L-x)
Wr(x)= - (1 -
GJ Cosh)iL
I
t
t
T Sinhp(L-x) t
Br(x) _- i (2-16)
-------------------------
GJ CoshpL
t
To Sinhp(L-x)
6ýý`ý
pr CoshpL 1
27
2.9 PARTIAL WARPING RESTRAINT

So far, the boundary conditions are taken to be simplified extreme cases of
warping behaviour for the end conditions in a bay of open section chassis frame.
The boundary conditions discussed so far are:-
i) free warping behaviour of the beam cross section in the case of unrestrained
torsion (St. venant).
ii) completely restrained warping at some section of the beam in case of restrained
torsion.
In a real chassis frame the cross members are restrained at the joint by the side
members. Therefore, the boundary conditions in the joints are partially restrained from
warping and the connection is partially built-in as shown in figure (2.5b). The degree
of partially restrained warping can be assumed to be directly proportional to that of
free warping in the joint obtained by equation (2-1), e.g see reference (23) of the
bibliography. The constant of proportionality, (K), which is called the warping
restraint factor, is determined by the degree of restraint provided by the side members.
Thus: -
dO
Wp =K* WF =-K ass ---------------------- (2-17)
dx
Where (Wp) is partial warping displacement of the cross section,

(WF) is free warping displacement of the cross section.
When K=1, the cross sections of the cross members of the chassis frame are
completely free to warp and, when K=O the cross section is completely restrained
from warping in the joint.
The warping restraint factor, (K), can be obtained by dividing the area under
the partial warping curve round the cross member section (which can be obtained
from a finite element analysis of the detailed joint or from experiments) by the area
28
under the free warping curve round the same section. This is similar to the approach
used by Lee (37). The method was chosen in preference to taking simple ratios at
single points. The justification for this may be seen from comparisions made in
chapter (7) for every joint examined.
Comparing equation (2-17) with equation (2-1) we get: -
dO T
=K ------------------------------------ (2-18)
dx GJ
This gives a new boundary condition for equation (2-10) in solving for (dO/dx), and
the modified rate of twist of the beam is shown by Alade (19) to be: -
d8 T Coshp(L-x)
_ (1 - (1-K) ) ------------------ (2-19)
dx GJ CoshpL
As in the case of free warping, the angle of twist is given with the proper boundary
condition being 0=0 at x=0.
Hence:-
T (1-K)(Sinhp(x-L)-SinhpL)
9=-
GJ pCoshpL
and the total angle of twist (Or,) at x=L would be: -
TL pL - (1-K)tanhpL
8ý ( ----------------- (2-21)
GJ pL
Hence, the effective torsion constant for the cross member is then given by either: -
TL
Jp GA ---------------------------------- (2-22)
.rp
29
or;
pL
Jp=J( (2-23)
pL - (1-K) tanhpL
Therefore, equations (2-16) for partially restrained warping can be written as:-
Bp() _ (1-K) Bo(x)

(2-24)
6P(,,) = (1-K) 6r(x)
Measured values by Alade (19) for open section cross members in isolated joints
agreed closely with the predicted values using the above expressions.
30
a) Channel section b) I-section
c) L-section d) T-section
e) Z-section f) Top-hat section
Fig. (2.1) Typical types of thin-walled open section beams

31
ýý
Unrestrained torsion b) Free warping displacement

a)
Fig. (2.2) Channel section beam under pure torsion
Fig. (2.3) Shear stress variation of an open section

across the thickness under 'torsion
[St. Venant theory assumes a linear variation

of shear stress through the thickness)
32
a) Generation of sectorial area of an open section beam
b) Channel' section c) I-section
Fig. (2.4) Distribution of sectorial area

33
a) Completely inhibited warping b) Partially inhibited warping
Fig. (2.5) Warping inhibition in open section beams
End with
X inhibition
warping
Fig. (2.6) Boundary conditions for the chassis

frame cross members
34
m
klýl B=2M*h p*ws
P'. IP"
/h
-e
/M
/'
G'
./Pp
Two bending moments b) Longitudinal load

a) opposite
M*e B=M*h
/h
ticc, eh
tiý
G'
'
Cc,
c) Pair of normal loads d) Pair of lateral loads
Fig. (2.7) Loads that introduce a bimoment into a channel section

35
a) Total torque
\EN.
ý R
/
Ty
Tr
St. Venant torque b) Torque due to bimoment

c)
Fig. (2.8) The distortion of an I-section due to torsion

36
Table (2.1) Comparision of the sectional properties
Items Sectorial properties Conventional properties
center of shear centre centroid

properties
co-ordinate sectorial co-ordiate X, Y co-ordinate

required X, Y co-ordinate
principal J(o dA Jxy dA

sectorial area
Jxo) dA,
statical JywdA fx dA, JydA
moment
2nd moment of
sectorialarea JO Jx2dA, Jy2dA
dA
shear centre centroid

CHAPTER THREE
STIFFNESS MATRIX. OF
THIN-WALLED OPEN
SECTION BEAMS
37
3.1 GENERAL
Mainly there are two matrix computer methods which may be applied to analyse
vehicle structures. The first is called the force (flexibility) method, while the second
is called the displacement (stiffness) method.
An excellent comprehensive bibliography together with an account of the

historical developments of these methods may be found in a paper by Argyris (41).
Stiffness method is the primary method used in computer analysis of structures.

One of its advantages over the flexibility method is that it is conducive to computer
programming. Once the analytical model of the structure has been defined, no further
engineering decisions are required in the stiffness method in order to carry out the
analysis. In this respect it differs from the flexibility method, although the two
approaches have similar mathematical forms.
In the flexibility method the unknown quantities are redundant actions that must
be arbitrarily chosen, but in the stiffness method the unknowns are the joint
displacements in the structure, which are automatically specified. Thus, in the stiffness
method the number of unknowns to be calculated is the same as the degree of

kinematic indeterminacy of the structure.
In this chapter the stiffness method is developed on the basis of writing joint
equilibrium equations in terms of stiffness coefficients and unknown joint

displacements. The method is formalized into a procedure for assembling the overall
stiffness matrix of the structure from individual member stiffness matrices.
3.2 FUNDAMENTAL CONSIDERATIONS

In chassis frame structures, the axial warping displacements due to plane
deformations will influence considerably the structure with respect to the strength
and torsional stiffness. When designing a chassis frame, very often warping at the
joints has been regarded as rigidly restrained, as shown in equation (2-15) of the
previous chapter.
38
In this chapter, however, the warping displacement at the joints is considered
as being completely transfered to the other connected parts as far as the joints are
concerned.
When warping inhibition is included in the matrix stiffness method, the stiffness
matrix must be increased in size. The rate of twist is added to the displacements, and
the bimoment is added to the loads. Hence, all the mathematically relevant quantities
such as St. Venant torsion, flexural twist and bimoment are derived and used to
modify the conventional (6) degrees of freedom beam structural analysis to the newly
developed one which has (7) degrees of freedom for each node. Accordingly, the
condition of bimoment equilibrium at the joints is introduced for the derivation of this
stiffness matrix.
3.3 EQUILIBRIUM AND COMPATIBILITY

As far as the chassis frame consisting of open section members is concerned,
warping in the structure tends to cause a significant gap between the theoretical
approach and experimental results. This is attributed to the difference of the
conventional concepts, regarding the transmission of the load between the open section
members, and those generally accepted.
Generally, it is considered that the torsional moment along a cross member will
cause only bending moment on the side member, as shown in figure (3.1b), without
causing warping. However, the actual behaviour shown in figure (3.1a) occurs such
that both the cross member and the side member experience torsional moment as well
as bending moment. This phenomenon can be regarded as a result of transmission of

warping produced in the cross member to the side member.
The aim of a static analysis is to determine the internal loads and displacements
of a structure when subjected to external loads. The basis for this analysis is that the
equilibrium of forces and the compatibility of displacements shall be maintained at
all points in the structure. Therefore, in order to interpret the joint behaviour shown
in figure (3.1b), the condition of bimoment equilibrium: ,"-at the joint must be
39
introduced, (EB; 0) in addition to that of torsional and bending moment equilibrium

(Y,M; =O) at the joint, as shown in figure (3.2).
Consequently, the relationship between bimoment and the rate of twist, i. e, plane
deformation, which is mathematically based on Vlasov's theory of thin walled elastic
beams, will be added to the conventional (12x12) beam stiffness matrix, giving a
(14x14) stiffness matrix for each element as will be shown later.
3.4 DERIVATION OF THE STIFFNESS MATRIX

There are four basic design quantities for thin walled beams. Two of them are
kinematical terms; these are the angle of twist (0) and the warping which is related
to the rate of twist (0'). The other two are statical items; they are the bimoment (B)
and the total torsional moment (Mx).
It is required mathematically to derive the relationship between these four
quantities. It shown in chapter two that the latter two items (B&T) can be expressed
in terms of the former kinematical quantities (6&8') as the follows: -
B - EIFO' ----------------------------- ----------- (3-1)
Tr =- EFO ----------- ---------------- ------------- (3-2)
T, =GTY (3-3)
As shown in equation (2-4), the total torsional moment (Mx) consists of the
sum of the flextural twist (Tr) due to the axial stresses and of the St. venant torsional
moment due to the non-uniform distribution of the tangential stresses over the
thickness of the wall.
40
Thus: -
Mx=Tr+Ty=-Er0`+G7A' (3-4)
Where (9) is (d8/dx), and by differentiating w. r. t (x) we get: -
Ere"-G78r=o
0""-11,0'=0 (3-5)
GJ
Where; (}1) is dimension constant, u2 =
Er
The solution can be written in the form: -
eýx,= Cl + C2x + C3 Sinhplx + Cy Coshpx -------,
and, by differentiating this we obtain: -
6"(x)= CZ + }i C3 Cosh}ux +p C4 Sinhpx -------4 (3-6)
Hence by further differentiation of 8'(x)
equation (3-1), and equation (3-4) become: -
B(x) GJ (C3 Sinhpx + C, Cosh)ix) -------
Mx(, = GJ CZ ------- ------------- ------------

)
Having found the relationship between four basic design quantities (Ax, 0, B., Mx)
in the above equations. The four unknown constants (C CZ, C C, ) can be obtained
41
by applying the following boundary conditions at both ends of each beam element as
shown in figure (3.3).
Where at x=0,0=0, B=B1 Mx=-Mx, ---------- --- (3-7)

,
at x=L, 0=92 B=B2 Mx=Mx2 (3-8)
and; ,9 =9 2, , -------------
Applying the first boundary conditions as in equation (3-7) to equations (3-6) get: -
B,
B, =C, +C4 C, = 8, + ----
GJ
solving
yields C2
Nix,
the
e',=C2+PC3 values GJ
---
-
of the (3-9)
constants
B1=-C, GJ C3=1 (8'1+ 1)-

j
G
B,
Mx, =-C2GJ C4 =- ------
GJ
Substituting the constants obtained in (3-9) into equation (3-6) get: -
B, Mx, 1 Mx, B,
ß(x) e' + . x+-(8, + ) Sinhpx - Coshpx
GJ GJ u GJ GJ
NU' Mxl pB,

0'(x)=- + (0', + ) Cosh}ix - Sinhpx
GJ GJ GJ
GJ Mx1
B(x) (8ý, + ) Sinhpx + B1 Coshpx
u GJ
Mx()_-Mxl
Rearranging all the above equations in terms of ( 8,0', B, Mx ),

42
111 Sinhpx
0(X)= 0, +(- Sinhpx ) 0', + (1 - Cosh)ix)B, (x - )Mx,
GJ - GJ ---i
1
(Coshpx) 0', - (p Sinhpx)B, (1 - Coshpx ) Mx,
GJ GJ ----
(3-10)
GJ 1I
B() (Sinhpx)8", + (Cosh}ix) B, -u (Sinhpx) Mx, ------
MX(x) _- (1) Mx, ---J
The above equations give the load-displacement relations at the ends of the beam.
Now by applying the second boundary conditions as in equation (3-8) to equations
(3-10) we get the following: -
Suih)iL
e, (1 SinhpL)e'1+ (1-CoshpL)B, - (L- )Mx, -(3-11)
e2= +
GJ GJ
p
11
6"2 = (CoshpL)8, - (p Smh}iL)B' (1 - CoshpL)Mxl (3-12)
GJ GJ ---
GJ 1
B2 =- (SinhpL)8'1 + (CoshpL)B, - (SinhpL)Mx- (3-13)
------
Mx2 =
- (1) Mx, (3-14)
------
As far as the stiffness matrix is concerned, all of the member forces have to be
arranged in terms of the node displacements, i. e,
From equation (3-11) and equation (3-12), the following expressions can be obtained:
-
Mx, = K19, - K26'1 - K, 92 - K202 ----------------------- (3-15)
B, = -K2 1+ K38'1 - K282 + K48'2 (3-16)

------------------------
43
Where;
GJ}i SinhpL
K, -D ---------------- (a)
GJ (1 - CoshpL)
K= ------------ (b)
ZD
GJ(pL CoshpL - SinhpL)

K
K3 ------ (C)
pD
GJ( SinhpL - pL)
pD
D=2 (1-CoshpL) + pL (SinhpL)
Since there is no external torque applied along the beam element, the internal torque
is the same at all points along its length, therefore: -
Mx=Mx2=-Mxl
Hence, equation (3-15) becomes:-
IMIx2=- Kiel + K26ý1+ K, 82 + K20 2 (3-17)
Finally, from equation (3-13), the expression for (B) can be obtained as:-
B2=-K28, +K, A', +K202+K30,2 (3-18)
Consequently, the above equations can be combined in the matrix equation as

follows: -
44
Mx, K, -K2 -K, -KZ 6,
B, K3 K2 K4 0
-KZ
--------------- (3-19)
MX2 -K, K2 K, KZ 82
BZ -K2 K4 K2 K3 82
Where (KKZ, K3,K4) are given in equations (a), (b), (c), and (d) respectively.
The load-displacement equation for warping can be expressed as:-
(Mi BiMý2B2)T=Kw(ei e'i 8202)
Where (Kw) is the stiffness sub-matrix for the warping terms.
Therefore, the additional 7th degree of freedom due to the bimoment effects and
the rate of twist are incorporated with those due to torsional moment along the axial
direction of the beam.
Consequently, the beam stiffness matrix has been written using the direct stiffness
method, as shown in figure (3.5). The positive load and displacement direction are
shown in figure (3.4) with four discrete forces used to indicate the bimoment. The
sign convention is chosen in order that a positive twisting angle (8, ) will be
associated with a positive twisting moment (Mx, ), where the other three degrees of
freedom (ß',, A2,8'2) have a value of zero and positive twisting angle (82), will be
associated with a positive twisting moment (Mx2), where are zero. The same
rule will be used for the warping mode and the bimoment, i. e, a positive warping
mode (8', ), will be associated with a positive bimoment (B, ), where (882,8'2 ) are
zero, and the same for (8'2) and (B2).
3.5 TRANSFORMATION FOR STIFFNESS MATRIX

Compatibility and equilibrium are the basis for assembling the global stiffness
45
matrix. So far the element stiffness matrix is established with reference to the local
co-ordinates. It must be transformed into the global co-ordinates. Therefore, it is
required to derive the transformation matrix associated with (7) degrees of freedom.
Through the transformation matrix, all the quantities such as displacements and forces
pertaining to the local co-ordinates can be related to the global co-ordinates by taking
into account the geometrical relationship between the two co-ordinate systems.
Many chassis frames consist of plane grillages. For this case, the procedure to
get the transformation matrix is such that Y-axes of the global and local co-ordinates
are parallel, i. e, when a grillage structure is analysed.
g=[T1]g (3-20)
Where;
g; denotes the quantities with reference to global co-ordinates.
g; denotes the quantities with reference to local co-ordinates.
[T;]; is the transformation matrix.
Referring to figure (3.6), the following relation can be written as follows: -
Pr -r ---------------
Mx = Mx Cos4 - Mz Sind ------

------------------- (3-21)
Mz = Mx` Sin4 - Mz' Cosh ----1
--------------- j
The direction Cosines of the local X-axis may be written in terms of the projections
as follows: -
46
LX'
Cosh = ------------------1
---------------------- (3-22)
ZL-7,
Sine = -------------------I
Where;
L= (X, -X; )2 + (Z;-Z1)2
Hence, the transformation matrix for a grillage member can be expressed as:-
11 000
[Tj = 10 Cos$ -Sind 0 ----------------------- (3-23)
J0 Sind Cos4 0
0001
A more general approach for the (3) dimensional case is shown in figure (3.7).
If the components of a vector quantity (V) are (x, y, z) in the local axis system, and
(x', y', z') in the global axis system. Then the relationship between (x. y, z) and (x', y', z')
is: -
x ' 11 m, n, x
' 12 m2 n2 y
y =
z 13 m3 n3 z
Where (1,,m,, n, ) are the direction Cosines of the local (x,y, z) axes w. r. t the global
x-axis.
(12im2,n2) are the direction Cosines of the local (x, y, z) axes w. r. t the global
y-axis.
(1,,mn3) are the direction Cosines of the local (x, y, z) axes w. r. t the global
z-axis.
47
(V) must be (or approximate to) a vector quantity for this transformation to be
valid. Forces, moments, small rotations satisfy this requirement.
Or, expressing grillage transformation matrix in equation (3-23) in more generalised

form for node (i), which has (7) degrees of freedom we have: -
1, m, 0
l2 0
m2
001 (3-25)
11 ml 0
12 m2 0
001
1
A transformation matrix for the beam element (i j) will be: -
[ETJ0
(3-26)
cri;] _
0 (Z',
]
Consequently, the transformation from local to global axes for the (14x14) element
stiffness matrix will be: -
[K; = F'I) i j1L[TIj]r (3-27)

jlIc --------------------------------
Where;
;j],, is Global element stiffness matrix. and,

[K; is Local element stiffness matrix.
J]L
48
N,
, \BENDING TORSION
a)
0.114,
TWISTED
TORSION
b)
Fig. (3.1) Mode of load transfer
ý(_ -
(Q)tr 0 {b) 4=o
Fig. (3.2) Moment and bimoment equilibrium at a joint

49
Mzl
Pzl
Bl
Mxl "^2
PxI L Px2
Fig. (3.3) Force and displacement components in local coordinates
mx1
e,
./ e2 7ý,ýý
1Vý2
B'
0
B2
x
8=
1
L
Fig. (3.4) Sign convetion for loads and displacements in

a warped channel-section beam element
50
Px, C, 0000 0 0 -C, 0 0 0 0 0 6 x,

PY1 C2 000 C6 0 0 0 0 0 C6 0 SJ1
-C,
Pz, C, 0 -C, 0 0 0 0 -C3 0 -C, 0 0 SZ,
Mx, K, 0 0 -K2 0 0 0 -K, 0 0 -K2 6x,
My, C4 0 0 0 0 C, 0 C8 0 0 @Y,
Mz, C5 0 0 -C6 0 0 0 C9 0 0z1
B, K3 0 0 0 K2 0 0 K4 0
Px2 C, 0 0 0 0 0 0 Sx2
PY2 C2 0 0 0 -C6 0 5y2
Pz2 C, 0 C, 0 0 Sze
Mx2 SYMMETRIC K, 0 0 K2 0z2
My, C4 0 0 eye
Mz2 C5 0 0,
B2 K3 8z
Where, the following are the convectional stiffness co-efficients
C, = E*A/L C6 = 6*E*ZJLZ
C2 = 12*E*ZJL3 C7 = 6*E*Ys/I2
C3 = 12*E*Y11L3 C8 = C,/2
C4 = 4*E*y; /L C9 = C5/2
CS = 4*E*Z; /L
And, K,, K2iK3,K4 are the co-efficients concerned with bimoment as
given in equations (a), (b), (c) and (d) respectively.
Fig. (3.5) Stiffness matrix for beam element with bimoment terms
51
Z
Mz2ý\w
e12
`\
my-
Mzl
Zý
MxI
exl 1
13,81.
x
Fig. (3.6) Coordinate system for a grillage
Fig. (3.7) A generalised coordinate for 3D frame

CHAPTER FOUR
THIN-WALLED
CLOSED-SECTION BEAMS
52
4.1 GENERAL
Warping effects in thin walled closed section beams are less important than those
in open sections. If closed section cross members are used in chassis frame, the
problem of torsional stiffness due to warping inhibition would disappear since there
is no significant difference in the torsional stiffness of a closed section beam which
is fully restrained against warping and one which is totally free to warp, unless the
beam is very short, which it would not be in this application.
However, the compatibility of any warping displacement in the cross member
and the rate of twist in the side member has to be ensured at the joints, even when
warpless sections, such as circular or square sections, are used.
The assumptions used in thin-walled open section beams such that the cross
section shape remains constant after deformations and that there is negligible shear
deformation, do not hold for thin walled closed section beams as they lead to
kinematically impossible deformation.
The additional displacements arising from the deformation of the cross section
can be added to the displacements assumed for open sections beams, while the
corresponding so-called lateral bimoment can be added to the internal loads. These
relations are more complex than those developed for open section beams.
In this chapter the load-displacement relations for torsion in closed sections will
be derived for rectangular box beams, as they are the most common sections used in
practice.
4.2 WARPLESS CLOSED SECTIONS

The free warping of a closed section subjected to torsion is well known and
documented by various authors (17,37). The formula for the warping displacement
of an arbitrary closed section is; -
53
J(llt)ds
aw, T (5 as jp ds) + Wa (4-1)
ww=las ds+ wa = 2AG t - 2A ---
where (p) is the perpendicular distance of the tangent at a point on the perimeter
(S) to the shear centre as shown in figure (4.1) and, (W. ) is the value of warping
displacement (W. ) where (S=O), and (A) is the area enclosed by the mid-line of the
tube wall.
Since {j(1/t)ds} is a constant for any given section of a tube, the warping
displacements will be a function of the values of (t) and (p) at the point considered
and at all the points from some arbitrary datum on the section to this point. If (pt)
is constant, (t SW/as) must be constant, but (t) is always positive, while (We) is a
continous function having opposite signs for different values of (s). Therefore, in order
for (aW/as) to be constant it must be zero.
Hence, the condition for zero warping is that (pt = constant) round the perimeter.
This means that thin walled beams whose sections are circles, squares, triangles and
other regular ploygons will not warp if the material thickness (t) is constant since all
these cases the distance of the tangent from the shear centre is constant.
If it is particularly important that no warping effects are present and the shape
of the beam is not regular the thickness can be varied (in steps if necessary) to keep
(pt) constant.
The free warping of rectangular box subjected to torsion is shown in figure (4.2).
The variation of warping a round the section profile is linear and the axial
displacement of the corners of the box section is; -
Tbh
W, = t (-- -) --------------- (4-2)
8hbG --
t2 t,
54
4.3 COMPLETELY RESTRAINED WARPING

The theory for estimating variation in warping along the length of a uniform
rectangular section thin walled beam subject to a constant torque is well known (e.g
see Megson (17)). It is derived using the more general coordinate system shown in
figure (4.3), and can be summarised as follows; -
d2W. T bt,-ht2
_ u`Zw` (4-3)
dx2 hbB E bt1+ht2
Where;
8Gt, t2
ý`2
BE(bt, +ht2)
1
B, =6 (ht1+bt2)
The general solution for equation (4-3) can written as;-
Tbh
W, = Cl Cosh}icx + C2 SinhpA + (t (4-4)
8hbG
where (x) is measured from the built-in end and C, and C2 are constants determined
from the boundary conditions.
The last term of equation (4-4) may be recognised as the free warping of the
corners of the beam as shown before in equation (4-2). Other important results such
as (dcp/dx) are derived as a matter of course in the development of the theory for the
variation of (We) along the length of the box section as; -
55
dcp 4WW(bt,-htz) T
=+ -------------- (4-5)
dx hb(bt, +ht2) hbG(bt, +htz)
For a uniform beam with complete warping inhibition at the built-in end. Thus; -
WW=O at x=0
Also at the free end the direct stress is zero since a pure torque is applied.
Therefore; -
awc
6r(x =U
)-E
ax
Substituting these boundary conditions in equation (4-4) gives; -
Coshp(L-x)
Wir = Wý (1 -) --------------------- (4-6)
Cosh)i,,L
The variation of direct stress is given by; -
6r
awc
=E ax
Differentiating equation (4-6) and substituting gives; -
Sinhpc(L-x)
6r )i `F Coshp,L
Subsituting equation (4-6) into equation (4-5) get;-
dcp Tbh T(bt, +ht2)2 Coshp, (L-x)

(4-8)
dx 2h2b2G t2 t, 2h2b2Gt,t2(bt, +ht2) Coshpc
56
The first term of this equation corresponds to the rate of twist in the
unconstrained rectangular section tube as predicted by the Bredt-Batho theory. The

second term represents the reduction in the rate of twist due to axial constraint.
Rearranging equation (4-8) and integrating with the appropriate boundary

conditions over the length (L), the angle of twist ((p1) at the free end where (x=L),
thus can be obtained as;-
TL pcL- [(bt, -ht2)2/(bt,+ht2)1 tanh}iL

cp` { } (4-9)
GJ, --------
u,L
where (Je) is the torsion constant, which is given for a thin walled closed tube by
the Bredt-Batho theory as;-
4A2
J`
S
where (A) is the area enclosed by the mid-line of the tube wall and,
as
s=!
t
Hence, for a tube of length (L), which is constrained against warping at one end,
the effective torsion constant (Jr) would be; -
TL
Jr = ---------------------------------------- (4-10)
r G(pr
and subsituting for ((Rr) from equation (4-9) get; -
)1`L
J.r = J. { } --------- ---- (4-11)
p,L-[(bt, -ht2)2/(bt,+ht2)2]tanh)i,L
57
4.4 PARTIALLY RESTRAINED WARPING

In chapter (2), section (9), it was seen that for open section thin walled cross
members the assumption that partial warping was directly proportional to free warping
was valid for the sections analysed. The same assumption is made in the case of
closed rectangular section cross members.
Thus;
W =K*W ------------------------------ (4-12)
Where (K) is the warping restraint factor, (Wa) partially restrained warping
displacement and (W, ) the free warping displacements.
This gives a new boundary condition for equation (4-4) in solving for (We). Hence,
the modified expression is; -
(1-K)Cosh}i, (L-x)
Wcp = WCP [1- (4-13)
Coshj, L
From this result, the direct stress expression can be given as;-
Sinhp, (L-x)
(fp = )icEWc(1-K) (4-14)
Coshp,L
substituting the result of equation (4-13) into equation (4-5), then substituting for
(Wa, ) from equation (4-2) and rearranging gives; -
dcp T (1-K)(btl-ht)2Cosh}i, (L-x)

----------- (4-15)
-=L1-] GJ (bt, +ht2)2 CoshpL
dx
The angle of twist, ((p, ), of the beam at (x=L), relative to the partially restrained
end may be found by integrating equation (4-15); -
58
TL p,L-(1-K)(bt, -ht2)2/(bt,+ht2)2tanhp,L
(P" [L] --------- (4-16)
GJ
Hence, the effective torsion constant for the cross member is then given by either;
TL
JCP (4-17)
G
or;
p,L
Jcp=J, [ ] ------ (4-18)
p L-(1-K)(bt, -ht2)2/(bt,+ht2)2tanhp,L
4.5 THE STIFFNESS MATRIX

In this section a stiffness matrix including warping effects as a seventh degree
or freedom is developed for the closed section member. The total deformation of
thin walled closed section beams can be separated into three parts, as shown in figure
(4.4). Each can again be divided into two parts, the shape function depending on (s),
and the magnitude depending on (x), the distance along the beam.
Therefore, the warping displacement (u), of any point can be written as the
product of the two functions; -
U= X(S) W(x) (4-19)
The terms which depend only on the dimensions of the cross section, as shown
in figure (4.5), can be integrated in turn to give the following coefficients;
-
59
1
JA x2dA b2h2(f1+f2)
a, = =
24
i
äs
jA )dA =2 (b2f,+h2f2) (4-20)
a2 =
1
JA ýl dA =2 (b'fl-h2f2)
a, =
where; f1 = ht f2 = bt2
Table (4.1), summarizes the cross section constants and the load-displacement
relations for closed section and compares them with those already derived for open
section thin walled beams.
The load-deformation relations for closed section beams are considerably simplified
if the lozenging and the lateral bimoment are neglected as in reference (39).
Therefore, using the same procedures as in chapter (3), section (4), the four
basic expressions can be derived as;-
1 a4 1
cpcx>
= +
cp, a, (- Sinhp x)W, + CoshpCx)B, - a42Sinhpx)Mx,
PC a,Ep. 2(1- a,Ep 3(px-
1 a,
WC(X)
_ (Coshpx) W, - (Sinhp,,x)B, -Z Ep, (1-CoshpCx) Mx,
a,EpC a,
a`
B, (Sinh)ix)W1 + (Coshp,,x)B, (SinhpCx)Mx, I
(X)= -Ea1P, -
Mx(x) _ (1) MX1

'
60
where ;-
48G t, t2
P2
E (ht2+bt,)(f1+f2)
a3
a4 =-
a2
The constants a a2 and a3 are given in equation (4-20).
The load-displacement equation for warping in closed section beams can be
expressed as;-
(Mx, B1 Mx2 BZ IT = Kwc { cci Wl 92 W2 IT ----------------

(4-21
where (Kv) is the stiffness sub-matrix of the warping terms, and can be obtained
from equation (4-20) as;-
[K1
-K2 -K, -K2
Kwc = K3 K2 1e (4-22)
-K2
-K1 K2 K1 K2
-K2 K4 K2 K3
where Kl KZ K3 and K4 are given in table (4.2) with a comparison with those
, ,
already derived for open section beams.
61
ickness
point s
Fig. (4.1) Generation of warping displacement of closed section
))p
oo
(4.2) Warping of rectangular box section

Fig.
62
Fig. (4.3) General coordinate system of thin-walled tube
Z
ý2
-. - X
Fig. (4.4) Displacement function of a box beam

63
Iz ax
as
b/2
---;. _y
El
)/2 b/2
Fig. (4.5) Displacement function in the plane

of the cross section of a box beam
64
Open cross section Closed cross section
u=0) 8, u=XW
ü =Jrds xyz
r=lAýSZdA a,=SAXda
B=Ere' B, =Ea, w
JA Bc dA
acs
B= as U1-ß = JA
6ýs = (B1a, )x
as = (B/(YS)Ws
Table (4.1) Summary of functions used in the analysis of open and

closed section beams
Open cross section Closed cross section
GJµSinhpL Ea, }k2SinhpL

K,
{ 2(1-CoshpL)+p. LSinhµL } { 2a42(1-CoshgL)+}. tLS inhpL }
GJ(l-CoshpL) Ea,a4}.
tý2(1-Coshp L)
K,
(2(1-CoshµL)+µLSinhpL) } { 2a42(1-Cosh},
tcL)+VcLSinhpcL )
GJ(µLCoshJL-SinhµL) Ea,p, (pLCoshVL-SinhpL)

K,
µ{ 2(1-CoshµL)+µ. LSinhµL } { 2a42(1-Cosh[.
tLL)+pcLSinh}. tcL )
GJ(SinhµL-µL) Ea,}.tja42Sinhp L-kL)

K4
µ(2(1-Cosh, JL)+VISinh1L) 12a42(1-CoshpcL)+pcLSinhg L}
Table (4-2) Summary of constants used in stiffness matrices of open and

closed section beams
CHAPTER FIVE
CHASSIS FRAME JOINTS

65
5.1 GENERAL
The stiffness of joints in a chassis frame can have a significant effect on
bimoment distribution in it. Joints can be classified as rigid or flexible. In rigid joints
all the member displacements are fully transfered to the other members, but in
flexible joints the displacement transfer at the joint has to be interpreted with respect
to the joint flexibility.
It is well known that direct load which does not act along the axis of the
centroid of the cross section causes bending moment. Similarly, normal loads which
do not act through the shear centre cause torque, and as shown in chapter (2-6), direct
loads which do not act through the points of zero warping may cause bimoment.
Finally, it is also shown in the same chapter, that couples made up of normal or
lateral loads whose plane does not pass through the shear centre produce bimoments.
Therefore, it is shown in this chapter that there are different beam axes for the
various generalized forces and displacements, and these have to be taken into account
in order to avoid serious errors when considering joints.
5.2.1 RIGID JOINTS ASSUMPTIONS

In chassis frames structures, the elastic properties of the joints are as important
as those of the beams. As shown in the previous section, it is not correct to use the
intersections of the centroid axes as a node without noting the effect of the
intersections of the torsion centre axes and the zero warping lines along the beams.
The zero warping lines are lines parallel to the axis of the beam passing through the
points of zero warping in the cross section as shown in figure (5.6). Taking rigid joint
assumptions into account when assembling the total stiffness matrix, the effect of
different axes must be included. This can be done by transforming the non-coincident
load and displacement components into the axes defined for the node.
5.2.2 JOINT COMPATIBILITY

The ends of beam elements at the joint can be defined by the intersections of
the centroid axes. The intersections of the torsion axes and the zero warping axes
will not be at the ends of the beam elements defined in this way. The zero warping
66
axis is defined as the axis formed by the intersection of the plane containing the zero
warping lines in the flanges and the plane containing the centroid and the torsion axes
as shown in figure (5.6).
Therefore, the node must be defined by three points (as introduced in reference
39), these points are the torsion centre (T), the centroid (S) and the warping point
(W). Such a node is called (TSW) node. The warping point (W) is the point where
the zero warping axes meet the end plane of the beam as shown in figure (5.6). The
lack of coincidence of the axes between a channel section cross member with
horizontal web and a channel section side member with vertical web are shown in
figure (5.1), where the centroids of the two channels (S) and (Sc) are separated
vertically by the distance (h3).

.
When assembling the total stiffness matrix, the stiffness matrix [<J relating to
the (Tc Sc Wc) node must be transformed to the matrix [K] relating to the (TSW)
node. The relationship between the loads and the displacements is well known and
can be written as;-
[P]=[K][d] (5-1)
or; -
[P,]=[K] [dr.] ------------- ------------ (5-2)
where the subscript (c) is denoted to (Tc Sc Wc) node, i. e, the node at the end of
the cross member.
But; -
[ý']=[H][Pý] --------------------------- (5-3)
or; -
67
[dal=[H]T[d] (5-4)
where [H] is the equilibrium matrix to allow for the (TSW) node.
Substituting equation (5-2) into equation (5-3) get; -
[P]=[H] [Kr] [dC] ------------------------------ (5-5)
From equation (5-5) and equation (5-4), the following expression can be obtained; -
[P]=[H][K][H]T[d] --------- ------------- ------------ (5-6)
Comparing equation (5-6) with equation (5-1), the following obtained; -
[KJ=[H] [I] [H]T

--------------------------------- (5-7)
Equation (5-7), gives the necessary relationship between the stiffness matrix relating
to the (TSW) node and that relating to (T, S, W, ) node.
Combining equation (3-32) and equation (5-7), a comprehensive stiffness matrix for
cross members in globalised system can be written as;-
K'jJ0=[Tij] 1a ]T[ 1
lijý L11ij1 1 ij]T
Where;
[T; is the standard transformation matrix between local and global axes
j]
[I-Lj is equilibrium matrix for beam (ij) including (TSW) nodes.
This is a clarification of the approach in reference (39), where the 'equilibrium and
the standard transformation matrices were combined in a single so called
"transformation matrix". In reference (39) the transpose of this matrix is incorrectly
refered to as the 'Transform' of the matrix.

68
5.2.3 EQUILIBRIUM MATRIX

The equilibrium matrix of the (TSW) node depends on both the profile of the
cross sections and the orientation of the beams meeting at the joint. For commercial
vehicle chassis frames where the beam elements meet at right angles, the equilibrium
matrix can be straightforward enough for practical use.
The side member is taken as a channel section with vertical web and the flanges
pointing towards the centre of the vehicle. The cross members are taken as a channel
sections and may have; -
i) Vertical web with the flanges pointing either to the left or right.
ii) Horizontal web with the flanges pointing either up or down.
The derivation of the equilibrium matrix for each of these cases will be dealt with
separately.
For the system shown in figure (5.2), with a vertical web channel cross member
with the flanges pointing to the left, i. e, case (i) above, the equilibrium matrix may
be written as;-
1 00 0 0 0 0
0 10 0 0 0 0
0 01 0 0 0 0
[H] = 0 -hs -RT 1 0 0 0 (5-8)
-----------
hs 0 ST 0 1 0 0
0 00 0 0 1 0
0 hSSa, 0 -S, Ra, 1
-hs
where the constants (h3, RT, S. Rw and S, )are given in figure (5-5).
r,
When the corresponding axes at the two beam ends do not coincide, the rotation
about one of the axes will make a contribution to the translation of the nodal point.
69
For instance, from figure (5.3c), the displacement (Sys) is at a distance (hs) from the
centroid axis through (S) where the rotation is (O) about the y-axis of the side
member. Therefore, this gives a displacement of (-hS8) in the y-direction of the cross
member (ye).
It is clear that when a direct load has an offset, it will produce a moment. For
example, the force (Py, in figure (5.4c) gives a moment of (-h3Py.) to be added to
,)
the moment (My) in figure (5.4a) as shown in equation (5-9).
The effects on bimoment behaviour of couples made up of normal or lateral loads
as well as moment whose plane does not pass through the shear centre are also
included in equation (5-8). For the system shown in figure (5.2), with a horizontal
web channel cross member with flanges pointing down, i. e, case (ii), the equilibrium
matrix can be written as;-
100 o o0 0
010 o 00 0
001 0 00 0
[H] = 0 -(hs+rT) 0 1 00 0 (5-9)
-----------
hs 0 ST 0 10 0
000 0 01 0
0 (hS+rT)SW0 Sw 0 -(hs+rw) -1
A general equilibrium matrix can be obtained by combining equation (5-8) with
equation (5-9) as follows; -
1 0 0 0 0 0 0
0 1 0 0 0 0 0
0 0 1 0 0 0 0
[H] = 0 -(hs+rT) -RT 1 0 0 0 ------------- (5-10)
hs 0 ST 0 1 0 0
0 0 0 0 0 1 0
0 (hs+rr)SW 0 +Sw Rq, - (hs+ra,) ±1
70
where in all cases the dimension (ha) is positive when the centroid (Sc) is above the
centroid (S).
The equilibrium matrix given in equation (5-10) can be applied to all situations
if the sign of the terms in row (7) of columns (4) and (7), is taken to be the upper
sign for cross members with vertical webs and the lower sign for those with
horizontal webs. The displacements of the nodal points at the end of the beam
element are shown in figure (5.3), while corresponding components of the loads are
shown in figure (5.4). The distances from the centroid axes and their sign convention
for the joints used in the general equilibrium matrix are given in figure (5.5).
5.3.1 FLEXIBLE JOINT ASSUMPTIONS

In the previous section, the joints are taken as rigid. In fact the joints deform
under load as well as the beams. Therefore, the displacement transfer at the joints
has to be interpreted with respect to the joint flexibility.
In short beams, joint deformation is more important than in long beams. Joints
of chassis frames where the length of the beam elements is of the same order as the
cross section dimension of the beam can be regarded as short beams. There are
flexibilities associated with the various member end-forces, but the rate of twist
flexibility is a major interest of this research.
5.3.2 RATE OF TWIST STIFFNESS

The rate of twist stiffness of the joints has considerable influence on the torsional
stiffness of the whole chassis frame as well as the stress distribution in its members.
The rate of twist of a joint can be included in the analysis of chassis frames by the
matrix stiffness method by a rate of twist stiffness coefficient as will be shown later
in the chapter on joint elements, section (5.3.4). This coefficient can be defined as the
bimoment transmitted through the boundaries of a node between a side member and
a cross member divided by the change in the rate of twist between those boundaries.
Therefore, the rate of twist stiffness coefficient of a joint can be represented by the
following expression; -
71
B
(5-11)
i' =
ee'
The stiffness of various joints can easily be compared using this simple quantity,
and it is also used to introduce the joint element stiffness matrix as will be shown
later. This is the reciprocal of the "rate of twist flexibility" introduced by reference
(39).
5.3.3 BENDING AND TORSION STIFFNESS COEFFICIENTS

Bending and torsion stiffness of a joint can also each be represented by a single
coefficient. This coefficient can be defined as the moment transmitted through the
node boundaries between a side member and a cross member divided by the
difference between the twist of the side member and the slope at the end of the cross
member or the slope of the side member and the twist at the end of the cross
member meeting at the joint.
These stiffness coefficients can be directly inserted into the stiffness matrix of
the joint element, as will be shown in the next section. Reference (45) gives a
definition of a stiffness constant for a joint with bending flexibility and shows how
it is included in the beam element stiffness matrix.
As far as this research concerned, joints are regarded to behave rigidly except
for the rate of twist where they regarded to behave in a flexible manner.
5.3.4 THE JOINT ELEMENT

In order to introduce the rate of twist stiffness of a joint into the matrix
displacement (stiffness) method of analysis, the definition of the (TSW) node,
considered in section (5.2.2), must be completed by including a node boundary at
the end of the cross member. Since beams are represented by their centroid axes,
the end of the cross member lies at the centroid of the side member as shown in
figure (5.7).
72
It is necessary to introduce the joint elements to take into account the effect of
the whole joint. The stiffness matrix for such a joint element can be obtained from
a finite element analysis of the joint area (see chapter 6 ). Reference (39) suggested
on the basis of experience that the length of the side member included in the analysis
should exceed the width of the cross member by around 70 percent of the side
member height as shown in figure (5.7).
To find the internal loads in the joint element, the displacements at one boundary
are fixed and unit displacements made at the nodes of the other boundary. The unit
displacements are ;-
1sx öy sZ ex ey eZ e'}=
while the internal loads at both boundaries are found by finite element analysis to
be;-
{Px Py PZ K M, M B)T
Assuming the node has zero length, the joint element stiffness matrix can be
written as;-
K,
x
Key
Ký
ýX
(5-12)
j,, Y
0 Ka
K,
Note: (Cross coupling terms may be present, but these have been neglected in
the above equation)
The equilibrium matrices defined in section (5.2.3) do not apply when a joint element
is used because the axes offset is automatically included in the joint element.
73
MEM
SIDE
I. Is
ol ,2
cenUoid
ýýý
waýlTtä
ýý
re
ý,
ý,
t<ý
(5.1) A joint defined by a (TSW) node with a horizontal

Fig.
web channel cross member
Fig. (S. 2) A joint with a vertical web channel cross member

74
Sz tz
Y
TSys
Sr
8z
SW
a) A vertical channel section side member with flanges pointing towards the centre
t
Z.
xý SZ.
SY. exý
T,
Sxý
04
yc
rN
eyc
b) A horizontal channel section cross member

with flanges pointing down
rw
X.
8; Szc
by. W. ýý
S
ý1
QYc
c) A vertical channel section cross member

with flanges pointing to the left
Fig. (5.3) Displacements of nodal points at the ends of beams meeting

at a joint, used to build up the equilibrium matrix [H]
75
a) y Pz
Mx
PY my Px
-ý-
ST
Mz
Sw
b) xc Pzc
Px, Cj Mx.
My.
Py,
T we .\
Y.-*-- rT
rw
zc
xý
Mz,
Pxc
PZC
My, W
ý
y° S N
Tý
Ra, RTý
to the displacement in fig. (5.3)

Fig. (S. 4) Loads corresponding
76
1
Y
N X
1
a) L. H. S. verticz
side member
Tý
Tý We S
y -`t-ý
Sc
we
rT
b) Horizontal ' c) Vertical

cross member "rw cross member
.ý
x
t° . --ý
X
d) R. H. S. vertical
side member
Fig. (5.5) Distances of torsion axes and warping point from the centroid
axis for various joint configuration, see fig. (5.3)
77
zero
Wareing lines
W'
zero
Warping axis
Fig. (5.6) Warping lines and warping axis of

a channel'section beams
hnimAnr; f c of
Fig. (5.7) Example of a joint element

CHAPTER SIX
THEORETICAL INVESTIGATION OF THE

EFFECT OF LONGITUDINAL LOADS
78
6.1 GENERAL
This project investigates the stress distribution in the side members of a ladder
chassis frame subject to longitudinal load. Hence, in this chapter the bimoment created
due to the longitudinal load and some basic terms and concepts related to bimoment
are discussed briefly.
6.2 LONGITUDINAL LOADS APPLIED OUTSIDE

SECTION PROFILE
If the longitudinal load is not applied on the profile of the cross section, then
the value of the bimoment produced depends on both the load position and the
geometry of the connection of the loading point to the section.
If a force (p) is applied at point (A), parallel to an open section beam away
from the shear centre as shown in figure (6.1). If the force is transmitted to the
section through a rigid arm fixed at the contour (D), and if this force lies in a cross-
sectional plane, the force will cause a bimoment equal to the product of the force
(P) and twice the area (TMDA). This rule may be demonstrated as follows;
-
B =p o), + Mh (from figure 2.7)
B=peh+pzh
Where;
co= principal sectorial area of point (D)

M=pz
o)D =eh
Hence; -
B=p{(e+z)h) (6-1)
79
but;
( (e +. z) h) =2 Area (TMDA) = WWA (6-2)
Where;
co, = principal- sectorial area of point (A) in which the longitudinal load (p) is applied
Therefore; -
B=PU'A (6-3)
From the above equations, it is clear that the value of the bimoment produced
due to the longitudinal loads depends on the position in which the rigid ann carrying
the force is connected to the member cross section.
To demonstrate this effect, two finite element models were created (see figures
6.3/6.4). In both of these, a longitudinal load was applied to a channel section side
member at the same offset position profile via rigid arms, thus applying similar
bending moments and axial load. However, the rigid arms were connected to different
positions on the channel section profile.
The different bimoment effects may be seen in those figures is due to different
area "of (TMDA), i. e different principal sectorial area (CO,

) of the point outside the
cross section profile as shown in equation (6-2).
The principal sectorial area ((os) distribution of a channel section is shown in

figure (6.2), while the relevant constant such as warping constant ( IF ), and torsion
constant (J) for a channel section are quoted from reference (40) as follows; -
80
h2 b3 t (4h + 3b)
tt wSZds = ---------------- (6-4)
6(h + 3b)
2 (b + h) t3
--------------- ---------------- (6-5)
3
6.3 TOTAL STRESSES DUE TO LONGITUDINAL LOAD

The formula for total stresses at any point along the side members of chassis
frame when a longitudinal loads applied can be defined as;-
6T=6A+6Y+6Z+6S (6-6)
Where;
P
ßA =Ä
--------- (direct stress)
M,
6,. =I --------- (moment stress)

rr
(6-7)
MY
6 _ (moment stress)
Iu
B cos
(warping stress)
ßs=r
Where; (y & z) are the coordinates of the point in which the stresses were calculated.
81
connecting
point (A)
oss section
Fig. (6.1) Bimoment created due to longitudinal load
r =-ý 1h
ýýý h
(b-e)h
a) cross section dimesions b) principal sectorial area

of a channel section of a channel section
Fig. (6.2) Principal sectorial area (ob) of the model used

82

83

CHAPTER SEVEN
FINITE ELEMENT ANALYSIS

84
7.1 GENERAL
Modern engineering projects have become extremely complex, costly and subject
to severe reliability and safety constraints. For a proper understanding analysts need
mathematical models that can be used to simulate the design of the projects.
The finite element method has become one of the most popular of these methods.
The method has successfully been applied to the solution of problems in linear and
non-linear regions for one, two and three dimensional domains. It can easily handle
discontinous geometrical shapes as well as material discontinuities.
Finite element analysis is used to predict the effectiveness of design concepts
of automotive chassis frames and body structures, and to evaluate the time and cost
feasibility in early design stage, and to reduce and eliminate the development period
and cost in accordance with the current trend of short life cycle of an automobile.
The finite element system used for this research work is known as I-DEAS
supertab. This commercial software package is presently available on the Cranfield

institute of Technology Computer Centre.
In this chapter a descriptions of the structural and finite element idealisation of

isolated joints of a ladder frame subjected to pure torque are presented. Also presented
is a finite element idealisation of complete laboratory chassis frame under longitudinal
loads. The incorporation by the author of the theory described in the previous
chapters, into a special finite element program which may be used in the preliminary
stages of chassis design will be discussed in the next chapter.
7.2 FINITE ELEMENT SYSTEM

For this project, a static finite element analysis has been performed using the
SDRC "Model solution" program to get the warping displacements and stresses at
the joints and a complete chassis frame. The main characteristic of supertab is to
use a wavefront solution alogrithm and dynamic core allocation technique to minimse
processing time and computer memory storage requirements. It was important, with
85
this project, that the elements were numbered so that the difference in number
between adjacent element is as small as to enable the models produced to be run with
the core storage available. More in depth information and theory of the I-DEAS
system are given in references (42-43).
7.3 STRUCTURAL IDEALISATION OF A JOINT

A ladder chassis frame subjected to torsional or longitudinal loads exhibits an
anti-symmetrical behaviour. This can be used to idealise an isolated joint of ladder

frame and reduce the computing cost and time. In this section, a description of such
an idealisation is given.
7.3.1 CROSS-MEMBER IDEALISATION

The cross member in a ladder chassis frame subjected to torsional or longitudinal
loads may be treated as a beam whose ends are each partially restrained from warping
by the side members. Half the cross member span can be considered since the
warping is symmetrical about its mid-span and reaches its maximum value at mid-
span. Therefore, half the cross member can be analysed, and its boundary conditions
become; -
i) one end of the cross member partially restrained from warping by the side
member, and
ii) the other end is free to warp ( since the bimoment is zero at the mid point of the
cross member ).
7.3.2 SIDE-MEMBER IDEALISATION

It has been shown by Nuttall (20), that the bending in the side members is zero
at mid-span approximately of each bay of a ladder chassis frame. This corresponds

to a point of inflexion which can be modelled as a simple support. Therefore, the side
member in a joint can be considered as a beam whose ends are free to warp, and
which is simply supported.
86
7.4 FINITE ELEMENT IDEALISATION OF THE JOINTS

A finite element analysis of an isolated ladder chassis frame joint is performed
to obtain the stiffness matrix of the joint elements decribed in chapter(5.3.4).
Figures (7.1/7.2) show the finite element idealisations of four joints whose
dimensions were chosen such that they correspond to those being used to build the
finite element chassis model shown in figure (7.15). As far as the joints were
concerned, the following geometric characteristics were introduced.
i) Joint No. (1), has a rectangular section cross member welded to the web of a
channel section side member.
ii) Joint No. (2), has a channel section cross member rivetted (with a plate used to
reinforce the joint), using Huck bolts to the zero warping points in the flanges
of a channel section side member.
iii) Joint No. (3), has a channel section cross member rivetted using Huck bolts at
the zero warping points in the flange of a channel section side member.
iv) Joint No. (4), has a channel section cross member welded to the web of a channel
section side member.
For the typical length of side members, (315-345)mm was chosen in such a way
that it would not cause any undesirable effects to the joint (see figure 5.7). For the
same reason, the side member was supported at the zero sectorial area, i. e zero
warping points, of both flanges with a strong pin jointed bars which were mounted
at the shear centre of the side member. Again the cross member was mounted at its
shear centre by pin jointed bars connected to the zero warping points.
The main concern for the joint element stiffness is the warping displacements,
87
i. e, those along the longitudinal direction of the cross member. Therefore, to

determine the joint element stiffness matrix by finite element calculation, the side
member section passing through the centroid of the cross member section has to be
fixed, as shown in figures (7.3/7.4).
For an accurate joint analysis, the ends of the beam elements and the boundaries
of the joint element should be straight lines after deformation, as they represent the
warped ends of the cross sections, and the shape of these cross sections should be
undeformed.
The main characteristics of thin shell elements in I-DEAS supertab are that the
linear thin shell four node element is formulated using classical thin shell equations
and does not include shear deformation effects through the thickness. The formulation
of parabolic and higher order thin shell elements includes bending and membrane
behaviour as well as the effects of shear deformation.
Usually, for a straight-sided structure with a flat surface and constant thickness
such as a channel section, a linear four node element can be an efficient choice (see
reference 43). Therefore, the finite elements chosen for the joint analyses were thin
shell four node quadrilateral linear elements for all of the members.
Beam elements could be attached normal to thin shell elements because, with
respect to the nodal degrees of freedom, both elements were compatible with each
other. Therefore, beam elements were used for the idealisation of Huck bolts between
side members and cross members for joint(2) and joint(3).
Rod elements were connected to zero warping points of both ends of the side
member and the free end of the cross member. These rod elements were mounted on
the nodes lying on the shear centre line of each member. These elements were used
rather than beam elements so as not to cause any other bending or torsional effects.
To fix the joints, the nodes which are at the shear centre lines of each member
should be restrained corresponding to the three translational degrees of freedom.

88
The application of a tip torque to the free end of the idealised cross member
can be introduced by applying shear forces along the flanges in opposite directions,
i. e, to the nodes which coincided with the zero warping lines; so as not to cause
any other force incorporation. The common value for the pure torque was (300)N x
the height of the cross member section.
7.5 RESULTS FROM JOINT FINITE

ELEMENT ANALYSIS
A finite element analysis was carried out on four typical isolated ladder chassis
frame joints, three having a channel section cross-member, the other a closed section
cross member. The material properties and dimensions of the members and their cross
sections are given in table (7.1).
The deformed shape of the joints are shown in figures (7.5/7.6). The stress
distributions due to partial warping inhibition at the connections along the longitudinal
axes of the cross member and the side member are shown in figures (7.7/7.8/7.9/7.10).
It is clear from these figures that the stress build-up in the cross member towards
the connection is due to the bimoment created according to warping inhibition.
Once all the warping displacements and the transmitted bimoments at the node
boundaries were obtained by finite element analysis, the change in the rate of twist
at the node boundaries of the joint element of each joint can be calculated. The
transfered bimoment can be obtained either from the reaction forces on the restrained
plane in the side member (see figure 7.4), or from the ratio of the partial warping
displacement to the free warping displacement (W, /WF) at the attached end of the side
member. Values obtained from both these methods agreed closely. Hence the joint
element stiffness matrix of each joint can be obtained as will be shown in later.
In the process of this project, as far as the joint element stiffness is concerned,
the most important item was how to get a proper distribution of warping
displacements, and obtaining the transfered bimoments which are reacted at the joint
element boundaries. Values of mid-plane warping displacements round the cross
member section at the partially built-in end (x=0) were determined for each joint.
89
The results are presented in figure (7.11) for the joints having a closed section
cross member. The corresponding values for the joints having channel section cross
members are presented in figures (7.12/7.13/7.14).
As mentioned in the previous chapters, the warping restraint factor (K) is defined
as the ratio of area enclosed by the profile of partial warping displacements to that
of free warping displacements.
Therefore, using equation (2-17), the warping restraint factor (K), can be obtained
as follows; -
WF
WF
Where (We) is partial warping displacement of the cross section,

(W. ) is free warping displacement of the cross section.
It has been shown that the finite element values of warping factors at the
partially built-in
end are in close agreement with the theoretical values and the
measured values have been shown in references (20,22).
Again from the same equation (2-17), by knowing the sectorial area (() of any
point on the cross section and its partial warping displacements, the rate of twist at
one joint element boundary can be calculated as follows; -
WP
_K
wS
The other boundary of the joint element is fixed along the longitudinal axis of
the side member, i. e, the rate of twist at that boundary is zero. The transfered
bimoments can be obtained at the boundary from the reaction forces around the cross
section.
90
Hence, equation (7-2) can be combined with equation (5-11), to obtain the joint
element stiffness coefficient due to the rate of twist stiffness as follows; -
B Wp
Ka (7-3)
--------------------------------------
K co,
The warping restraint factor (K), as well as the rate of twist joint element
stiffness of each joint are also given in table (7.1).
7.6 FINITE ELEMENT IDEALISATION OF A COMPLETE

CHASSIS FRAME
A finite element analysis of a complete ladder chassis frame was performed.
The dimensions of the model chassis frame, shown in figure (7.15), were chosen
such that they correspond to those being used for experimental chassis frame shown
in plate (9.1). Consequently, the same finite element types, and beam elements used
to idealise the joints (discussed in section 7.4), are used in this idealisation.
The model chassis frame was supported with four supports, and appropriate
restraints and boundary conditions were used. These were chosen to match an
experimental validation discussed later. Three load-cases with the longitudinal load
in different positions relative to the shear centre, and hence applying different
bimoment, are considered as shown in figures (7.16/7.17/7.18). The load cases and
loading positions are described in detail in section (9.6) of chapter (9), page (127),
and they are shown in figure (8.4), page (119).
The common value for the longitudinal load applied was (10) KN. The deformed
shapes of all cases, and the longitudinal stress distribution along the side members
of the chassis frame due to the longitudinal loads applied in position (2), i. e, position
where the longitudinal loads produces bimoment are shown in figures (7.19/7.20/7.21).
The results of the complete chassis finite element analysis are discussed and compared
with other results in chapter (9).

91
a) Joint I
= applied force
_3-
-ý = translational
restraint
b) Joint
Fig. (7.1) Finite element idealisation of joints type 1&2

92
a) Joint
--- = applied force
--ý = translational
restraint
ýýý
D
ýöÖ
b) Joint No. 4
Fig. (7.2) Finite element idealisation of joints type 3&4

93
-P; --a ---ti. --
(7.3) Displacements due to bending flexibility

Fig.
Fig. (7.4) Displacements due to rate of twist flexibility

94
a) Joint
b) Join
Fig. (7.5) Joint deformations

95
a) Jo
b) Jo
Fig. (7.6) Joint deformations

96
97
0
ý._ z
G
r.,
4J
r
C
0
n.
N
H
....:......
.
4J
to
v
aý
i co
v
"'i
zaQ Lr) nn to clj n

ö
1n n co
mo
Q,
cm6 M S?
LO 11) D vn co C) rn rn ol
{/} N O N
C14 LO
co 1n h tD M] 1ý Oy 92
fn En
0 En
/aii
98
m
to
`O
1x
<
E co n dNN OL cti
N 1n 01 h t'7) O1 1n
.d n N r-T
r QT [D fV N tÖ
h N
N m to V4 0) co N N
°o to N rn
LO Eli
Q Nr
tý
1 - co rý N J N p> ) ßt1 ýýý
ý% 1III
Z I
NX
w
H fv)
N
f tý t x.
t f,.,........... r ý
t., ý +)
» ..
..................
ýJ
ý r r
j r
(U
ul
U)
4)
4J
U)
0
N
1A 54
N
N
N
Nhh
q
to O
hN n Öf
to Orn `:
cli O
r co Cý
0) C. tr) Iý cc W) N
bJ to ''t
(n 0
N co
C)
N1 to a) 1_ý
N 1OM 11 11
W
99
V.
v
rn
to m 4q
pi 0 to to
NNnn 0 rlý co W) co
a + to O- OON
ri
N ^ 1ý I1 nnn C4 pn W) W)
N0 U') 0
N CO to 0)
hN .41 (A
Ln
P) 1[) 1 co V)
a.
Ol 1A
a I
N
X
a
U)
N
W 0
d'
z
U)
4)
0
.r,
0
.,.,
4,
U
b
a
0
41
d
'0
0
.,.,
.0
...,
1.4
4)
U)
U)
to)
a)
U
U)
O
m
PI)
(D r
co
10
G.ý
N
O
w
N tO
AP- 1
4 -q Ol Go w Ull LO t- w)
4) R?
N Ui
i +h O
V 0 O +
^ co v n
O c0 u) U) PO) O co W
to CNy F) t0 N 1) N t0 N In
9:) F)
h 1, Nh t0
10 ýA 1l to aD m
U) _ _
y
, I)
100
I
LO O
I 'ý 0 14
4J (1)
I
Ef) U
N (D
C> -+- '0

(1)
N
LO
U)
LO 00
UU
0
O
Z
l"
'o
C 0
rn
0
L
O
,o00
4- E
E
N
C U)
Q)
E Ow
Q) o it
C) 0
cTi f W lJ_
r ll l
.
LL: O 0-
Q m
Lo Z
N
ffý i;
mm 0
"r
a
wW
ft OU
Q) W
U- LcL
"r % V)
16
Q V)
O
C- O
CU
U
0
c'1
f i
ri
ý" L
r 0
ýý t
a)
0)
"r
LL-
34-3
i i i O
O O O OOOb a)
U
O tf) O 00010
N `-' 1ÖL
IIN
ww 1N3W33V7dSIG ONIde VM
101
r- w
0 (1)
vN
Hf
- OEn O
co
0 O
z N
4)
c
. r-
0
O
L-
0
0
E
E
C
(1) °w
E J
()
0
L0
CO
r-
0
Q
U) öZ O
"r Co
U
0) W
(/)
C
"r
Q
O 0
Cc$
U
N 0
N
C)
v
"r
IL
.0N
(D 1-8
U) NO `- N 17 a)
U
WW 1N31NlOd-1dSIa ONIdeVM
102
I0
C ý+
Oa,
L) a)
o ')
U,
NO
a-
OU
Cr)
0 0
d-
z
4)
C
"t-
0
r) 0
N
V-
L
O
4- E
OE
4) O
U)
C
Q) W
E J
Q) LL
U
O0 O
r
(Ti 00
Q
Z
"r
0
1D f-
O W
O) (/)
C
"r U)
Q_ U)
O
r-1
r C
N
"r
LL-
v (N N d- cp v
U
ww IN1W33V-1dSla ONIdeJVM
103
o Q)
N
d' N
0) o
C:4 w
ou
Q
d rn
0
0
00
c
"r
O
"r)
C
C- N
0
E
E
(C) ^
C
Q) W
E J
(L) V-
O
C)
(13 ft:
r
d
U)
"r
0
Ü
0)
W
Cl)
C
.r
Q
C- O
0
U
0
N
rf
N-
0
0)
"r v
It_ R
.,.,
ýý Av
aý N
O LO 0 LO O LU O ýý
If) O
N `- `- 000 aý
Cam! U
WW IN1 NIOV1dSIG ONI&NM

104
aö
I?ý
ti
U)
41
1-1
(is
4)
a
x
a)
TI
a
U)
as
w
1A,
rx,
o1,
U
0
w
0<, zs
` a)
ta
0
ýC
a)
h E
}' as
'y 14
44
.Nli
N
K 10
tu
a
In
1-
4
V 03 -.
w
105
Fig. (7.16) Finite element idealisation of model chassis frame

under longitudinal loads, case -1-
106
a)
>= applied force
b)

InIT
= annlied force

108
a)
24.82
18.99
13.17
7.34
1.51
Longitudinal stress distribution

-4.31
Y
X
-10.14
Fig. (7.19) Deformation and stress distribution for case (1), when
the longitudinal loads are applied at position (2)
109
a)
19.05
9.89
2.77
-5.32
-13.46
Longitudinal -21.62
Y
X
-29.78
a
16.28
11.79
7.30
2.81
-1.68
-6.17
Longitudinal distribution
-10.66
111
joint cross member warping restraint change in the stiffness due to
number dimension factor (K) rate of twist rate of twist

(WxHxTHK) using eq.(2-17) (t8. ) (I(. )
mm mm'' N. mm3
1 50x100x3 0.85 1.0864x10" 0.480x109
2 45x168.8x1.6 0.02 4.4485x10'5 0.1205x1012
3 60xl75.2xl. 6 0.10 2.3419x10` 0.209x10"
4 45x90x1.6 0.20 2.7391x10 0.861x10'°
side member dimension (60x18Ox3.2)
v=0.31
E= 205 KN/mm2
Table (7.1) Rate of twist stiffnesses of the joints, obtained by F.E. analysis
CHAPTER EIGHT
THE "A. SAFE PROGRAM"

112
8.1 GENERAL
In this chapter, the developed theory of thin walled beams and joints described
in the previous chapters has been incorporated by the author into special purpose
finite element program which may be used as a design tool in the preliminary stages
of chassis frame design. This program is called the A. SAFE program. The chapter
contains the flow chart of the program for analyzing chassis frame structures by the
direct stiffness method considering bimoment effects.
The stiffnesses of the joint elements for the joints described in chapter seven are
calculated from the finite element output. The bending moments and bimoment
distribution due to longitudinal loads applied to the chassis frame in different positions
were obtained using the A. SAFE program.
Laboratory tests to validate the theoretical stresses obtained from this program
are discussed in chapter (9).
8.2 A. SAFE PROGRAM

Because of the high cost of building and testing prototype truck chassis frames,
and the necessity of shortening the time from design to production, finite element
analysis is used to predict the effectiveness of design concepts. Finite element

calculations may be used early in the process, to provide data for finite element
modelling, right through to analysis of the data from vehicle trials.
Finite element shell element offers detailed analysis of the structures, but
extremely increases the computing cost and time. At the early stage of structure
analysis, it is desirable to have a quick and inexpensive method for the purpose of
cost-effective planning and conceptual design. Simple modelling with beam elements
is frequently used for this purpose.
Nevertheless, reference (39) claims that finite element analysis of chassis frames
should only be used for those areas where normal analytical methods can not be used,
such as at the joints. In particular, beam elements where the cross section remains
113
constant should not be analysed using a finite element method according to this
reference.
A computer program called STRU is presently available at Cranfield Institute
of Technology, which uses conventional six degree of freedom beams to analyse
structures.
A modification has been made to the STRU program to make it capable of
calculating displacements, loads, and stresses at chassis frame members considering
warping restraint effects.
Therefore, a new computer program is written using direct stiffness method. The
new program is called A. SAFE which stands for Al-Hakeem program for Structural
Analysis of chassis Frame considering bimomemt Effects. The program includes the
warping inhibition effects in thin walled beams. The capabilites of A. SAFE program
are;-
i) rigid joints with bimoment, taking the effects of different axes offset.
ii) flexible joints with bimoments, using joint properties obtained from finite element
model.
iii) analysis of thin walled box section beams considering bimoment effects
iv) calculates the stresses at the ends of beam element.
Therefore, in case of the analysis of the chassis frame with flexible joint
assumptions, the joint element stiffness matrix obtained from finite element analysis
can be supplied to the program.
All of the chassis frames to be analyzed by this program are assumed to consist
of straight, prismatic members. The material properties for a given structure are taken
to be constant throughout the structure.

114
Only the effects of loads are considered, and no other influence is taken directly
into account. The program is designed to handle in a single computer run any number
of loading systems for the same structure. Double precision for numerical accuracy
is used in the program.
Although the program is only used for the longitudinal load case in this thesis,
it is generally applicable for other chassis load cases, including torsion, bending etc
...
and combination of these. The program flow chart is shown in figure (8.1), while the
subroutines used are shown in figure (8.2). As mentioned before the program can be
used as a design tool in the preliminary stages of chassis design, figure (8.3) shows
the program incorporation into chassis design programme. Input data to the program
consists of; -
1- Introductory data-title of the job, elastic constants, etc
...
2- Nodal data- number of nodes, bandwidth, nodes sectorial area, nodes coordinates,
constraints and forces. Unconstrained structural nodes have seven degrees of
freedom.
3- Element data- number of elements, element identity number, cross section

dimension. Separate subroutines were written and included for channel and
rectangular sections to evaluate cross sectional area, second moment of area,
torsion constant and warping constant.
Output of the program consists of a listing of all input data as a preliminary
check and the displacements for all seven degrees of freedom at all structural nodes.
Constrained nodes were stated as such in the displacement list to avoid confusion with
zero displacement and the correct list of constraints provided a second check on the
data input. Using the calculated displacements and element stiffnesses, the output also
contains the forces including bimoments and torques etc, at the ends of each element
in the local coordinates.
The stresses at each node at a point on the cross-section with a given sectorial
area is provided in the output list.

115
An example for a simple grillage structure of input and output data decks,
annotated to explain each data item, is included in appendix (A).
8.3 THEORETICAL CHASSIS FRAME MODEL UNDER

LONGITUDINAL LOAD
A theoretical chassis frame model was constructed using the A. SAFE program
with the same dimensions as the finite element model decribed in the previous
chapter. As mentioned before the same load cases and boundary conditions were used
with the longitudinal load in different positions. The common value for the
longitudinal load applied was (5) KN to each side member of the chassis frame, see
figure (8.4). The bimoment input for offset longitudinal loads was calculated in
accordance with the methods of chapter (6).
The bimoments, bending moments and stresses along the side members of the
theoretical chassis frame model were calculated by A. SAFE at sufficient nodes to
account for the gradients of stress to be expected. The computer program A. SAFE
calculated the bimoment and bending moments as well as stresses. The bimoment
results for the three load cases were plotted in figure (8.5), while the corresponding
bending moment results were plotted in figures (8.6/8.7/8.8).
Whilst it is desirable to show the correlation of finite element methods with
experimental results, confidence may also be obtained by comparing experimental

stress results those obtained from the A. SAFE program. Hence, the A. SAFE model
had nodes at equivalent positions to the strain-gauge positions used in the
experimental model, see chapter (9).
The stresses along the side members obtained from the A. SAFE program were
compared with those obtained from finite element analysis and experimental results
as will be shown in chapter (9).

116
PROGRAM A. SAFE,
Icommon dimension
I read and write title
read number of beams, number of nodes,

maximum bandwith and element identity number
read joint stiffness element, or

axes misalignment parameter
initialise variable and arrays
read beam dimensions
call subroutine OSEPR, or

CSEPR for beam properties
write beam properties including sectorial moment

of inertia (1), and joint transformation parameters
transtormauon from local to global axes
calculate stiffness coefficients including added

bimoment coefficient of bimoment and total torque
I
assemble member. stiffness matrices where bimoment is incorporated
call subroutine PROD to transform member

stiffness matrices into global coordinates
assemble structure stiffness matrix
call subroutine solve to modify structure stiffness matrix

allowing for constraints, to solve for displacements and reactions
calculate beam end loads in local axes
calculate beam end stresses
write node displacements, member loads in local axes and beam stresses
STOP
END
Fig. (8.1) Flow chart of A. SAFE program

117
SUBROUTINE PRODI
dimension
matrix product
RETURN
END
SUBROUTINE OSEPRO
dimension
I read open section dimensions
calculation of area, J, I,, y, Ia, rI
RETURN
END
SUBROUTINE CSEPRO
dimension
read closed section dimensions
calculation of area, J, Iw, Ia, rI
RETURN
END
Fig. (8.2) Flow charts of subrotines used in A. SAFE program

118
INCORPORATION INTO CHASSIS DESIGN PROGRAN1NE
START
problem
(stiffness or stresses)
initial design
redesign run A. SAFE, with

(beam or joint parameters) rigid joints assumption
NO required
stiffness&stresses
YES
F. E analysis experiments
(detailed joints) (tested joints)
redesign run A. SAFE, with

(joint connections) flexible joints assumption
NO required
ýýj stiffness&stresses
YES
F. E analysis
(complete chassis) complete chassis tests
redesign NO required
(detail problem) stiffness&stresses
YES
complete truck tests
NO required
stiffness&stresses
YES
Fig. (8.3) Flow chart shows the incorporation of A. SAFE program

into chassis design programme
119
a) a)
10
". -1 N
10
". -1
La CA
Lö
i
i
Ofi
Ofi
ti
O ýy
QQ
O
u.
ASO
oz
a) N
ü]
tö O
U2
0
r-q
w
0
0
4)
0
a
10
r-
N
N
OD
-,
w
120
co
0
O) 0
"r 0
1-
Q)
a) 0
U) 0
00
T-
ü i
ca 0
O 0
c0
r-
--------
------
------
------
------ Ir-
C O
"r
_Ü J.
d- v
V- ry
4-3 W
"f- m
i
0)
r- 1 OW
0
i
ýZ
1
1
1
N
1
r-
1
W
0 OM
Ow
O -r
Q)
ü CD
C Oz
O
O0 J
00 Q
"r
z
0
C-4 fl) Op
"r O
L
OO
UUU
-1,
U) 13 0
i3
i3
"0 0
0
4)
C
()
E
0 0
E 0
N
m
U'N
OD Co CO c0 CD CO 0 Co Co t0 (0 cam' co, (0 co' gyp'
0
OO000O0O0O0OOOO
x--- c- c- == c- x- c- c- c- c-- c- x-
0) IC) d' r`) N `- (D cV r'_) d' U) cD I- 00 m
"r
L1.
(Z**' N) IN3NOW18
121
NT
OD O
O
0
O) cV
"r
(4--- 0
0
Q) 00
a) r-
U)
O
a
O
(0
C
O O
"r 0
Ic- W
"r
m
O
OW
CL O
N
4-3
0
Cis U)
OW
a O=
O
ü Oz
(ö OO
O 00 J
J
Z
1- O0
O
i ýý co
0
z 0
O
a) 0
U) -4-
cd
U
4) 0
0
N
Q)
E
0
10 v o v
u1 a1 vl
0
O 0 O O .1
O .1
O O O O to u tn U)
'O t- ý- a-- r
O O O O
0 C q Q (D Ln D
co LO 0 Ln O
11) O 1C) `; N c ti
0)
. I- (WW*N) 1NIWOW
LL-
122
00 C)
0
0
N
a)
"r
(i- 0
0
Q) co
Q) T-
U) .
11-0
0
--
0
I'
N
C
0 CD
. r-
o
4-) S w
.r m
H1
0 CD w
Q o
- - - - - - - - - - - - N
w
0
N U)
CD
- CD w
o=
Q.
CD
"0 oz
co 00
0 co
J
J- N z
N
C) p
o
0
- - - - - - 7 - - - i - - co 5
Z 0
°-
a) / - o
0
U)
(i3 - - - - -
U
C)
z 0
-
-
C N
____
(1)
E
0
0
in to *0 vo in in In in to In in Un
O Lo OOO C) 0 Ln O LO © Lx) C) LO O
co j ÖNN
cl Lci vi ti4 to
°' (WW*N) IN3WOW

.
L
123
co
0
0
0) 0
N
4-
' 0
Q) - 7 -
0
Co
Q)
CV
_7 O
O
C
0
O
"r
O
"r- w
m
0
/
Q
- -
Ow
- - - - - - - - O
N
W
C
U)
j O
OW
Q O=
ü 0
OZ
7
(0 -
O - - OO
__;r -
J 0J
r
m Oz 0
O
0 '-
ý (! )
. 0
aD
7
O
- - O
d
-- ---
c0
0
O
c 0
a) N
E
0
dll-l le) 0
O 0 0 ö
co to
GO O 0000Ö
L6 ýN
1n o 1n Cl Lo o u o
L6 Nt
0)
Lo
.1 (WW*N) INýIWOW
Li
CHAPTER NINE.
EXPERIMENTAL INVESTIGATION
124
9.1 GENERAL
This chapter describes the experimental approach used to determine the stress
distribution in the side members of an experimental chassis frame under different
cases of longitudinal loads applied to the side members. The experimental chassis
frame model was constructed with the same dimensions and constructional details as
the various cross member to side member joints as the finite element models decribed
in chapter seven.
The purpose of the experimental test lies in the confirmation of the theoretical
results obtained by Vlasov's bimoment theory and finite element analysis.
The discussion of the measured and corresponding finite element values with the
theoretical results of stresses obtained from A. SAFE program will be given in a later
chapter.
9.2 MATERIAL PROPERTIES

Tensile test samples of the mild steel was cut from the same sheets which were
used to make the members of the experimental chassis frame model. They were strain
gauged by two mutually perpendicular resistance strain gauges placed on each face
of the test piece to measure longitudinal and lateral strains.
A tensile test was performed and the stress and longitudinal and lateral strains
were recorded for each load increment. The average strains in the longitudinal and
lateral directions were determined and the stress-strain curves are shown in figure
(9.1). The mean values of poisson's ratio (v)
and the modulus of elasticity (Young's
modulus) were found to be (0.31) and (205 KN/mm2) respectively.
9.3 CONSTRUCTION OF THE TEST MODEL

The test chassis was fabricated of mild steel to be representative of typical truck
chassis construction. The general nature of the chassis model is shown in figure
125
(7.15). Channel section side members (size 60x180 mm) were connected by several
different types of cross member and cross member to side member joint.
For the same method of a fabrication as actual chassis frames, cold working
rivets were replaced by Huck bolts. The flanges of the channel open sections of the
cross members for joints No. 2 and No. 3 were attached to those of the side members
by using Huck bolts.
For joints No. 1 and No. 4, the cross member was MEG welded to the side
member web by mild steel electrode. A fillet weld having equal leg length dimensions
of (6)mm, was used for welding all round the cross member section. A (6)mm fillet
weld was used as it is thought that this is a typical size used for welding of members
in ladder chassis frame construction. Care was taken to ensure that the weld size was
uniform around the channel section and consistent in both joints. Care also was taken
to keep the distortion of the side member web caused by the welding process to a
minimum.
The box section cross member consisted of standard cold-formed (ERW)
structural steel section while all of the channel sections used for the project were
formed from mild steel sheet by a bending machine with a proper radii of bending
and consequently showed good forming condition at the free edge of the bent part.
The length of the side members was restricted by the maximum length of the
available steel sheet as (2000)mm. All of the holes for the mounting of the cross
members were designed to be in the line of zero warping along the longitudinal
directions of the side member flanges.
9.4 STRAIN MEASUREMENT

Axial constraint direct strains were measured on the surfaces
of the side members
of the chassis frame by single electrical resistance strain gauges; the gauges were
aligned parallel to the longitudinal axis of the side members. The gauges were of
U. S.A., Micro-Measurment Division manufacture, with (5)mm gauge: 120 ohm
resistance, gauge factor = 2.09 .
126
The strains were recorded using strain recording unit. Central dummy gauges
were used to compensate for surrounding temperature variations during the tests and
to complete the full bridge circuit. For each gauge position, the recording unit
registered a stable reading.
The gauges had to be attached to the surface of the side member sections after
the cross members were welded to avoid damage. The strain gauges were attached to
the inside surface of the side members sections to avoid damage during handling and
testing.
The gauges were positioned such that each gauge placed at a point on one
channel section side member had a corresponding gauge placed symmetrically on the
other side member. This was done to check for anti-symmetry in the tension and
compression readings. Figure (9.2) shows the positions of the gauges around the
channel section for both side members. A gauge position is labelled with an upper
case letter and the corresponding anti-symmetrical position labelled using the same
lower case letter. The measured strains were converted to stress by multiplying by the
pre-determined elastic modulus of the side member material.
9.5 SUPPORT AND LOADING CONDITIONS

Plate (9.1), shows the general arrangement of supports and the loading conditions
for the chassis frame under longitudinal loads. Four brackets were attached to the side
members of the chassis frame. The chassis frame was horizontally positioned from
these brackets by four rigid beam supports. Rose joints were used at both ends of
each rigid beam support to idealise pin joints. The system of rigid beams and pin
joints was arranged so that it did not contribute to the stiffness of the chassis frame
and could easily and accurately be idealised for inclusion in the computer program.
Constraints were introduced, in such a way that these would not distort the
measured stresses by influencing internal forces in the chassis frame. The chassis
frame was subjected to longitudinal loads. These loads were applied by hydraulic
jacks connected by strain-gauged rods to the support brackets on the side
members
of the chassis frame.
127
9.6 TESTING PROCEDURE

The horizontal positioning of the chassis frame on the four rigid beam supports
with rose joints at the brackets was confirmed using a bubble-in-glass inclinometer.
Initially, in each test, zero strain readings were taken under the self-weight of the
frame and the support and loading apparatus. The chassis frame was subjected to
longitudinal loads by hydraulic jacks connected to the support brackets on the side
members as shown in plate (9.1).
In order to investigate the bimoment effects on the stress distribution in the side
members of the chassis frame, different longitudinal load-cases with the longitudinal
loads in different positions relative to the shear centre, and hence applying different
bimoment, were considered. Three load-cases were used as follows;
-
i) Load-case (No. 1). The longitudinal load (p) was applied to the side members at
joint (No. 4), where the cross member is a channel section. The test was carried
out with the load (p) applied in two positions relative to the shear centre.
When the load (p) was applied in position (1), the load would only introduce direct
load, and moment about the z-axis. Neither moment about the y-axis nor bimoment
would be introduced when the load was applied in this position as shown in plate
(9.2a). However, when the load (p) was applied in position (2), direct load, moments
about z-axis and y-axis as well as bimoment would be introduced to the side members
as shown in plate (9.2b).
ii) Load-case (No. 2). The longitudinal load (p) was applied to the side member
between joint (No. 3) and joint (No. 4), where there is no cross member attachment
as shown in plate (9.3). For this case, the test was carried out with both loading
positions as used in case (No. 1). Hence moments and bimoment would be
introduced to the side members.
iii) Load-case (No. 3). The longitudinal
load (p) was applied to the side
member at
joint (No. 1), where the cross member is a box section
as shown in plate (9.4).
128
The test was also carried out with both loading positions used in the previous
two cases, hence same moments and bimoment would be introduced to the side
members due to longitudinal loads.
In all tests with the above load- cases, strain readings were obtained at each load
increment of (1) KN at low loads when loading and unloading. The direction of the
applied load was then reversed and the measurements repeated. The test results
showed that the tension and compression readings were linear and consistent with the
load applied in opposite directions. This confirmed the linear elastic behaviour of the
structure at low loads. The load was increased until (10) KN, i. e, (5) KN for each
side member.
The stress distribution due to longitudinal loads for each station, i. e, the flange
and web of the side members, see figure (8.2), were recorded. For each load-case
when the longitudinal load was applied at position (1), i. e, when no bimoment is
created due to longitudinal load, the stress distributions were plotted in figures
(9.3/9.5/9.7). The stress distribution when the longitudinal load is applied at position
(2), i. e, when it produces a bimoment, were plotted in figures (9.4/9.6/9.8) for each
load-case.
The total axial stress around the cross section of the side members obtained from
A. SAFE program with those experimental values are plotted in figures (9.9/9.10/9.11)
at selected cross sections, see figure (9.2).
The comparison between the theoretical stresses obtained from finite element
analysis and A. SAFE program with those experimental values has been establised for
the side members of a complete chassis frame. Different load-cases with the
longitudinal load in different positions (position 1 and 2) relative to shear centre, and
hence applying different bimoment, were considered. The discrepancy between
theoretical and measured values of the stress increases towards the bracket
attachments. Nevertheless, even in the worst case (case No. 1), the discrepancy
between A. SAFE program and the measured values is only (19%) at the loading
brackets. For the region between the brackets and the joints there is a good
129
agreement. The reasons for the discrepancy between the theoretical and experimental
approach can be regarded as due to; -
i) Manufacturing inaccuracies in the positioning of bolts, heated structure because
of welding process,...etc.
ii) The friction due to the high tightening forces of the bolts which may have
partially inhibited the warping at the position of the loading brackets attachments.
iii) The stiffness coefficients of the joint element used in A. SAFE program are
approximate, because of the assumptions used and the way of obtaining them (i. e
from F.E analysis which itself has been built on simplifying assumptions).
iv) only joint flexibility due to bimoment is considered in A. SAFE program, while
flexibilities due to other type of loads, joints regarded to behave in a rigid
manner.
v) approximations used such as ignoring the deformation of the profile of the
section, lozenging and the effects of the curvature shape at the web-flange
connection of the channel section. Those approximations are used to simplify the
mathematical formulae developed to calculate the stiffness sub-matrix due to
warping effects.
From figure (8.5), it is clear that the external bimoment applied to the side
member due to the longitudinal load (case 2) is shared nearly equally between the two
parts of the side member. Therefore, the maximum bimoment near the loading bracket
of the side member for this case is lower than the other two cases, that gives a
maximum bimoment stress distribution near the loading bracket of the side member
for this case lower than the other two cases.
It is also noted that the stresses predicted by the A. SAFE program which uses
Vlasov's bimoment theory for beam analysis and finite element analysis for joints are
closer to the measured values than those obtained by the finite element analysis of
the whole chassis frame. This supports Beermann's claim that finite element analysis
should only be used for those areas where normal analytical methods can not be used,
i. e, joints. In particular,
beam elements where the cross section remains constant,
should not be analysed using the finite element method.
130
a
an
0
o'
0
My
o Q'
0
ö
C-
a)
ö
4J
(n
C> (,
C)
a,
c -.+
"ý cl)
C
U)
o 'A
C-
v a,
C) E
C>
In
1.,
0
441
AS
4J
(D
14
U)
rn
.w
0
0
1
131
.. a 1o
ýO0
0 -4 ON
-4 ai I
tn 1r4
rý 4J -H
(D W (CS
", I a;
zs tr to
i
w
i
0
.. -4
Oo
M
10
ýo
Q 0
04
v a)
rý .ý,
J Si
ýO 4J
Oll
t;
-4
fam.
u1
132
A) NODES AT THE FLANGE, STATION A (SEE FIG. 9.2)

40
35
30
25
: 20 F-- TOTAL STRESS
15
0
cn
lo-
Li
e
V) 5
-5
-10
0 200 400 600 800 1000 1200 1400 1600 1800 2000
POSITION ALONG THE SIDE MEMBER (MM)
" EXP.
o F. E.
- PROG.
B) NODES AT THE WEB, STATION B (SEE FIG. 9.2)
9.0
7.5
0
6.0
---u- 7777w
4.5 \olb
3.0
1.5
w
U) 0.0
TOTALSTRESS
-1.5
-3.0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Fig. (9.3) Stress distribution for load-case No. 1, the

when
load (p) is applied at position (1)
133

50-
45
40
35
30
TorAL SMM
25 Ma
---- SMESS'No T
20-
15-
lo-
5 f
------ --- --------
0
-5
-10
-15
-20
0 200 400 600 800 1000 1200 1400 1600 1800 2000
s EXP.
o F. E.
--- PROG.
10
C
cV
0
-1C
I TOTAL STRESS
--- STRESSND OMOMENT
-2C
"
-zý
-30-j-
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Fig. (9.4) stress distribution for load-case No. 1, when the

134

3(
2:
2(
5
0
li 0
9 e
5 ý
-1. TOTAL STRESS

-
-21
-2 5
-3 0
0 200 400 600 800 1000
1400 1600 1200 1800 2000
POSITIONALONG THE SIDE MEMBER(MM)
o EXP.
o F. E.
- PROG.

6
5 A 0
4
N
\
1-1
U) 0
(I) \ \
W
i.I
F
- TOTAL STRESS
0 200 400 600 800 1400

1000 1200
1600 1800 2000
Fig. (9.5) Stress distribution for load-case No. 2, when the

135
) NODES AT THE FLANGE, STATION A (SEE FIG. 9.2)
0/
5
0
5
ö " 0 0
5 TOTAL STRESS
-
--- STRESS NO MOMENT
O
0 200 400 600 800 1000 1200 1400 1600 1800 2000
e EXP.
o F. E.
PROG.
10
C
o
-5
Cl)
ui
'--
-1G
- TOTAL STRESS
---- STRESSM BMOMENT
-15
-2C
0 200 400 600 800 1000 1200 1400 1600 1800 2000

136
NODES AT THE FLANGE, STATION A (SEE FIG. 9.2)

30
25
20
- TOTAL STRESS
15
10
in 0
U,
5
cr
U,
0
0
-rl
-ic 0 200 400 600 800 1000 1200 1400 1600 1800 2000
e EXP.
o F. E.
-PROG.
8) NODES AT THE WEB, STATION B (SEE FIG. 9.2)
7
6
00
5
3 I
U)
En
W20
1
TOTAL STRESS
-1
0 200 400 600 800 1000 1200 1400 1600 1800 2000

137
A NODES AT THE FLANGE, STATION A (SEE FIG. 9.2)

U
"
5
O
0
TOTAL STRESS
Q STRESS.NO EIMOMENT
-----
rJ
f5
.0 600 800 1000 1200 1400 1600 1800

0 200 400 2000
" EXP.
o F. E.
- PROG.
E3) NODES AT THE WEB, STATION B (SEE FIG. 9.2)

10
0 ` -ý----
--------o
cC
0
L
1
vý 0
- TOTAL STRESS
--- STRESSNO 69AOMFNT
-1
_2c
0 200 400 600 800 1000 1200 1400 1600 1800 2000
(9.8) Stress distribution for load-case No. 3, when the

Fig.
138
0.004 5.135 1 3.031
_0_o08
1
all values in (N/mm2)

" Experimental
---- Theoretical
0.008
-. -_i -0.004
7.511 7.145
i) section -A- ii) section -C-
tý®
21.769
2.087
-2.872 -10.462
1
l
1
1
1
i
I
8.684 9.686 ý1 13.611
20.931
1
`ý
L--
iii) iv) section F-

section -D- -
Fig. (9.9) Total stresses around selected cross sections of the side member,
see fig. (9.2), for loading-case No. 1, when the load (P) is applied
at position (2).
139
i ý;
3.464 3.128
-0.001 0.000 P
I
I
t
1
all values in (N/mm2)

s Experimental
---- Theoretical
0.000 ýa
0.001
i 7.427
7.562
i) Section -A- ii) Section -C-
0.220 9.058
-0.783 tQ el
i
-28.496 i1
0/
iz
I,
Jill
Ip \ýýýý
ai -10.497
ý i- "0.059
10.325 10.682
t
w
iv) Section -F-

iii) Section -D-
at position (2).
140
f.
0.151 8.731 i
7 -5.929
-0.302
all values in (N/m
" Experimental 1\
----Theoretical
0.302 1"
-0.151 ___.
14.096 13.724
i) Section -A- ii) Section -C-
4.962 iý3.650
2.055
-3.683
1
1
1
1
6.732 6.748
9.528 i9.710
t
e
iv) Section -F-

iii) Section -D-
at position (2).
141
ii
1/
1ý1
r
ýaýý
"ý
ý.
4J
L
142
a) position -i-
b) position -1-
Plate (9.2) Loading-case No. 1

143
i i '.
"F,
!t,
rä
A4
a) position -1-
,,r ý.w
b) position -2-

144
w'i 4'ý dnýký.

wýh
u
iý"n.Wrrtk+ý c; ýä.'ý 5rýý. ý 'ý,ýaVrMýný
r
.,,.nt, ý!.; rS;ý'ý*x '; ý,-a.
r::
"1, ý
a) position -1-
ý, ý' 7_ ýsý ýaýýýý`.

'öß'11
b) position -2-

CHAPTER TEN
CONCLUSIONS
145
10.1 GENERAL
Two load cases for commercial vehicle chassis frames are;-
a) Torsion due to anti-symmetric vertical loading of the chassis.

b) The longitudinal load due to braking or acceleration, acting through the spring
hanger bracket.
These loadings are relatively independent of each other, and in practical cases
may vary considerably in magnitude under different dynamic conditions. The chassis
dimensions and material would vary with different designs.
This project dealt with the problems of these loadings. The analysis of the stress
distribution in open section side members of commercial vehicle chassis frames under
the effects of the previously little studied longitudinal loads which may act on truck
chassis has been investigated.
In this chapter the correlation between theoretical and experimental approach is

discussed. Suggestions for the optimum design of a spring hanger bracket are
discussed from the point of view of longitudinal loading.
10.2 SUMMARY
10.2.1 THEORETICAL INVESTIGATION

The theoretical work undertaken throughout this research is to develop adequate
formulae for the analysis of stresses due to longitudinal loads in commercial vehicle
chassis frames. In particular, the following theoretical studies were conducted: -
a) Thin walled beam theory introduced by Vlasov was refined and rederived for
appropriate loading and boundary conditions.

b) Longitudinal stresses were derived accordingly.
c) The governing differential equations of thin walled beams of open and closed
146
section were also derived and each term of flexural and torsional loading was
identified.
d) A thin walled beam finite element stiffness matrix was introduced for open and
closed sections incorporating bimoments, and detailed discussion was then given.
e) The equilibrium matrix of rigid joints with the effect of different axes offset was
introduced for different cross section profile and orientation of the beams meeting
at the joint.
f) Joints were solved by a finite element program I-DEAS and by the theory adopted
in this thesis using flexibile joint assumptions.
g) A complete finite element model of a chassis frame under longitudinal loads was
solved for different load positions.
h) A new computer program which incorporates all the above developed theory of
warping inhibition effects in thin walled structures such as chassis frames has
been written and tested for a complete chassis frame under longitudinal loads.
i) The bimoment introduced on the side member of the chassis frame due to
longitudinal loading for different bracket connections and different loading
positions was studied experimentally and the results were used to validate the
theoretical and finite element results.
10.2.2 EXPERIMENTAL INVESTIGATION

The experimental investigation provided several interesting results. It was the
experimental evidence reference which first revealed the presence, magnitude and
extent of warping stresses in the side members due to the longitudinal loading. This,
together with results from the finite element analysis, lead to the conclusion that the
side member web allows the distortion in the cross members ends because of the out
of plane flexibility of the side member web.
To further assess the effects of loading position along the side member in the
experiment, a series of tests were undertaken on the chassis frame model with the
longitudinal loads applied at different locations along the side member. For each load-
case two different load positions were investigated.
147
When the longitudinal load (p) was applied in position (1), it was expected that
only direct load and moment about the vertical axis would be introduced into the
chassis frame side members. But, when the load was applied at position (2), a
bimoment as well as direct load and moment about both vertical and lateral axis
were expected to apply to the chassis frame side members.
10.3 DISCUSSION
10.3.1 THEORETICAL ANALYSIS

The ladder type normally used for chassis frames of commercial vehicles appear
to be simple structures. They are frequently analysed by simple methods. However,
accurate analysis of the stress distributions, are actually quite difficult. Rough
estimates based on simplified assumptions are justified, when dimensions of the main
members are required in early stages of design.
Therefore, beam elements have been developed which include the effect of
constrained warping as shown in chapter (3). These elements are principally

formulated for thin walled open section beams.
Closed section thin walled beam cross members are only used occasionally in
chassis frame designs, but a simplified method of analysis can be used ignoring the
deformation of the profile of the section as shown in chapter (4).
The matrix displacement method can be used to analyse a chassis frame with
full continuity of bimoment at the joints, i. e, rigid joint assumptions. In this case the
cross member stiffness matrix has to be transformed using the equilibrium matrix as
shown in chapter (5), in order to maintain compatibility -at the joints.
The equilibrium matrix written out in detail in equation (5-10) can be used for
one end of the beam element comprising a cross member when only half the frame
is analysed because of symmetry. If the symmetry condition can not be used, each
148
cross member has two equilibrium matrices on the main diagonal. The signs of the
coefficients in the equilibrium matrices are found from figure (5.5). The beam
have been represented by their centroid axes, and the lengths by the distance
elements
between the intersections of these axes.
In the case of flexible joints, the displacement transfer across the joint has to
be interpreted with respect to the joint flexibility. A joint element has been introduced
and the stiffness matrix of the joint element was obtained from detailed finite element
analysis of joints.
10.3.2 FINITE ELEMENT ANALYSIS

A finite element method approach was considered in chapter (7), in its favour
this method is extremely versatile and can be used to analyse virtually any linear
elastic structure under any loading conditions.
From the point of view of a simple model of joints or chassis frames under
investigation the method was quite easy to use. The method also had the added
advantage that stresses were calculated throughout the whole chassis frame members.
The disadvantages of the finite element method are many but are generally
outweighed by its versatility. The main disadvantage is the cost, the finite element
method can only practically be used on a large digital computer. Data preparation
can also be tedious and time consuming and although this has improved over recent
years it can still take a considerable time to be prepared.
The method also requires a high degree of understanding on the part of the
engineer, the choice of element types, the application of load, choice of restraints,
mesh design and element density are but a few of the factors which need to be
considered for any particular analysis.
Stress discontinuities across element boundaries are a problem. Particular care was
exercised across folds in the shell structures analysed.

149
10.3.3 A. SAFE PROGRAM

The simplified finite element program developed in this research (see chapter 8)
employed Vlasov's theory, the resulting linear differential equations were solved using
the matrix displacement method. This method was easy to program and could be used
on a microcomputer without difficulty. The program also executes very quickly and
is therefore cheap. As indicated before the program can be used as a design tool in
the preliminary stages of chassis design.
10.3.4 GENERAL
The question of whether open or closed section cross members are most efficient
is not one that can be easily answered. It is obvious that fewer or smaller closed
section cross members are required to achieve a given frame torsional stiffness with
acceptable stress distribution in chassis frame members. This is really a question of
structural optimisation, with which the present analysis is not concerned. The methods
and the program presented however, provide analytical tools which may be used as
part of an optimisation technique.
10.3.5 THE SHAPE OF THE GRAPHS

It is seen in chapter (8), that the shape of the graphs in figure (8.5) for the
bimoment distribution at points along the side members due to longitudinal loads
acting on the side members of the chassis frame test models are straight lines while
the theory suggests curves of hyperbolic functions (sinhpL & CoshpL). However,
because of the chosen cross-sectional dimensions (see table 7.1) the value of the
dimension constant (p) becomes very small (p = 2.708x10` mm'), while the range of
(pL) is (0 - 0.542). In that range the shape of the hyperbolic sine and cosine is nearly
a straight line. So the shape of the bimoment distribution graph at the points along
the beam is approximately a straight line. The graphs would-be curves if different
cross-sectional dimensions are chosen.
Again the graphs of the stress distribution at the points along the side members
(station A and station B) are straight lines instead of curves and that difference is for
the same reason which is discussed above for the bimoment graphs.
150
10.4 PRACTICAL SUGGESTIONS

In commercial vehicles, spring hanger brackets are generally attached to the side
member chassis frame. These components can apply longitudinal loads on the side
members of the chassis frame.
Since it is proved in this project that quite large bimoment stresses as well as
bending and direct stresses would be produced due to the longitudinal loading, it is
very important to choose an optimum design for these components on the side
members of a chassis frame from the point of view of longitudinal loading. Bimoment
stresses depend on ;-
i) Creation of bimoment due to application of longitudinal load at points where (co.,)

is not zero as shown in figure (6.1).
ii) For the same position of application of longitudinal load, different bimoment
effects are caused by different connections to the section profile as shown in

figures (6.3/6.4).
Therefore, careful design of the attachment of the spring hanger brackets is
recomended in this thesis, which gives a minimum value of principal sectorial area
to the points where the longitudinal load is introduced.
Another factor which affects the warping behaviour and has been investigated
in this thesis is the position where the longitudinal load is applied, i. e, position (1)
or (2), (see figures 7.16,7.17, and 7.18). This factor is important because it has a
significant effect on the bimoment stresses as well as the bending stresses due to the
bending moment about the lateral axis of the chassis frame (i. e. axis 'Y', see figure
5.4). The optimum place for applying longitudinal load is position (2), because the
side member would be subjected only to stresses due to direct load and moment about
the vertical axis of the chassis frame, while the other components of stress due to
bimoment and moment about the lateral axis of the chassis frame are zero.
Some typical spring hanger brackets are shown in figure (10.1). An alternative
spring hanger bracket which has been designed to minimize the bimoment created due
151
longitudinal loads taking into consideration the bolt effects mentioned above, is shown
in figure (10.2).
10.5 SUGGESTIONS FOR FURTHER RESEARCH

Further research analysis is required into ;-
1) Vlasov's thin walled beam theory can be used to derive a stiffness matrix for
different types of closed section cross members (other than square or rectangular
sections which have been investigated in this thesis).
2) The determination of the optimum torsional stiffness of a ladder frame taking

into account the stress concentrations at the joints.
3) The program presented in this thesis would need only slight modification to
analyse tapering members or combinations of tapering and constant section
members.
4) Calculation of the stress distribution for a combined set of torsional and
longitudinal loads. The patterns of the stress distribution for a range of ratios
of torsional and longitudinal loads and the changes in the stress distribution with
changes in the ratio could be calculated. This is likely to be a worse casethan
when torsional and longitudinal loads are applied separately.
5) Calculation of the stress distribution in chassis members for different chassis
designs, i. e, different range of ratios of flange width to web width with thickness.
6) Measurement of the magnitude of the longitudinal loads transmitted to the side
member of actual truck chassis during dynamic tests on the road.

7) Measurement of the stresses developed in the side member of an actual truck
chassis due to longitudinal (braking) loads during dynamic tests on the road.
8) Comparison of the stress distribution in chassis members obtained from theoretical,
laboratory experiments, dynamic tests and finite element models for different
designs.
152
ý
a) b)
(4
ýý ý0
.
.0 ýo
:T
Fig. (10.1) A variety of typical spring hanger brackets
S.
C
ý; axis
Zeto/ý.
Fig. (10.2) Alternative design and attachment of

a spring hanger bracket
REFERENCES
153
1. VLASOV, V. Z.
"Thin-walled elastic beams"
English translation, National Science foundation,
Washington, D. C., London, oldbourne press, 1961.
2. ZBIROHOWSKI-KOSCIA, K.
"Thin- walled beams from theory to practice"
Crosby Lockwood & Son Ltd., London, 1967.
3. HANKE, M.
"Thin-walled beams in Automobile Engineering"
Part 1:'Theory of simple frames', Auto. (Prague), NO. 1,1959.
Part 2: 'Theory of plane joints', Auto. (Prague), No. 8,1959.
Part 3: 'Torsion of ladder frame', Auto. (Prague), No. 4,1960.
4. ZAKS, M. N.
"Stress state in joints of a twisted Automobile frame"
NANII Proceedings, No. 61,1963 (English translation of Russian original)
5. ZAKS, M. N.
"Calculation of the distortion of Motor-Vehicle chassis frame, taking into
consideration the rigidity of the joints"

Automobile Industry, No. 4,1964 (English translation of Russian original)
6. TIMOSHENKO, S. and WOINOWSKY-KRIEGER, S.
"Theory of plates and shells"

Edition 2, McGraw-Hill Book Co. Inc., 1959.
7. KOBRIN, M. N., KILIMNIK, L. SH. and TITOV, A. A.

"Investigation of the stress state and durability of vehicle frame side member walls
at the point of load transfer"

Avtom. Prom No. II, 1969 (English translation of Russian original)
154
8. SETTLER, W.
"Fundamental reappraisal of the chassis frame"
Automotive Design Engineering, Vol. 2, Data sheet 14,89, October, 1963.
9. COOKE, C.J.
"Strain Energy theory applied to the chassis frame"
Automotive Design Engineering, Vol. 2, Data sheet 14,89, October, 1963.
10. TIDBURY, G. H., MARSHALL, P.H. and ROACH, A. H.

"Torsional stiffness of commercial vehicle chassis frames"
The XII FISITA Congress, Barcelona, 1968.
11. TIDBURY, G. H.
"The torsional stiffness of an open section thin walled beam in terms of bimoment
and the generalisation of partial warping inhibition boundary conditions"

Motor & Vech. Conf. Belgrade, 1974.
12. ROACH, A. H.
"Warping inhibition in commercial vehicle frames"
M. Sc thesis, Cranfield Institute of Technology, 1966.
13. ERZ, K.
"Uber die durch unebenheiten der fahrban hervorgerufene verdrenung von
strassenfahr zeugen"
A. T. Z., No. 4 April, No. 6 June, No. 11 November, No. 12 December, 1957.
14. AWUDU, G.
"Warping inhibition in the joints of vehicle chassis frames"
M. Sc thesis, Cranfield Institute of Technology, 1968.
15. MEGSON, T. H. G. and ALADE, G. A.

"Structural analysis of ladder frames under torsion"
Proc. Instn. Mech. Engrs., Auto. Div., Vol. 190,1976.
155
16. MEGSON, T. H. G., ERGATOUDIS, J. and NUTTALL, J.

"Partially restrained warping of open and closed section thin walled beams"
Proc. International Conf. thin walled structures, Univ. of Strathclyde, April 1979.
17. MEGSON, T. H. G.
"Linear analysis of thin walled elastic structures"
Surrey University Press, 1974.
"Extension of the Wagner torsion bending theory to allow for general systems of
loading"
The Aeronautical Quarterly, Vol. XXVI, August 1975.
19. ALADE, G.A.

"Structural analysis of ladder frames under torsion"
Ph.D thesis, University of Leeds, 1974.
20. NUTTALL, J.
"The torsional analysis of ladder frames"
21. DATOO, M. H.
"Stress concentrations in the joints of a ladder frame subjected to torsion"
22. ALVI, M. S.I.

"The stress distribution in the joints of vehicle chassis frames subjected to torsion"
Ph.D thesis, Cranfield Institute of Technology, February 1978.
23. TIDBURY, G. H. and ALVI, M. S.I.

"The theoretical and photoelastic investigation of stresses in chassis frame joints"
Proc. Instn. Mech. Engrs., Vol. 198D, No. 3,1984
156
24. SHARMAN, P.W.

"Optimum stiffness weight design of peripheral and ladder frames"
Proc. Auto. Div. Instn. Mech. Engrs., Vol. 182, pt. 2A, No. 3,1967-68
25. LASEVICH, L. G., SKOLNIKOV, M. B. and PODLEGAEVA, T. D.

"Selection of optimum sections of commercial vehicles chassis frames"
Avtom. Prom., No. 2,1975.
26. SHARMAN, P.W.

"Torsion of chassis frames"
Automotive Engr., 1, Oct. -Nov., 1976.
27. SHARMAN, P.W.

"Some aspects of the structural analysis and design of commercial vehicles"
Ph.D thesis, Loughborough University, 1974.
28. SHARMAN, P.W.

"The effect of joint flexibility on the torsion of a vehicle body"
Proc. Instn. Mech. Engrs., Vol. 196,1982.
29. SHARMAN, P.W.

"Analysis of structures with thin walled open sections"
Int. J. Mech. Sci., Vol. 27, No. 10, pp665-677,1985.
30. TIDBURY, G.H.

"Matrix force analysis applied to chassis frame stiffness"
Automotive Design Engineering, Vol. 2, Data sheet 25,81, September 1964.
31. TIDBURY, G.H.

"Stress analysis of vehicle structures"
Advanced school of Automobile Engineering, Cranfield, Note 1,1965.
157
32. MARSHALL, P.H.

"Torsional stiffness of ladder frames"
Advanced school of Automobile Engineering, Cranfield, M. Sc. thesis, 1965.
33. ROBINSON, J.
"Automatic selection of redundancies in the matrix force method"
A. I. A. A/C. A. S.I Joint meeting, Ottawa, Canada, October 1964.
34. ALI, R., HEDGES, J.L. and MILLS, B.

"The application of finite element Techniques to the analysis of an Automobile
structure"
Proc. Instn. Mech. Engrs., Vol. 185,1970-71.
35. HEDGES, J.L., NORVILLE, C. C. and GURDOGAN, O.

"Stress analysis of an Automobile chassis frame"
Proc. Instn. Mech. Engrs., Vol. 185,1970-71.
36. TRIMAN, R. S.
"Incorporation of bimoment in matrix structural analysis"
M. Sc. thesis, Cranfield Institute of Technology, 1982.
37. LEE, W. G.
"Bimoment effects in a chassis frame considering joint flexibility"
38. BEERMANN, H. J.
"Joint deformations and stresses of commercial vehicle frame under torsion"
I. Mech. E. Conference paper, C178/84,1984.
39. BEERMANN, H. J. (ED. TIDBURY, G.H. )

"The analysis of commercial vehicle structures"
Mechanical Engineering Publications Limited, London, 1989.
158
40. AL-HAKEEM, A. H.
"The effect of torsional and longitudinal loads on the stress distribution in open
section commercial vehicle chassis members"

41. ARGYIS, J.H.

Appl. Mech. Rev. 11,7,331,1958.
42. SDRC
"I-DEAS level 4.0 Supertab-Engineering analysis Pre- and Post- Processing user's
Guide"
Structural Dynamics Research Corporation, Milford, Ohio, 1988.
43. SDRC
"I-DEAS level 4.0 Supertab-Engineering analysis Model Solution and Optimization
user's Guide"
Structural Dynamics Research Corporation, Milford, Ohio, 1988.
44. PERRY, C. C. and LISSNER, H. R.

"The strain Gauge primer"
McGraw-Hill Book Co., New York, 1962.
45. WILLIAM WEAVER, JR. and JAMES M. GERE

"Matrix analysis of framed structures"
2nd edition, 1980, D. Van Nostrand company, New York.
BIBLIOGRAPHY
159
1. MEGSON, T. H. G.
"Aircraft structures for engineering students"
Edward Arnold, 1972.
2. AL-SHEIKH, A. M. S.
"Behaviour of thin walled structures under combined load"
Ph.D thesis, Loughborough University, 1985.
3. BEERMANN, H. J.
"Warping torsion in commercial vehicle frames taking into consideration flexible
joints"
Int. Journal of vehicle design, Vo1.1, No. 5, pp397-414,1980.
4. ROARK & YOUNG

"Formulas for stress and strain"
Fifth edition, McGraw-Hill Book Company, 1975.
5. ROMANOV, F.
"Investigation of stress concentration in thin walled elements of chassis frames"
6. RUSINSKI, E.
"Torsional stiffness of chassis frames with point-welded nodes"
7. TIDBURY, G. H.
"Vehicle structural analysis-a survey"
IntJoumal of vehicle design, Vol. 1, No. 2, ppl65-172,1980.
8. CARVER, G. C.
"Truck chassis frame considerations in equipment mounting"
S.A. E., No. 760291,1976.
160
9. BEERMANN, H. J.
"Static analysis of commercial vehicle frames: A hybrid-finite element and
analytical method"
Int. Journal of vehicle design, Vol. 5, Nos. 1&2, pp26-52,1984.
10. MEGSON, T. H. G., ERGATOUDIS, J. and NUTTALL, J.
"Analysis of closed section thin walled beams subjected to partially restrained
warping"
Proceeding of the fifth International specialty Conference on cold- formed
steel structures, St. Louis, University of Missouri-Rolla, 1980.
11. ROMANOW, F., PALUCH, Z.

"Experimental studies on the durability of the elements of vehicle frames under
complex state of loads"

FISITA Congress, Vienna, May, 1984.
12. RAO, B. V. A., RAMAMURTI, V. and GANESAN, N.

"Torsion of prismatic shells by finite difference approach"
Journal of Aeronautical Society of India, Vol. 25,1973.
13. TAKAHASHI, K.
"A torsional strength analysis of truck frames using open section members"
S.A. E. paper, No. 710595,1971.
14. BAIGENT, A. H. and HANCOCK, GJ.

"Structural analysis of assemblages of thin walled members"
Journal of Eng. struc., Vol. 4, pp207-216, July, 1982.
15. BARSOUM, R. S. and GALLAGHER, R.H.

"Finite element analysis of torsional and torsinal-flexural stability problems"
Int. Jour. for Num. Meth. in Eng., Vol. 2, pp335-352,1970.
161
16. MIZUGUCHI, KISHIMOTO.

"Experimental analysis of torsional strength of ladder type truck frame"
MHI Technical Bulletin, Vo1.6, No. 5,1959.
17. NAKAMURA
"Siffness of ladder type frame"
Nissan Diesel Technical Bulletin, No. 47,1985.
18. OKUMOTO, YAMANE, HARADA

"Strength of ladder type truck frame"
ISUZU Technical Bulletin, No. 79,1988.
19. OHNUMA, ET. AL.

"Stiffness of truck frame and its prediction method"
JSAE paper, No.882161,1988.
20. KAZUO AO., SATORA OHNUMA, TOSHIHARU MATSUI &KUNIO HARA

"Basic study on truck frame tosional stiffness"
MITSIBISHI Motors Technical review, No. 2,1990.
21. MOUNIR M. KAMAL, and JOSEPH A. WOLF, JR.

"Modern Automotive structural analysis"
Van Nostrand Reinhold Co., New York, N. Y, 1982.
22. GURNEY, T.R.

"Fatigue of welded structures"
Cambridge University Press, 1968.
"Analysis of semi-trailer chassis subjected to torsion"
162
24. TIMOSHENKO, S. and GOODIER, J.N.

"Theory of Elasticity"
McGraw-Hill Co., 1951.
25. BRESLER, B., LIN, T. Y. and SCALZI, J.B.

"Design of steel structures"
John Wiley and Sons, New York, 1968
26. ODEN, J.T.

"Mechanics of elastic structures"
McGraw-Hill Book Co., New York, 1967.
27. DALLY, J.W. and RILEY, W. F.

"Experimental stress analysis"
McGraw-Hill Book Co., 1965.
"Increasing semi-trailer torsional stiffness could enhance operational safety"
Automotive Design Engineering, May, 1970.
29. TIDBURY, G.H.

"Design problems of truck bodies"
Automotive Design Engineering, October, 1962.
30. BOLLAND, G.B.

"Flexibility and stiffness matrices for an open-tube warping constraint finite
element"
Department of mechanical Engineering, Lanchester Polytechnic, Sep.,1971.
31. REILLY, R.J.

"Stiffness analysis of Grids including warping"
Journal of the structural Division, ASCE, Vol. 98, July, 1972.
APPENDIX (A)
Al
Example
J1 J2 J3
3 7 9 11
1
z
R
K node in
xy plane
4 8 10 12
2
J1 J2 J3
Fig. (A. 1) A simple Grillage structure.
El. Node Node Element type Joint Joint type

No. (i) (j) No.
El 1 3 side member J1 flexibile joint
E2 2 4 side member J2 rigid joint

with axes offset
E3 3 7 side member
J3 rigid joint without
E4 4 8 side member axes offset
E5 7 9 side member Bandwidth=[ (Nj - Ni) + 11 *7
E6 8 10 side member El. Identity No.; -

1= side or cross member
E7 9 11 side member without axes offset
2= cross member with
E8 10 12 side member axes offset
3= joint element
E9 3 5 joint element
Sec. Identity No.; -
E10 4 6 joint element 1= any section with
given properties
Ell 5 6 cross member 2= channel section
dimensions
E12 7 8 cross member 3= box section
dimensions
E13 9 10 cross member
Table (A. 1) Elements used for the Table (A. 2) Joints used for the
Grillage structure grillage structure
A2
Input data I Decription

----------------------------------------------------
'
SIMPLE GRILLAGE STRUCTURE Title
2.05E+05 0.31 Young's modulus, Poission ratio
13 12 35 No. of El., No. of Node, Bandwidth
1 side member, E1. Identity No. =1

13 i-Node, j-Node
000 300 0 00 200 0 coord. of i-Node, j-Node, k-Node
11.37 -90 1800 11.37 -90 1800 (y, z, c) of i&j Nodes for stress calcul.
2 channel section, Sec. Identity No. =2
60 180 3.2 section dimensions
1
24
0 400 0 300 40 000 200 0
11.37 90 -1800 11.37 90 -1800
2
60 180 3.2
1
37
300 00 700 0 00 200 0
11.37 -90 1800 11.37 -90 1800
2
60 180 3.2
1
48 1
300 400 0 700 4 00 0 0 200 0
11.37 90 -1800 11.37 90 -1800
2
60 180 3.2
1
79 I
700 00 1200 0 00 200 0
11.37 -90 1800 11.37 -90 1800 I
2
60 180 3.2
1
8 10
700 400 0 1200 400 0 0 200 0 I
11.37 90 -1800 11.37 90 -1800
2
60 180 3.2
1
9 11
1200 00 1500 00 0 200 0
11.37 -90 1800 11.37 -90 1800
2 I
60 180 3.2
1
10 12
.
1200 400 0 1500 400 00 200 0 I
11.37 90 -1800 11.37 90 -1800
2
60 180 3.2
3 joint element, El. Identity No. =3

35 i -Node, j-Node
300 00 300 1 00 200 0 I* coord. of i-Node, j-Node, k-Node
0.85E+10 I* joint stiffness coef. from F. E. analy.
3
46 I
300 400 0 300 399 0 0 200 0
0.85E+10 I
A3
1
56
300 10 300 399 00 200 0
40 50 1600 40 50 1600
1 * any section type, Sec. Identity No. =1
250 0.25E+03 0.3E+06 0.5E+05 0.8E+08 * Area, J, In7, I,., r
2 * cross m. with axes offset, E1. I. No. =2
78
700 00 700 400 00 200 0
35 45 1500 35 45 1500
20 20 15 15 -5 -5 25 -25 -8 8 * offset i&j
parameters of Nodes
2
45 90 1.6
1
9 10
1200 00 1200'400 00 200 0
25 50 1250 25 50 1250
3 * box section, Sec. Identity No. =3
100 3 50 3 * box section dimensions
1 1 1 1 1 1 1 * Rest-, 7 deg. fre. for

of each Node
1 1 1 1 1 1 1 1= free
0 0 0 1 1 1 1 0= fixed
0 1 0 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
1 1 0 1 1 1 1
1 1 0 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
0000000 * Loading
0000000
0000000
0000000
0000000
0000000
0000000
0000000
5000 000 0.25E6
-0.5E6 -0.1E8
5000 000 -0.5E6 -0.25E6 0.1E8
0000000
0000000
A4
H
z
4 ooaooooomco a%
H .4 .+ .+ . -4 .x .+- .4o00
V3 +++++++++++
Zwwwwwwwwwww
O mmmmmmaomoýnr
V-vvvvevvO. 4O
a. O. mm of OA m 0% 0 0% rv
V' Ln ul Ln ui IA ßf1 u1 .n Ou . -1
z %O %0 b to %a b %D \O OrC
M %0 %0 býO%01O to %0 Oa0N
00000000000
ýO %O bVbV %o %o Ln -n %o
No00oOOOOOOO
+++++++++++
I. WWwwWwWwwWw
z rn o. rn o. rnm rnrnot00
wCCC '0' QvPCOON
OQOOOOOOOO (A
Q co co co co mwN co ON to
'y co m co co co co w cD Obv
.
0% C1 m C% 0101 C% 0% Orr
zNNNNNNNNV11f1M
NOOOOaOOoooo
rrrrrrrruýor
>. 00000000000
+++++++++++
Hwwwwwwwwwww
z cm 0% OI O1 C% O, a. ON ONN
W vi to vi to ui Ln an in oi o%
:4 rv NNNN rv NNO In "
0 to %O%o%Ob %o %o 0001. -4
i. a. (1 01 O 43% cI (1 01 O co c4
.
N00000000000
C V. Cv VW CCMMb
'J.. 00000000000
O .......... .
UWWWWWWWWWWW
wNNNNNNNN U) V. (.
.7M ý+f rý1 tý1 f'1 Iý1 Al fýl NNN
OOOaOOOOOOO
NN fV NNNNO co O
ul ill U) Ill IA IA Ul IA Ow O
C.
] (h ma(71 ONC% 0Ný
1t1I1 00000000000
t 0000000C. 0 C. 0
N000000000OO
$. wC N
oo". +
w I. 1
to ma NN
E -. 4 "O 41 ý' 000 C) 0000000
1a m-. + u1 00000000000
01 -4 9Z %4 2Z N OOOOOOOOOfOO
W0a0w 4 OOOO 471 O C.
1. UCUW1 c. . V' .M q' V'
as 1
V
1a E'+ EC1 b Ln OOOOOOOOOOC.
1md4m OO 00000000000
$4 LA + :F
U) W
x VC. 0 N C. 0000000000
a a+ E o X aaaooo00000
xu I1". ß f+] M f+1 rrN NUf Mor N
1ý -to A to ON
O "i co
.1w to
E NMr 00000000000
N V 00000000000
F
U a ., OOOOOOOOOCO
E, uuu N
11111 O
v)
N
W E
(7 a 'C N OOOO0000000
as
ä 00000000000
ca a ...........
a ON E W
y S RO .- 0oo roo
M W N O O
C. OO
4D Q
aaaQ
H NOZ E.
E. . f+ a
14 P
ä
0 U)P
=HW w 00000000000
Oo0000000C. 0
0
0 00 N a4 00000000000
H a -+
x OOOOOOOOO
N P.
f'1 NI rr rv rv 1+1 rN
a Q
a a .a _4 .4
E.
a a V
Q a
O
H
a' H
V1 h ý+f .r a0 Oý O .y .1 ýO a0 0
t*. E
N
14 x
a hl U) N( Vr m O) O Ln r 0)
E a
P.
Z
O . -1 N (N mot' .0 IO rm ti N (N
a
w
A5
OOOO0000 .0w C> C> ko W%0 %0 NN IS'i In IA IA IA IA

OOOOooOOOo00 0000000000o0
tttttttttttt 111111111111
WWWWWWWWWWWW WWWW.. W ...
OOOOOOOOOOOO '.O 'o fV IN 01 ON ývm 0% 10 %0
OOOOOOOC. OOOO vc co co rr In In .r .1N IN
OOOOOOOOOOOO -4 'o 'D " 'n wig) mm ascm
C. OOOOOOOOOOO ýC to ýD w IA Ul OO In M . -4 11
OOOOOOOO. -4 .4OO
OOOOOOOOOOOO
000000000000
111
ý4 -4 -4 -# -1 -4 -1 ý4 -4 -4 -4 ý 4 -4 44
00000000%a %0 C, <> v. rcw -rcvwmmIn f

CD OOOOOOOOOOO 000000000000
1111111
++++++++++++
ww ca WWWWwwwWW
wwwwwwwwwwww
t, r, Iý IN «9 cm Co Co f+l M f"1 (9
0000000a000a
OooCD CC V' Ve OO V' " 0% 0% Ol 0%
o0000000
01 O\ 01 01 Co a0 M 0% C. C. 0% o.
000000000000 NN NN 1N tN Nm
U) o c> en OOOO
Oo00000oN 1N
NN NN NN (N M f+11r1M
00000000 N fV OO
00000000CD 000 00000000000 O
IIIIII
ý4 ý4 ý4 14 H ý4 -4 1-4 -4 -4 .4 -4
000000000CD 00CD
NOOOOOOOOOOOOO
ý;: OOOOOOOOOOOOO '0 10 OO
0000000O Vr V' CVrVVVW q' H1 IA IA M
N OOO Cl OOOOOOOO OOOOOOOOOOOO
OOOOOOOOOOO co cl ++++}+}+++++ 11 III1II1
WWWWWWWWWWWW WWWWWWWWWWWW
00 0000000000 Choi m cl M1 Me0 IV aT V -W -w
OOOOC. OOC. OOOOO

000000000000
OOOOOOOC. OOOO
.4 . -4 .-
NNNNNN . -4 .-$ 400
01 01 to \O
acc
to SQ
NO 000000000000 000000000000 IA IA f. '1 In IMM IA t010 to to
Hooo oo00000000 Oo oooo00IAInoo mmmmmmMMý. ý I. a
y
00000000000 I[1 O C. OOOOOOOOOOO OOOOOOOOOOOC.
N 1I 1111
14 14 . -/ . -1 rl . -1 .i 11 11 14 . -I .1 .t .i . -1 . -I .4 . -1 .1 r/ . -r 11 -I . -1
00C. 00000ooo00
NOOOOOOOC. OOOOO
$ooooo00000000
00 0 oo 0 no
000 0 o I 000000000000 vrc. rcraww -ra. NN
0 00000000000 000000000000
++++++++++++ 1IIiIII1
WWWWWWWWWWWW WWWWWWWWWWWW
0000000000000 000000000000 NNSO oOCt0 orrenIn
NOOOOOOOOOOOOO 00000000000o at at rrv7Of 01 NNNN
FOOO OO O OO O OOO O OO OOO OO OOOOO IA I V1 tf1 OO %D to tO %O O%O1
a"
000000000000 . '+rrrrSO%ooONN
4D OOOOOOO000 IA0 000000000000
OOOOOOOOOOOO OOOOOOOOOOOO
t
oOOOC. O Co OOoOOO
NO000O0C. 00000 co
.. 4 N .. { . -4 .y 14 -4 . -l .. d .. 1 -4 .y . -1 . -4 -4 1-4 .. l -4 .$1 .4 .y 14 "4
1A OOOOOOOOOOOOO
0 00
000 0 000000
N
00 00000 0000 O .4H OO -w 'r . -4 . -1004 .l
OOOoOOOOOOOC. OOOONNOOC. OOO
OOOOOO0000000 + ý. }. I I++ 111 1++
.ý} .F . 1. } ý. ++}
."ý000OOOOOOOOOO
-10000000000000 wWWWWWWWWW1W WWWWWWW[. ] W [d (a7 w
. OOOOOOOOOOOO to %D c> CI r V'N N OONN
y 0 00 0 00 O O O OOOOC. OOOO 'D %o OON LA .irlOO f" 1 f" 1
0 0 Cl 0o co o C. 00 000000000 01 01 00 1f1 orro0 01 01
000000000000 V' a' OO IA . -1 to to OOTm
OOOOOOOOOOOO NNOOMrNNOO7 V'
OOOOOOOOOOOOO OOOOOOOOOOOO OOOC. OOOOOOOO

HOOOOOOOOOOOOO IIII
Ho000oOooOo000
y
. -1 . -1 OO 11 11 .ý . -1 oO If .d .i . -1 Co O r/ FHNoO rl .l
000000000 co 0 to co
N
N
W OOOOOOOOOOOOO OOOOOOOOOOOO NNOM to MNN V' M. -i. If
. -1 OOOOOOOO co OOOO OOOOOOOOOOOO 000000000000
äo000000000000 }+++++++++++ I I+ IIIIItIII
CC fl WwWww 14 wwwWW IA wwwwwWWwww 67 W
OOOO co OOOoý c>
z oo 0000000000 N O Ul O . -I 000 -4 Nr IA Ol
000000000000 F -T NO in co IA MN. -l IN CO
0000000 00000 4 CO N0 .4 . -1. -1 CO Nm 0050
a OOOOOOOOOOOOO
00C. 0000C. OOOO 14 In rO to In 5o r . -l co . -1 MM
000000000000 0 r0 1nco'a2 I Nrnrn
ýooooo00000000
.Fo00000000000o
N OOOOOOOOOOOo V OOOOOOOOOOOO
0 a"""" z ä
1-4 C. ooIn I1IItI1
0 000 00 0o o
E H oN . -1 . -1 .i "-1 . -1 "4ý . -1 O . -1 "-/ -4 i -4 . -4 . -1 . -4 . -1
04
a
a
~
C4 000oOo000000o F Q
0 .ao00oOC. 0000o c) o N
NoOOOOOoOOOOOO W 00000000avOo 0' 0000 NN .
"J.'" W' 000000000
m OOOOOOOOOOOOO
000 O C. OOONNaOO c OoOoo
c> O o
z N A
}++++}++}}}} W +}++11I1II11
a Z
wwwwWwwwWwWW
OO ID OOoOo0oOo
WWWWWWWwWwWW
a
E A. ' OO0000000000
IA
Z
oaOO
OOOOM
co a V' 'V' C V' ýP V'
. -1 CO 00 50 5O 5D W
OO0000000000 0 OOOO tO V' IIm f"1 IA Pl
E CO OO0000000000 H OOOO In IA
h r at a ul ýO b co a G ei OOmm
z f"'1 V' co OOOOOCD OO inLn C> 4D F Oa 0 N
H e-1 .i "i rl 4 ............ orr. 4. -4NNN
... .. "......
0 0 OOOOOOOOOOOO a OOOOOOOOOOOO
I") H N t"1 V' r co 01 O f'1 d' Mr 01 a
rl o 1
-4
IA rl . -1 .y .4 .r . -1 . -1 .0H4
f'1 C ul to r co 01 O ri N MI
.3HN
w .a .4 'a .4
A6
ombIt 0. ßc 0000000 00eo000 C. OOOU1NN O co OOUlNN

O VBW o, O V'N 0000000 0000000 W Co OOmwm aO OO 0% co m
O IA OOc Vf 0000000
.i 0000000 . -1 OOOI-O N ei000r ON
O a0 Ul an O V" 01 0000000 0000000 Ln 000 co N 1- Ln aOOWN(,

OM IA OObo c0 to M co c0 to m co
aminroao IN aT N 1N aT N
IA I Hf c, 'd M 1(1 14 .i ". Ul ý4 -4
NOm 1, W 1I W
I in ý4 ko Pi a
a 4
a,
N
W U)
W W
V U U
U V
a
0 a C4 a
k. 0 0 0 0
w w k" w
Z r- OOOO CDOr OOOOUlNN OOOOLnNN
O O VV 1(1 0% N 0OOOOOOM N
/f1 . -4 000000 vi ý(1 0000 Tco Oh O x00000.001
. -4 O . -1 Ill c> N ti op ý-1 00 000 00 "4 OOOOOOO
. -10001-0 N "-1 OOOt- ON
ca O10t!;; Nr 00 01 0000000 O OOOOOOO m 1(1 OOO a0 N f- tl' UOOO I,
o f'1 IA ONO C1 <O N
IA co b t+l eo CO %o r"1 co
oC. Ifß NN 001 O O N -w c4 N VrN
u1 I cnN Nm+1 v p.,
NN fn O . -1
a NN Ri M a m W
W IN W 1
00 P7
w W
m W W
0000000% 00CD CD 0001 OO%D IA N O1 . -1 00w IA N 0% . -1 0 m%0 U1 0

0000000 0000000 O co 1fl 01 NmN O CO tn O% NmN
r1 T
O0 000 O V" U1 p1 O V" N
o fi 000000 Nf O. ti NNNN 'D 0. -11! ) NNN 'O O. -$ U100 7 12
000000 Co 000000 e0 0 IA tn aO N O1 1N 0 IA IA Co N 01 IN 010 tom 0
IA d" 0%
^A O aO NbN «1 o1 O co "A 1D N 1+1 Cl OM IA OO
N 10 O
N O U1 NN %0 If1 O IA NNb
N tn Ori IA N000
N IA 1 V1 NNN IA IA N IV N "" 1A 11MO
W "-1 W W 1N r1 l
"-1 «1 . -1 W NM4 W fV O W1 IN
i ä
c4 f4 1N P NIN W'
4 tn . -1 %D
< 1
ýt 1m
W W 1 1
W y y
U u W W
U U
a a a u
w 0 0 a
O
ö
r, , w p,
-- 0000000 0000000 OO'0 In ON. OO1D AON.

f"1 000 0000 0000000 -1 -1 O f"1 'C N NOM
N O a0 tn OICD I0O
OOOo0oO ap O Co an O1O CoO Cl O IACtNPl
oOOOooO o "-{
. -4 12 N OO tO O . -1 Ifl 1N oo 1O O u1 ONim
r1
-4 0000000 N 0000000 Pl OUl Ul co (D NI Q. OLn Nap O N1f1 1N OID IA(A N fN co
O °D PN OM /A to N1 1N OM IA ONO %n
c> 1 ul f4 " I! 1
v O1 IAN C 1(1 OmU)N N O0%
v `ý IA N . -1 N1 ". " IA I Hfl NM - 1A P'1 NNM
1
1N 111 1
a . -. NNMO
W W a 1 WINN
W
W
w r
W W x'
IA
(4
$ x
A7
IF
CD, Om rn UoON ti m
OONNP V' OO "0 1fl m
V100 ýO t0 Hfl iA OOM 0M
OCD COm000CD r%D 0%

NN11
MMrraQC. OOmN
N1-4 -4 -4 -4 0% (A OOM %0%0
10 OO0. O\ OOOOMMM
O C. OOOOOC. NN 01
11 I .11
. $. -ý. -4
000o Ch moa 10 N .4
. -1 C. C. a0 a0 OOO CD Cp .M
NOOoOMMOOý.
. -1 M
0ov0C. 00 co
.4
I) mMW co .i e-4 O CD 10 r %0
". +-4 . +m ml Ifl0O N. -lN
WAooc, 4 NaV. Oooro
OOONNOOOOO Ifl O
z11 .4
11
OONNrrOO 1O MN
NOOC ao WOO Vw U1
EOONN Oý Oý O C. . -1 1O O1
0O . -1 . -4 V -T O co .iNr
0o rn Q. eeo oý mr
oovmMnoor0N
Eoo u2 NNoo m12 o
00Pa0o00o
uý rv.
-11
OONNNNOC. Of 01
.i
XOONNNNOO V' aD a0
0. OOMM.. MOOM. ! ".
00 to to U1 U10 Q0 r10
OO to tO l+1 (toO &A , -1 CO

x'100 rr bý00 OHO Ill op
NOO in in PPOO f"'1 tO P'1
00 in in 00 w00N to 01
1
OONNPP in 1f1 Ot co N
NOO to bPPPP in to \O
NOO in in o7 M
II1I1b,
a0 Hf N1 P9 ("1
OOtn1i! 10000NN at
Ww
O0
14
OO to to PPOO %D N 11 NN
.inroorN.
OO in NI 01
4. yoomPý
01 OOP
R -A '0q ai
, -1 P9 awMU
w>Nm
ooinin iioo. -4Pw
1 W0 Id Ow
1 k. S. RUW
oo ný a4
Ln . -4 n1n*o rto
I CO CO PPN
tn 49j:
.iOO ri .4 .4N E -f
W O OOMI in in aro
G .OO
N........
maao1
. m i. u1
0 OOOONNOOO 1f1 O
O co, OO fl r4 N MOOOOQ,
fn oooLnCh z
x=C. 0
a. I1 a ar U
aD 00001.001 NOO0 inert to V. 000 at in in
.4OOOrON f1 OOOPMM
,4OOOO in P 1xUu
11 1". 4A 111
OOP WOO to 1+1 N
U1000 co Nr .iO 06 rr at t0000IA U10 NO O in IA 000
o OP 1tP
to to ic0 N P%a co in OM 0 r; 43 OO OP P OO. to at
41 m
NPN in M .i ir 01 01 N
-I N
i .4 -4 W 0Po I in to r OOOO
ý
.4 . -1 OO . -1 Nr UI11I1
a c>
i
14 N "ý II11
I
N N N >. O OOO PP O Or Oin

W f7 C
a a N...o00oPo0o .. rv
0 0 .
I1
4, w
z OONNNNOO ti Ot Ot
OOOONNN ý+ Mf OOO ýo OP - in OOO ul rn Oý X0ONNNNOOP
"wOOOm CO ON 00 rv OOO o0 w
MI O1 %D O POGO Ot Ifl in N aOO 1`Y 1' 1! "I OO li 1ý
rl o0oro rv P'1 00oPM 1+1 .4 , -1 OOO O1 l P w N,,,,,,,,
0oO N
in c> OOwNr r .HOOOrr at at to
oo in in nvl poo. tio
in 1fl O N
a0 tD in c0 NP to /II
aD n1 O in o [a
NPN P in in "-1 in r 0% Ct a'
in .4- ... I Oar o .., iwr N
rv' 1o
a
N1 N ýi mPr Op 01 O N 10 a7 O
I in a, . -4
D7 O a
'F' w M a0 at O in r Ol
.4NiPr
W z 0) .4
" .1N !nc in '0 r co .aN in
1.7

Al-Hakeem Thesis 1991

Uploaded by

Copyright:

Available Formats

Al-Hakeem Thesis 1991

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Al-Hakeem Thesis 1991

Uploaded by

Copyright:

Available Formats

CRANFIELD INSTITUTE OF TECHNOLOGY

Academic Year 1990-1

ALI HASHIM AL-HAKEEM

In the name of God, most Gracious Most Merciful

"Of Knowledge it is only a little that is

communicated to you, (0 men! )"

my son Hashim, my daughter Haneen

in other words "great gifts are from great men".

- Mr. JC Brown, my supervisor whose excellent tuition, valuable advice, continuous

assisted me, I am much obliged.

shoulders, may I ask for the user's indulgence.

- "perfection is exclusively a divine attribute'.

produces high longitudinal stresses.

This is an analysis of the stress distribution in the side members of commercial

complete chassis frame models are validated against experiments performed on a

2. WARPING THEORY OF THIN-WALLED

3. STIFFNESS MATRIX OF THIN-WALLED

4. THIN-WALLED CLOSED-SECTION BEAMS

4.2 WARPLESS CLOSED SECTIONS 52

S. CHASSIS FRAME JOINTS

6. THEORETICAL INVESTIGATION OF THE

7. FINITE ELEMENT ANALYSIS

8. THE "A. SAFE PROGRAM"

1.1 Typical commercial vehicle with a ladder chassis frame 16

thickness under torsion 31

channel-section beam element 49

section of a box beam 63

channel cross member 73

5.3 Displacements of nodal points at the ends of beams meeting at

a joint, used to build up the equilibrium matrix [H] 74

axis for various joint configuration, see fig. (5.3) 76

and experimental tests 104

8.1 Flow chart of A. SAFE program 116

chassis design programme 118

member, see fig. (9.2), for loading-case No. 1, when the

9.10 Total stresses around selected cross sections of the side

member, see fig. (9.2), for loading-case No. 2, when the

member, see fig. (9.2), for loading-case No. 3, when the

2.1 Comparision of the sectorial properties 36

closed section beams 64

and closed section beams 64

9.1 General arrangement of chassis test rig 141

b cross section width

s cross section profile

0. angle of twist of an open section member

cpx angle of twist of a closed section member

In members subject to torsion or longitudinal loads, warping effects are more

or I-sections. An I-section is very efficient in providing bending stiffness but

continental counter-parts which employ closed section cross members or extensive

requirement, static loads can not be disregarded and should be analysed. It is

1.2 REVIEW OF PREVIOUS WORK

complete warping inhibition or free warping at the joint. An introduction to Vlasov's

work is given by Zbirohowski-koscia (2).

restraint by measuring stresses and rotations on members of individual joints. He used

perfect kinematic coupling between members meeting in the joint.

Tidbury/Marshall/Roach (10-12) dealt with the problem of determining

reported by Awudu (14).

Megson/Alade/Nuttall (15-20) dealt with the theoretical determination of warping

results by experimentally measured values.