The Structural Design of Bus Bodies: Paper
The Structural Design of Bus Bodies: Paper
Paper 9
The problems of designing bus and coach bodies are discussed. The matching of the body structure and
chassis frame causes problems which have been largely overcome on an ad-hoc basis. Simple theoretical
methods are suggested for estimating the stiffness in bending and torsion of the superstructure of composite
buses. The formulae are checked against computer analyses and plastic models when possible. The proportion
of load carried by the bodywork is estimated to find the stresses in critical members for any combined loading.
Although further analysis and full-scale testing are required, preliminary design calculations for composite
buses can be based on this method.
INTRODUCTION separate chassis and for this reason the attempt has been
THESTRUCTURAL design of buses has always received more made to incorporate some of the simple concepts used by
attention in technical publications on the Continent than Continental designers for integral buses in the analysis of
in Great Britain and it is of interest to speculate about the composite vehicles. These concepts are checked against
reasons for this. computer analysis and model visualization, where pos-
First, the number of people using buses as their only sible.
means of transport is higher in Eastern Europe than in the
West and this would account for the importance of bus S T R E S S ANALYSIS
design in those countries. Secondly, for several decades Methods proposed
British roads have enjoyed a higher standard of surfacing The most sophisticated analysis of bus structures seen by
than those of most other countries. Bad surfaces cause the author was published by Brzoska ( ~ in ) tPolish in 1955.
high torsional loading on vehicle structures and torsional This work is mainly concerned with integral buses and
loads, as will be shown, are carried more by the bodywork uses the best techniques of aircraft fuselage analysis
than the chassis. Therefore, unless torsional loads are available at that time. Brzoska analyses structures of the
small they must be carefully analysed by the designer of forms shown in Figs 9.1, 9.2 and 9.3.
chassis-type vehicles. For the type shown in Fig. 9.1, it is assumed that the
When bodies are strong enough to resist high torsional structure above the waist rail is so flexible, relative to the
loading, they will generally also carry the bending loads underbody which is reinforced by transverse tubes, that
and this explains the popularity of the integral vehicle no load is taken by the roof. Fig. 9.2 illustrates a fully
where roads are bad. It is notable that the ‘Olympic’ bus, integral structure where only the end bulkheads are stiff in
produced in Britain specifically for operation in rough their own plane and where the torsion loads are applied
road territories, is an integral vehicle. in that plane. The third type analysed by Brzoska is an
Another solution is to build a flexible body on a separate integral bus in which all the bulkheads are flexible in their
chassis and in some countries wooden bodies mounted on own plane (Fig. 9.3).
rugged chassis have a satisfactory life. All the structures analysed are open-ended tubes with
Because of the trend toward the integral bus on the no door openings. This has meant that Brzoska’s paper
Continent, several methods for stress analysis have been has not been widely used in spite of the fact that his ana-
proposed for it. It is probable that a substantial part of the lysis is more detailed than those of other writers. (He
bus industry in Britain will continue to mount bodies on includes the effects of the non-linear stress-distributions
in shell structures.)
The M S . of this paper was received at the Institution on 16th June Hungarian workers, notably Michelberger (2), have
1972 and accepted for publication on 26th June 1972. designed successful buses utilizing an underfloor grill
* Senior Lecturer, Advanced School of Automobile Engineering,
Cranjield Institute of Technology, Cranjield, Bedford. t References are given in Appendix 9.3.
---
,- -. I
1
I
,
. -.
/*
i
I n bending, the door-openings and window-pillars are (b) when all joints are rigid and the cant-rail beams
the critical areas and approximate formulae are given in are assumed to be completely rigid.
Fig. 9.7 for the maximum bending moment in the window-
In case (a) a bending moment arm equal to the total
pillars due to bending, and in the cant-rail and in door
height of the pillar is obtained while in case (b) the maxi-
sills as a result of the shear force. In Fig. 9.8 the method of
mum arm is only half that height. Erz suggested that an
analysis for torsion is shown. It will be noted that the
empirical factor (c) of two-thirds be used for computing
structure is treated as a thin-wall tube with its axis across
this bending moment. On the other hand the bending
the vehicle. The transfer of shear from the cant-rail to the
moment at the base of a portal with equal stiffness in all
waist-rail again puts the window-pillars in bending.
members and legs carrying a similar load in shear ( d )
Schemes for the complete analysis of the various types
suggests a factor of 4/7,approximately midway between
of bus structure discussed so far are given by Pawlowski
the alternative suggestions (Fig. 9.10).
(6) and new ways of incorporating torsion boxes are also
included (Fig. 9.9). When dealing with the problem of
It must not be assumed from the references quoted that
no stress analyses have been carried out in Britain bur in
transfering shear from the cant-rail to the waist-rail by
most cases these have not been published. Further, the
bending in the window-pillars, Pawlowski suggests the
references quoted have all been to simplified analyses
use of the two extreme values of bending moment :
without computers (although the floor-grillage idealiza-
(a)where there is a pin joint at the cant-rail or the tion of (z), (3) and (4) lends itself to a standard computer
cant-rail is flexible in bending, and solution).
Approx:
bending
moment
diagram
Q =&,D p=-!d
pr Err h
QU -- / " Q
Coach stress analysis - Erz +IL
loads due t o bending Qul, Loads due
Mu,=~'0.75 to
bending
Fig. 9.8. Torsion analysis assumes t h e coach t o be a Fig. 9.9. Pawlowski's open bus structure incorporates
thin-walled tube w i t h a transverse axis torsion boxes i n a new way
-:?;"
U
4
q = shear f l o w
a t cantrail
,---. --.
!
,a''
! M
1 moment
Waistrail
hI
z
iA
For an all-steel body the ratio of stiffnesses will be in the
d Portal frame order of 4.2 to 1, for an aluminium body 1.4 to 1 and for a
Equal stiffness wood-framed body it may be as low as 0.000 64 to 1,
members assuming the cladding only carries shear and the bending
4 stiffness depends only on the wooden members. These
ht M= Qh,
simple comparisons depend on two assumptions : first,
(a) With a pin joint or flexible cant-rail; (6) cant-rail and joints rigid; that there are no door-openings or wheel-arch cut-outs
(c) an empirical factor; ( d ) the portal frame assumption gives a over the length of the vehicle considered (normally
factor half-way between the others.
between the axles); and secondly that the floor cross-
Fig. 9.10. Bending moments in window-pillars members and the attached pillars and roof-bows form
bulkheads that are infinitely stiff in shear.
The first assumption is normally true for touring
Other workers, notably Alfredson (7), have published coaches but the overhang loads form a separate and
complete computer analyses of integral buses using finite important loading case for the side-frame over the rear
element methods. An early attempt at a similar analysis of a wheel-arch where its depth is reduced. The second as-
fictitious bus structure was carried out at the A.S.A.E. for sumption is not strictly true, more of the bending load
a student thesis (8). being carried by the chassis frame than would be indi-
Finite-element analysis can be made as complicated as cated by the ratios quoted.
the computer will allow but must always be used as a check It is clear from the sample bending moment diagram in
against a proposed design; when used for a complete Fig. 9 . 7 ~that a door-opening in front of the front-axle
three-dimensional analysis the method gives the designer will be in the area of low bending moment and shear
little guidance on how to improve a structure. However, (particularly for a rear-engined vehicle) and no difficulty
as will be seen, simple redundant frameworks can be ana- would be expected in providing an adequate structure
lysed by the computer programmes for side-frames and round the door-opening to carry the small loads.
grillages given in (9), and the results incorporated in non-
computer analyses to give a useful picture of how the main
loads are carried in a mixed structure. Vehicles with a door between the axles on one side
In buses for standing passengers the door-openings on
Bending one side need to be larger so that an asymmetric structure
results. A particular case investigated was the Ha'argaz
Vehicles with no doors between the axles bus, shown in Fig. 9.12. This vehicle was the subject of an
The bending loads in a composite vehicle will be shared early attempt by Hochberg (10)at the type of calculation
between the chassis frame and the bodywork in propor- described here. It is mounted on a chassis with the engine
tion to the stiffness of the two structures. In practice this in the middle and is typical of the all-metal bodywork used
means the stiffness of the side-frame of the body and the for both city and rough roads.
Proc lnstn Mech Engrs 1969-70 Vol 184 Pt 3M
Downloaded from pcp.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016
76 G. H. TIDBURY
that of the side-wall beam between the floor and the waist-
rail.
This rather surprising result was confirmed by tests on
polystyrene models (Fig. 9.13). The photographs are
double exposures, taken before and after the load was
applied. In the models no attempt was made to simulate
Diminensions in m m
the bending stiffness of the side-wall between the floor and
the waist-rail, but the window-pillars, cant-rail and door-
sill are in proportion to the respective second moments of
area in the actual structure.
In practice the bending moment on a vehicle is always
generated by vertical loads and the pure moment case
2500 considered above does not arise. Deflection of each side
was therefore compared under a single load midway
Fig. 9.12. A typical asymmetrical bus w i t h large doors: between the axles. The polystyrene models now showed
the first step in analysis was t o compare the stiffness
in bending of the t w o different body-sides about three times the deflection (Fig. 9.14).
The bending stiffness of the side-walls between the floor
and the waist-rail has been correctly represented in
Since the bending stiffness of the asymmetric body and Fig. 9.14 to the same scale as the pillars in Fig. 9.13. A
the chassis frame must be compared to assess the propor- calculation based on the assumption made by Erz (Fig.
tion of the load carried by each, the first approach was to 9.7b) on the way the shear force is carried over the door-
compare the stiffness of the two body-sides under pure opening gave a ratio of flexibilities of the two sides as
bending. The Erz assumption (Fig. 9.7a) was used for the 3.15 to 1.
side with a door-opening and it was found that thestiffness Since the deflection due to the window-pillars for a
of the window-pillars in bending, while transferring the bending load on the structure, calculated as in Fig. 9.70,
compression load in the cant-rail, was comparable with was small, it can be assumed that, for this particular
Fig. 9.13. Polystyrene model of the bus in Fig. 9.12. Double exposures (before and after loading) confirmed
that the stiffness of window-pillars in bending was comparable t o the whole side-wall up t o the waist-
rail. A pure moment was applied
vehicle, stiffness in bending of the side-frame includes cent increase in the bending moment of the cant-rail and
some contribution from the cant-rail, giving an increased door-sill, suggested by Erz in Fig. 9.7b, does allow ap-
bending modulus. This increase is shown in Fig, 9.15 as proximately for the deflection of this ring.
an arbitrary factor but it is based on the computer result The two sides with different flexibilities are joined by a
for a side-frame without a door, shown in Fig. 9.16. The series of rings and the flexibility in shear of these rings
computed deflection indicates that the ring round the will affect the overall flexibility of the body. If the sides
door-opening deflects in a complicated way. The 50 per of the rings are very stiff, the shear will be carried by the
Fig. 9.14. Model side frames with central load applied: all beam stiffnesses are t o scale
Fig. 9.15. The bending flexibility of the t w o sides of an asymmetrical bus compared
i
I--'
.'* -
Fig. 9.1 6. Computed deflections for side-frames under central load
roof-bows and floor cross-members, in the same way as rear ends. The assumption is also made that the panelled
the cant-rail and door-sill carry the shear over the door- structure below the waist-rail is stiff in shear compared
opening in the side of the frame. with the window-pillars.
The effect of one such ring at the centre of the wheel- By making three of the four vertical frames infinitely stiff
base can be calculated by the formula given in Fig. 9.17. in turn, the overall flexibility of the structure in torsion can
The flexibility (f3) of this ring can vary widely with the be found when the window-pillars in one frame (side- or
stiffness of the sides. The effect of adding the other rings end-frame) are deflected in bending. The formula is :
can only be found by a grillage analysis. e - e2Z2hI3
This has been carried out on the computer and the _
result compared with the result from the polystyrene M , 27A2b2E2I,. = f i
model as shown in Fig. 9.18. The computer gave the ratio The notation is defined in Fig. 9.8 and 2I,.is the sum of
of the deflections of the two sides as 1.64 to 1, and the the second moments of area of the window-pillars in the
model 1.8 to 1. To obtain the former it is necessary to flexible frame.
add the stiffnesses of all the rings in the structure and The overall body-shell flexibility is then :
4
further increase the stiffness (llf3) by a factor of 1.75. In
fact, the overall flexibility of the combined structure is f = iC= lf i
not very sensitive to the value of f3, and this analysis The formula allows for asymmetrical side-frames above
indicates that a good approximation can be obtained by the waist-rail, heavy door-pillars, panels incorporating
taking f 3 as the inverse of the sum of the stiffnesses of all route information, etc., on one side only. It does not,
the roof-bows and floor-members which transfer shear however, allow for the increased flexibility when the door-
between the two sides, as indicated in the final formula in opening extends below the waistline and further work
Fig. 9.17. would be required to find the correction for this effect.
Application of these formulae to the Ha'argaz bus gives Based on the above formula, the torsional stiffness of the
a ratio of body to chassis stiffnesses of 7-4to 1 so that 0.88 bus body analysed was 5000 times as great as that of the
of the total bending load will be carried by the body. chassis frame. The latter was estimated by the method of
Because of the effect of the rings joining the two sides, the (11) assuming no warping inhibition. Even if complete
bending load carried by the body will be much more warping inhibition is assumed in the joints of the chassis-
evenly shared than would be indicated by their flexibili- frame (say by the floor-panels acting as large gussets) the
ties. The ratio of the loads carried by the sides is 1-32 ratio of stiffness would be 500: 1 and it can be safely
instead of 3-15 to 1. assumed that, for metal bodies of the type analysed, all the
torsion is absorbed by the body shell.
Torsion To confirm the assumption that the roof and floor act as
The concept in Fig. 9.8 leads to a very simple analysis for shear panels, Fig. 9.19 shows the polystyrene model in
the torsional stiffness of the body-shell of a bus. The floor torsion. The lower model shows the sides connected by
and roof are assumed to be infinitely stiff in shear and all continuous sheets at floor and roof level, front and rear
deflection takes place in the side-frames and the front and bulkheads representing the actual pillar stiffnesses. The
Proc lnstn Mech Engrs 1969-70 Vol184 Pt 3M
Downloaded from pcp.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016
THE STRUCTURAL DESIGN OF BUS BODIES 79
4fz+f3(q%
O v e r a l l flexibility ’I =T= 2f3+f,+fz
Fig. 9.18. Polystyrene model of asymmetric bus with a central load: the t w o sides arejoined by the rings only
Fig. 9.19. The model in Fig. 9.18 under a 1 Ibf corner load (top) and, a t the bottom, the same model w i t h sides
joined by roof and floor shear panels and front and rear bulkheads; under a 5 Ibf corner load
torsionload has beenincreasedfive times over the load used Even within the framework of the present work there
at the top where the only connection between the sides is are some unsolved problems, e.g. the apparent ‘shorten-
provided by the rings. The vertical deflection of the load ing’ of the window-pillars at the front of the vehicle in
point (right-hand near side in the photograph) is seen to the computer-plotted side-frame deflection diagram in
be reduced in spite of the increased load. Fig. 9.16; and the effect of carrying the torsion-shear flow
The torsion test rig consisted of vertical supports at over the door-opening.
three of the axle support points and a vertical load applied In spite of these limitations it is hoped that the methods
at the fourth point. Sufficient horizontal flexibility was outlined will enable design calculations to be carried
provided at two of the supports to prevent constraint of out on chassis-mounted bus bodies with the same confi-
the torsional deflection, There was some horizontal dence as those for integral vehicles.
instability in the test rig which made measurement im- When the proportion of the overall load carried by the
possible but the vertical deflection of the support point various parts of the structure is known, the formulae in
(left-hand support in the Photograph) can be seen to be Fig. 9.7 can be used to calculate the local stresses in the
negligible. critical structural members.
CONCLUSIONS ACKNOWLEDGEMENTS
The work presented has demonstrated the use of simple The results in this paper supersede those of Hochberg
theory to estimate the proportion of the total load carried (10) and Tidbury (12)(which carried some results from
by the body structure of a particular bus. Before the em- (IO)), but the pioneering work of J. Hochberg in this field
pirical constants used in these calculations can be used must be acknowledged. In his present capacity as chief
elsewhere with any real confidence, it will be necessary to engineer of the Ha’argaz Company he has further contri-
analyse several other body-chassis combinations. Since buted by the provision of information on the structures of
this paper has only demonstrated consistency between the vehicle analysed.
simple theory and more sophisticated theory using a com- Much of the work reported here, including the manu-
puter and simplified models, it is necessary to carry out facture and testing of the polystyrene models, was carried
stiffness measurements on full-scale vehicles before the out by A. J. Campos as part of a Technical Essay for
methods presented can be used for detailed stress analysis. submission for a D.Au.E.
Proc lnstn Mech Engrs 1969-70 Vol184 Pt 3M
Downloaded from pcp.sagepub.com at PENNSYLVANIA STATE UNIV on May 10, 2016
THE STRUCTURAL DESIGN OF BUS BODIES 81
Bending flexibilities of side-frames (Fig. 9.15) where F is a normal force applied in opposite directions
Side with no doors. at each end.
The deflection per unit load at the centre of a simply I n this case F = half the unit load since the shear force
supported uniform beam is l3/48EIS,where I, is the effec- on each half of the side frame is half the centre load and
tive second moment of area of the complete side. An I = IU+IL.
upper value for I, could be obtained by considering the Therefore deflection in shear over the door-opening
beam as having the full depth of the side including the would be
cant-rail, waist-rail and other longitudinal members at 123
s = 24E(Iu
floor level, giving a value of 31 905 cm4. This assumes +IL)
that the window openings transmit shear as efficiently as It has already been noted that an empirical factor of 1.5
the continuous panelling below the waist-rail. If it is on the bending moment at the end of the beams makes an
assumed that window-pillars are ineffective in transmitting approximate allowance for the fact that they are part of a
shear a lower value for I, = 1', would be obtained by ring system. This factor becomes squared when applied to
including only the waist-rail as the upper flange and the the deflection. Therefore
floor level longitudinals (and any skirt runners) as the
lower flange. I', = 5700 cm4.
In view of the large difference between these values the
8door-opening = (I*5I2
24JWu 11,) +
and the flexibility f z of the whole side is f 2 = sbeam
computer result illustrated in Fig. 9.16 was used to find an
arbitrary factor which can be applied to I', to give the same sections+ 8door-openfng as in Figm
deflection using the simple formula Overall bending JIejcibility of an unsymmetrical body
e3
a=-- 48EI', - 2.88 x cm/unit load As shown in Fig. 9.17 two unit loads are applied to the
centre of the vehicle, the overall flexibility will therefore
6, (computed result) = 1.67 x cm/unit load be half the deflection given by these loads at the centre
For (6/2). Due to the shear stiffness of the structure joining
I3 e3 the two sides the force acting on each side-frame will not be
a, = - - 1.7 (approximately)
- a unit force but Fl and F2 where Fl+Fz = 2. Let 8,
48EIsy
be the deflection of the side with no door due to a force
This result will clearly only apply to vehicles of this type Fl acting on it and 6, be the deflection of the side with
and the ratio (1.7) would be reduced for larger window- a door due to F2. Then the shear deflection 8, = 6,-S,.
pillar spacing and increased for closer spacing. I t is If the shear flexibility is f3, 8, = (Fl-F2)f3.
hoped that it will give a good first approximation as do the But 6, = Flfl and 6, = F2f,. Therefore
Erz constants.
F2f2-F1f1 = (Fl-F2)f3
Side-frame with doors. 2(f3 +f 2)
This is treated as two sections of beam with the same or Fl = since F, = 2 -Fl
2f 3 + f 2 +fI
effective second moment of area as the side without doors
(I3) with a section over the door-opening subject to the and F - Z ( f 3 +f 1)
shear deflection only which is added directly to the bend- - 2f3+f2+fl
ing deflection. As for other approximations in this analysis The deflection at the centre will be the mean of the
recourse is made to comparison with model and computer deflections at the sides. Therefore
results to justify the addition of the whole of the shear
s=-% + s2z - Flf,+F2f2 - ( f 3 + . f J f i + ( f 3 + f l ) f 2
deflection over the door to deflection at the centre of the - 2 2 f 3 +f 1 +f 2
side. Therefore
Application of the unit load method to the two sections
8 f I f z + f 3f i( +7f 2)
of beam ll and 1, joined by a rigid section over the door
(12) gives the following integrals : Overall flexibilityf = - =
2 2f3+.fl+f:!
The angular deflection due to this pillar will be, by the (3) MICHELBERGER, P. ‘Wirkung der Turoffnungen auf das
unit load method, Kraftespiel der Omnibuskarosserien’, Periodica PO&-
technica (Budapest) Vol. 6, No. 2, 1962 (in German).
0pilla.r = a1 hl 2
Jo (3)
I,
(&) -
e2PI,h13
x2 dx = 27A2b2E(2
(4) MICHELBERGER, P. ‘Das Kraftespiel einer Infolge der
turoffnung unsymmetrischen Omnibuskarosserie mit
elastichen Quertragern’, Periodica Polytechnica (Budapest)
The angular deflection due to all the pillars will be the Vol. 7, No. 3, 1963 (in German).
sum of the deflections due to the individual pillars, i.e. (5) ERZ, K. ‘Uber die durch Unbenheten der Fahrbahn
hervorgerufene Verdrenung von Strassenfahrzeugen’,
e212~,h13
oi = 227A2b2E(2 Ir)2
A.T.Z. No. 4; No. 6; No. 11; No. 12,1957 (in German).
(6) PAWLOWSKI, J. Vehicle Body Engineering, G. H. Tidbury
Since all the terms except I , inside the summation are Ed. English Edition, Business Books, 1970.
(7) ALFREDSON, R. J. ‘The structural analysis of a stressed skin
the same for all pillars we can write bus body’, The Journal of the Institution of Engineers,
e212h132I, - e2Z2hl3 Australia, Vol. 39, No. 10-11, 1967.
8, = f, = (8) LEWIS, R. ‘Structural analysis of an integral passenger coach
27A2b2E(21,)’ - 27A2b2E2I , by matrix force method’, A.S.A.E. Thesis, 1966.
as in the text. (9) WARDILL,G. A. ‘Small computer procedures as tools for
structural designers’, Body Engineering Symposium, Cran-
field, 1970.
APPENDIX 9.3 (10)HOCHBERG, J. ‘Stress analysis of bus bodies’, A.S.A.E.
REFERENCES Thesis, 1967.
(I) BRZOSKA, Z. ‘Basic problems in the statics of self-supporting (11) MARSHALL, P. H., ROACH,A. H. and TIDBURY, G. H.
vehicle bodies’, Archwim Budowy Maszin, Vol. 2, No. 4, ‘Torsional stiffness of commercial vehicle chassis frames’,
1955, Vol. 3, No. 1, 1956 (in Polish). X I I Congress International des Techniques de l‘automobile,
(2) MICHELBERGER, P. ‘Die Untersuchung von Autobussen mit F.I.S.I.T.A., 1968.
Bodenrahmen oder Fahrgestell auf Verdrehung’, Actina (12) TIDBURY, G. H. ‘Integral structures for P.S.V.’s’, Auto-
Teknica Vol. 35, 1961 (in German). motive Design Engineering, September, 1968.