Sab
PROCEEDINGS OF THE 3" INTERNATIONAL GONFERENDE
POWER TRANSMISSIONS '09
12 GoToseR 2003, CHALKIpIK, GREECE
ou. A. MinaIBIS
LEAF SPRING SUSPENSION DESIGN OF COMMERCIAL VEHICLES BASED ON
‘THEORETICAL METHODS
Georgios SAVAIDIS, Alexandros SAVAIDIS, Michail MALIKOUTSAKIS, Nikolaos VAXEVANIDIS
Abstract: A finite element (FE) based procedure is proposed to evaluate, assess and, if necessary, optimise the kinemat
ies and stress-strain behaviour of spring suspension systems for commercial vehicles, It gives assistance to the design
engineerin refer t0 (a) important issues and details thas have to be taken into account during FE modelling of the sus-
pension system, (6) load configurations that have to be calculated and (c) standardized design loads that have to be
applied. Standardized wheel foad spectra (SPRILOS) are suggested to be used for the derivation of appropriate design
Toads for the theoretical analyses. The advantage of the proposed methodology is that it can be applied in a very early
stage of development, when prototypes are not available, since it uses only standardlized leaf material and geometric
configuration data, Furthermore, the recommended design load spectra according to SPRILOS seta reliable basis to
derive adequate test spectra for the series release durability approval of leaf spring suspension systems.
Keywords: Fatigue, automotive, leaf springs, suspension systems, finite elements
1, INTRODUCTION
Leaf springs are a very efficient solution fr the guide of
rigid axles from both a technical point of view and with
rogard to costs. The basic advantage of leaf suspension is
that axle guiding and suspension are incorporated in one
component (Figure
Fig, 1. Leaf spring based suspension and guiding system
of a non-driven front axle
Owing to its highly integrated functions leaf suspension
poses a variety of challenging technical problems. The
two basic problems are the guarantee of perfect position-
ing (optimized kinematics behaviou) and of an adequate
service life. The present paper focuses on these {Wo is-
sues, It shows @ finite element based methodology to
evaluate, assess and optimize both the kinematics and the
stress-strain behaviour of leaf spring suspension systems
of commercial vehicles. The corresponding design loads
are derived from standardized wheel load spectra reported
in literature, In addition, the wheel load data used setle a
reliable base for the derivation of appropriate fatigue test
Programmes applied for the experimental durability ap-
broval of the most promising design solution,
2. THEORETICAL ASSESSMENT BASED ON
FE ANALYSES
Following initial discussions among the responsible ena
neers, during which requirements and wishes are defined
and their economic realisation is cousidered, the design
engineering department brings its wide experience to bear
and designs «suitable springypile. Besides the desired
spring rate and camber, minimisation of maintenance
‘Work is also taken into account. In new generation springs
plain bearings in both leaf spring eyes have been dis-
pensed with. They are replaced by maintenance-free mo-
lecular bearings
‘Considering these topics and on the basis of simple ana
lytical calculations (1, 2] the design enginecring depart
‘ent proposes first Solutions for the spring and axle
kcinematics. In a second step, the axle kinematics and the
stress-strain behaviour of the spring must be calculated
and, if necessary, optimized, Since the springs are sub-
jected toa large amount of deformation, iterative, geomet-
Tically nonclinear finite element caleuations must be
‘made which consider changes in the geometry and any
contacts that may arise, The modelling procedure has
been initially reported in (3, 4}
Finite element mesh
Figure 2 shows a typical finite element mesh for a non-
driven front mle suspension system. It consists of the
springs with suspension, the axle beam, the steering
knuckles with wheels, the wheel steering arm, the steering
drag Tink and the drop arm. Whilst the springs and the
axle beams are illustrated in detail with volume elements,
the other components are modelled using bar elements.
309Fig. 2. Finite element modelling of the axle suspension
‘and guiding system
Axle kinematies
‘The objective of the calculation analysis of the axle kine-
‘matics is to minimise the influence of the leaf spring de-
formation on the roll-steer behaviour. Emphasis is given
(0 vertical loading and braking events,
An optimum situation is achieved if elastokinematics
‘movements of the front axle do not cause any reaction at
the stecring system. In the case of vertical deflections of
the springs this condition can be relatively easily achieved
by (1) modifications in the design of the wheel steering
arm or 2) a change in the postion of the steering gear
Therewith, optimised kinematic behaviour that is free of
Selfinduced roll-steer effects during vertical deffection
can be obtained,
The situation is considerably more difficu
braking, The acting brake force yields an "S*-shaped de-
ection of the spring (Figure 3), This "S”-shaped deflec.
tion ofthe spring causes a change in the kinematics ofthe
front axle system which can result in unwelcome sworn.
ing of the vehicle. In addition, the brake bending moment
causes the axle system to rotate about the Y-axis, The
resulting subsequent displacement of the wheel steering
arm, which is not situated at the instant centre of this
‘movement, in both dircctions, x and z, can lead to a roll,
steer effect, Further, as a result of braking, longitudinal
forees are acting inthe negative z direction causing tongi-
tudinal displacements. if the kinematics influences are
cotimised this effect can, however, be completely sup-
pressed [5]
in the case of
tvs arecton
Fee 74 i
Fre sont
fe Ay
Pig. 3. S-shape deformation and sres distribution curing
braking
310
Stress calculation
A sro of ving tet in th past in which stain gage
‘ete applied fo he sping lees, hve shone nae
arate loa sess ahd stain aie ia voice ee
Wiley (6,7) Figute 3 shows the stereo
along the lea lengt scoring fo the FE eae
th event of violet braking, Very nigh beseech
seu of th sings on be ober. Tis ype of ea
therefore considerably in uenes the sie eas
spring sytem, The magnitude of the loa st
Strain canbe restioted by usage of additonal canta
sing clement, the so-called idle and "5"
see Fig Hed tothe vohicle ame, whieh eee
the prevalng fres from he spring Ine thee
Owing to their frequency, pute verted! dfectons are
ao of grat import, inthis case eo ne
stesies due to vertical loading can be kept exec
wie ranges ofthe leaves its geomet dese
timed
A fiter ital oe stustion is onesded compression
tn ane to eloulate the leaf tehovour at ie tae
Touding, the FE model nmst be extended to ako go
vehicle fame. As ares of the asymmettal dee
ton that isto be expected the load on the sping fn,
influenced bythe rgiiy of the fame The oe
has not yt been cared oul and is therefor ot dey
tere
Inde ate ofthe designed sping lent opin the iity
af the wo sping eyes, which are neceaaey fos ap
Kinematics, the thikess ofthe ea has been cle leet
te sulicient ( Keep the acting lel sense belo he
alewabe (rea) vale
A ajor task forthe optimisation oF kinematisand expe
Gilly the suesstrain behaviour by means of Ts og
Knowledge of maximum design load, load values te
ray aise daring the various driving ianocuvren Desi
lead en be aceatly derived fo conespondny eked
load measurements, if availble for the aie seen
confgiation under cotsieration. Noweveh ip hens
early stuge of development such menses nesmell,
do not exis. In this sage the standardised deter eed
specs SPRILOS (SPRIng LOed Seuenes) [ye sek
gesied 10 be used to detive appromite aoe lok
Thereby, maximum vues for all oad compensa (oat
ca, longitudinal, transversal acting atthe wheel estat
Point ae proposed dependent on the pajead far sb
silty of occurence of P%, Le. only 1% of pose
‘operational loads may oveteed ihe proposed vluce
3. TECHNOLOGICAL INFLUENCING FAC-
‘TORS
‘Besides the mechanical stress-strain behaviour that can be
accurately calculated by means of FE analyses, techno-
logical factors resulting from the thermo-mechanical sur-
face treatment, e.g. residual stresses, and the surface al-
‘erations, ie., surface integrity states, strongly influence
the fatigue life of leaf springs and have to be taken into
consideration. These influences are briefly presented be-
low.
Surface integrity of fabricated parts is a comprehensive
oneept which describes the change of chemical and me-
(ee caw cae et RRu
i
ee
‘chanical propertios of material in surface layer, induced
‘during different manufacturing processes. It is usually
divided info 4 main categories [8]: (a) sucface geometry
factors: such a8 surface roughness, sretch, ele; (b) sur-
face structure factors: microstructure, dislocation density
and configuration, non-metallic inclusions, ete; (c) me-
chanical factors: micro-hardness in the surface layer, sur-
face strength, surface fatigue limit of material, as well as
residual stress distribution long surface layer and (4)
chemicaVelectrochemical factors: surface free energy,
electric poteney entropy, ete. Aer machining, the ma-
chine paris often undergo different manufacturing proc-
‘esses: heat-ieatment (quenching and tempering), grinding
(or even electro-polishing) and, in some cases, electro-
plating, The last two are, however, only exceptional used
in leaf spring constructions. All these processes cause
different surface integrity changes on the machine parts,
‘whieh have important influence on their mechanical prop-
ceties, especially on their resistance to failures initiated
from the surface, such as fatigue and sttess corrosion
cracking,
Despite considerable advances in understanding fatigue
ver the last 30 years, the role of surface roughness is not
fully understood. While changes in roughness affect the
fatigue life, the effect is difficult to separate from the
changes in surface chemistry and structure. The Lipson
curves [9], which provide a load (stress) factor depending
on the surface conditions, have been published for some
time, but they should only be used to capture qualitative
trends. The Lipson approach uses the fatigue notch factor,
,, which depends on the theoretical stress concentration
factor, material hardness (or strength), and notch size,
Although this approach is useful as @ first approximetion,
‘better model that encompasses the deteiled characteris-
ties of the surfaces and the fatigue crack grovrth behav-
‘our is needed to capture the lifetimes over a broad range
of conditions; see Ref. [10}
In conventional fatigue life prediction, surface quality is
accounted by using empirical correction factors that alter
the endurance life. Surface factors are defined according
to the machining process such as grinding, forging and
polishing. Within each category, surface measurements
yield geometric averages such as average roughness R,,
Taximum peak-to valley height Ray, and 10-point rough-
ness R, A surface factor is found by regression wi
or several such parameters, providing an empiri
the observed recinetion in endurance life. This can yield
accurate predictions when used with appropriate reduction
factors, Finding the appropriate ones, however, is not
straightforward due to the many alloys and machining
processes available as it s indicated in (11).
Jn general, any deviation from a polished surface reduces
endurance strength because the rougher surface provides
sites where locally increased stresses or inegularities in
the material structure promote the initiation of micro-
scopic cracks that can progress to fatigue failures. Menue
facturing processes, corrosion, and careless handling pro-
duce detrimental surface roughening
Fig 4 adapted from data in [12], shows estimates for the
‘endurance strength compared with the ultimate tensile
strength of stels for several practical surface conditions.
‘The data first estimate the endurance strength for the pol-
ished specimen to be approximately 0.5 times the
ultimate strength and then apply a factor related to the
surface condition,
tng 8h
Perrier
‘ie
EEE #E
‘wt ae
Fig. 4. Endurance strength versus tensile strength for
‘wrought steel of various surface conditions [12]
From the curves plotted in Fig, 4 itis evident that special
attention should be given to surface finish of critical eom-
ponents exposed to fatigue loading in order to benefit,
from the steel’s basic strength,
Besides surface finish, residual stress field developed
during the manufacturing process and the subsequent
thermo-chemical treatment of leaf springs is the main
surface integrity parameter which affects considerably the
fatigue lite
Fatigue failures typically initiate at locations of relatively
high tensile stress. Any manufacturing process that tends
{o produce residual tensile stress will decrease the endur-
tc strength of the component. Welding typically is
known as a process that produces residual tensile stress.
Grinding and machining, especially with high material
removal raies, also cause undesirable residual tensile
stresses, Critical areas of cyclically loaded components
should be machined or ground in a gentle fashion with a
possible subsequent thermal treatment
Processes that produce residual compressive stresses have
been proved to be beneficial. Amongst these processes
shot peening and stress peening is extensively used t
Improve the fatigue life of leaf springs [13]
‘The positive influence of shot peening on the fatigue
properties of metallic parts is attributed mainly to two
reasons: (a the formation of residual compressive stresses
at the surface or at layers near to the surface and (b) the
col working effect of the surface area yielding to a
higher strength, For a variety of applications not only in
the automotive but also in the aeronautical indusiry and
for biomaterials shot peening has grown up to an indis-
pensable treatment for cyclic loaded parts
A discussion on the development of residual stresses due
to shot peening in beyond the scope of present paper and
the relative literature is quite extensive; see for example
[14], However it should be emphasized that two different
patterns of residual stresses due to shot peening are en-
countered: Shot pening can produce compressive resid
ual stress distributions with a maximum magnitude atthe
surface, often observed in soft materials, as well as with a
au‘maximum magnitude below the surface, mostly observed
jn materials of medium harness and exclusively ob-
served in hard materials,
‘Two different deformation processes accounts for these
patterns,
One of these processes is the direct plastic elongation of
layers very close to the surfuce as a consequence of tan-
ential forces due to numerous shot indentations, This
‘Brocess is comparable ti hammering of the surface and is
indicated by an increase of surface roughness or surface
‘hardness. The elastic-plastic elongation of the surface
layer results in compressive residual stresses with a
‘maximum magnitude at the very surface, as illustrated in
Fig 5(b), if this effect isthe only or predominant one
The second residual stress generating process can be
Hertzian pressure which arises as a consequence of the
vertical force connected with the impact of each shot bal.
The theory of Hertz describes the course of normal
stresses below a surface on which for instance a ball is
Dressed statically; see also {14, 15], This type of residaal
stress pattern is illustrated in Fig 5(8),
‘Typical residual stress distributions in hardened steel
(J6MnCr5, 385 HV) after peening with shots of different
diameter and with different shot velocities are presented
in Fig. 6. Such distributions are quite favourable as far as
the fatigue resistance of ieaf springs is concermed; see also
(13),
Sesaerge
ALLL ey
Fig. 5. Schematic illustration of the formation of vesiduat
stresses as a consequence of two competitive deformation
‘mechanisms in shot peening [14]
Fig. 6. Residual stress distributions in hardened steet
‘fter peening with shots of different diameter and with
different shot velocities
312
Whilst residual stresses due to shot peening favours fa.
tigue resistance, the morphology of the peened surface
exerts an opposite effect; the inerease of sutfie rough,
ness introduces microscopic stress raisers that reduce the
crack initiation life compared to perfectly smooth spect
‘mens. This roughening ofthe surface by shot peening and
its influence on fatigue or corrosion fatigue properties ia
rarely described in literature, The commonly used rough.
hess parameters Ror R, ae not sufficient for examining
of the influence of surface conditions of shot peened
parts. In order to estimate their notching effect, it fs nec
ssary to have statistically based information ‘about the
real shape and depth of the different types of surface
flaws; see also [11].
4, DURABILITY APPROVAL
Though the technological influences are qualitatively ex-
plained, their quantitative calculation still lacks om acon.
roy. Therefore, the durabitity approval required for the
series release of a spring is nowadays performed exper.
mentally. The complexity of the test rig depends om the
future operational load types, the failure risk of the devel,
‘oped spring system and the causes that may arise to hus
mans and the environment due to its failure. Front axle
teal springs experience the most complex and damaging
loud configurations compared to rear axle springs. They
‘ppear during braking and comering manoeuvres a5 wel
4 from vertical loads during straight-ahead driving
‘The major task forthe durability approval isthe definition
of adequate fatigue test programmes. The standardised
‘design loud spectra SPRILOS [7] give valuable assistance
to derive appropriate test programmes
Besides the maximum (design) values for all load compo-
nents (vertical, longitudinal, transversal) acting ‘et the
Whoo! contact for Po=1%, SPRILOS contains proposals
‘ezarding the shapes of the load spectra for certain sums
of frequency of occurrence. It should be mentioned, that
design spectra according to SPRILOS are based on huge
hnumber of load and strain measurements during feld op.
«ration (especially focusing on west Eutopean road cond:
tions) and on test tracks, Thus, they set a comprehensive
land experimentally verified data base for the derivation of
appropriate test programmes,
A ‘ypical design spectrum shape for pure vertical loading
according to SPRILOS is illustrated with dotted line in
Figure 7. The dashed horizontal fine indicates the payload
of the axle
erica! ding stl sheed ering
learn
Fig. 7. SPRILOS- design spectrum and derived
{est spectrum at straight-ahead driving
(Re meer scztsrremmecmonmmie tee enea the case of vertical loading special attention must be
jven to design details concerning the position ofthe axle
‘eam relatively tothe frame and the leaf springs. This is
‘one with the aid of the force-deflection curve for the
ite suspension system.
igure 8 shows an example of a measured force-
leflection path for a front axle suspension system. It can
ve recognised that a vertical load causes a proportional
nerease in the spring deflection end, hence, deformation
ip tothe time of full impect ofthe parabolic spring on the
niddle and S-stoppers (buffers). These buffers restrict the
pring deflection so that a further increase of the vertical
‘ree value yields only a small increase in the spring de-
lection, The force is transmitted via the buffers, which
‘xhibit a considerably steeper progressive characteristic,
liectly into the frame of the vehicle. In view of this and
iv order to prevent the premature failure of parts of the
est rg (frame, axle substitution, et.) itis suggested to
mit the maximum load to the maximum values shown in
he force-deflection curve of Figure 8, since further in-
rease would not significant affect the deformation and
hhorewith the strength behaviour of the leaf spr
Moreover, small load eycles, which are calculated by
Iheoretical analysis using Miner's rule [16] in one of its
rarious modifications [17] to cause nogligible or insig-
rificant damage, can be omitted in order to reduce the
ready long period of testing
Fig. 8 Typical force-deflection curve of an axle
‘suspension system
The determination of appropriate test spectra to cover the
most damaging event of braking during straight-ahead
driving (vertical suspension combined with braking force)
's based on the maximum braking force appearing during
instant braking of the fully laden vehicle and the geomet-
tie configuration of the system spring-axle-wheel. Ac-
sording 10 SPRILOS, Gaussian spectrum shapes can be
assumed for vertical and longitudinal forces acting on the
wheel contact point [7], Figures 9 and 10 show design
(dotted ines) and testing (full lines) spectra regarding the
event of braking. For the derivation of the testing spectre,
the small load cycles with negligible calculated damage,
Were omitted inorder to reduce the testing time,
The less critical situation of backwards braking may also
be covered exemplary by means of the two load steps
shown in Figure 10,
‘Additionally, in order to take account of life-reducing
load sequence effets resulting from the stochastic nature
of the loading in service, following measures are pro-
‘posed: (1) the steps of the test spectra should be distib-
uted in a psendo-random manner over the time, and (2)
the number of cycles indicated in each step should be
reduced by a certain division factor to achieve muttiple
repeats of the load histories until the sum of the frequency
‘of occurrence of the test spectrum is reached.
ett ce
Fig, 9. SPRILOS spectra and derived test spectra for the
vertical load component at braking
Fig. 10. SPRILOS spectra and derived test spectra for
‘te longitudinal load component at braking
aint
It should be noted here that all above mentioned load as-
sumptions should be checked by measuring local defor-
‘mations in highly stressed regions of a prototype leaf
spring during one testing block. The same spring should
bbe installed in a protatype vehicle to determine the acting
local stresses and strains resulting from driving on “hard”
test tracks or fields.
5, CONCLUSION
A finite element based procedure dealing with the evalua-
tion and assessment of leaf spring suspension systems of
‘commercial vehicles has been presented. Modelling de-
tails for accurate calculation of the spring and vehicle
kinematics are discussed. Desiga loads for simulation of
the most critical and damaging driving manoeuvres are
obtained from standardized load spectra according 10
SPRILOS,
Besides the mechanical stress-strain behaviour that can be
accurately calculated by means of FE analyses, techno-
logical factors resulting from surface integrity states
strongly influence the fatigue life of leaf springs and have
to be taken into consideration. Surface finish and residual
stress field imparted by the manufacturing processes and
the subsequent thermo-chemical treatment of leaf springs
is the main surface integrity parameter which affects con-
siderably the fatigue life,
313.‘The proposed methodology of leaf design assessment uses
geometry, lond configuration and standardized leaf mate-
Fial data. Thus, it ean be applied in a very early stage of
evelopment to determine and assess the optimized sus-
pension design accurately. Development and, especially,
testing time and costs can be reduced to a minimum.
Furthermore, the SPRILOS spectra are proposed to set the
basis for the final experimental durability approval of the
suspension system,
REFERENCES
[1] Niemann G,, Maschinenelemente Band 1, Springer
‘Verlag, 1981
[2] Fischer F., Vondracek H., Warmgeformte Redern ~
Konstruktion und Fertigung, Hoesch Hohenlimbarg
AG, Druckerei und Verlag W. Stumpf KG, 1987.
[3] Savaidis C., Schmitt J,, Ricbeck L., Schwaiger F.,
Feivelmayer K., Betriebsfestigkeitsgerechte Fnt-
Wieklung von Parabelfedem im Nutafahezeugbau im
SE-Prozeb, DYM-Bericht 125, pp. 183-192, Darm-
stadt 1998
[4] Savaidis G., Riebeck L., Feitzelmayer K., Fatigue
Life Improvement of Parabolic Leaf Springs in the
Process of Simultaneous Engineering, Materials
Testing 41 (1999), 234.240,
[5] Riebeck L., Figentenkverhalten, Vergleich Versuct-
Berechnung, MAN-technical report, 1997.
[6] Grubisic V,, Fischer G., Automotive Wheels,
Method and Procedure for Optimal Design and Test-
ing, SAB Technical Paper Series 830135, Michigan
1983.
[7] Grubisic V., Determination of Load Spectra for De-
sign and Testing, International Journal of Vehicle
Design 15 (1994), 8-26.
[8] Field M, Kahles'J, Review of Surface Integrity of
Machined Components, Annals of the CIRP 20
(1971) 153-462
{9] Lipson C, Juvinall R.C., Handbook of stress and
strength, New York, Macmillan, 1963
[10] Andrews S, Sehitoghu 1, A computer model for
fatigue crack growth from rough surfaces Intema-
tional Journal of Fatigue, 22, 2000) 619-630.
[11] As SK, Skallerud B,, Tveiten B.W., Holme B., Fa-
tigue life prediction of machined components using
Finite clement analysis of surface topography, Inter
national Journal of Fatigue, 27 (2005) 1590-1596,
[12] Mott R., Machine Elements in Mechanical Design,
Pearson-Prentice Hall, 4 ed, 2004
[13] Savaidis G., Feitzelmayer K., Ostermann M., Light-
‘weight Construction and Improvement of Fatigue
Lifetime of Leaf Spring Suspension Systems. by
Means of Innovating Surface Treatment Technolo-
sies, Buropetn Automotive Congress, SAITS 01129,
Bratislava 2001.
[14] Wobifahrt H,, The influence of Peening Conditions
fon the Resulting Distribution of Residual Stresses,
Proc. 2" Int. Conf, on Shot Peening, 1984, 316-331,
[15] Vaxevanidis NM, Petropoulos GP, Pantazams
CN,, Surface properties of steel plates impacted by
spherical shots, Proc, 9" Int. Conference on Metrol~
ogy & Properties of Engineering Surfaces, (eds. B.-
au
G, Rosen and TR, Thomas), Halmstad University
Sweden, 10-11 September 2003, 2004, 232-239,
[16] Miner M.A., Cumulative Damage in'Fatigue, Jour
nal of Applied Mechanics 12 (1945), 159-164.
[17] Radaj_D., Vormwald M., Ermfdungsfestigkei
Grundlagen ftir Ingenieure. 3, Auflage, Springe
Verlag, 2007.
CORRESPONDENCE
‘Georgios SAVAIDIS, Assist, Prof,
‘Aristotle University of Thessaloniki
Faculty of Engineering
Lab. of Machine Elements
& Machine Design
54124 Thessaloniki, Greoce
gsavaidis@meng.outh.gr_
‘Alexandros SAVAIDIS, Di. Ing,
National Technical University of
Athens
Department of Mechanics
Zogratou Campus
15773 Athens, Greece
asavaidis@tellas.gr
Michail MALIKOUTSAKIS,
Dipl. Ing.,
Aristotle University of Thessaloniki
Faculty of Enginooring
Lab, of Machine Elements
& Machine Design
54124 Thessaloniki, Greece
mmalikou@auth,gr
Nikolaos VAXEVANIDIS,
‘Assoc. Prof,
‘School of Pedagogical & Techno-
logical Education (ASPETE)
Department of Mechanical Engineer-
, ing Educators
\ | 141 21 N, Heraklion, Athens, Greoce
vaxev@aspete.gr