Cheri An 2009
Cheri An 2009
Cheri An 2009
The manuscript was received on 4 February 2009 and was accepted after revision for publication on 29 May 2009.
DOI: 10.1243/09544070JAUTO1153
Abstract: This paper describes the non-linear modelling of a double wishbone suspension
developed to investigate the non-linear kinematics and dynamics in the closed, spatial
kinematic chain configuration. Analytical and ADAMS models are generated and kinematic
and dynamic characteristics of the models are investigated. The analytical model of the
suspension mechanism is an idealized four degree-of-freedom (DOF) model, with suspen-
sion members considered as rigid links and bushings taken as linear spring–damper
elements. The simulation results of the model subjected to a virtual kinematics and com-
pliance (K&C) test are compared with results generated by an ADAMS model, developed
based on parameters obtained from the vehicle manufacturer, subjected to the same virtual
test. The experimental K&C testing on the test vehicle presents a method of capturing the
kinematic characteristics of the suspension mechanism. The comparison of the simulation
and experimental results presented shows that the models are capable of simulating the
characteristics of the pre-existing suspension configuration of the test vehicle.
Keywords: double wishbone, suspension design, ride quality, kinematic and compliance,
K&C, ADAMS, DAE solver
investigated. Using the kinematic derivation to configuration to verify and evaluate the role of
perform static analysis in three-dimensional space, the different parameters in the performance. The
the components that have the highest contribu- equations of motion derived from the analytical
tions to suspension stiffness were identified. Attia model will be used to investigate the significant
[2] had developed a simplified kinematic and terms contributing to the transmission of forces
dynamic model of the double wishbone suspension through the suspension.
configuration in which he neglected the bushing The suspension configuration is shown in Fig. 1.
compliance. While useful at the suspension design The numerous viscoelastic bushings used to mount
stage, the model cannot be used for studying the the suspension elements to the chassis are labelled in
effects of suspension compliance, nor the vibratory this figure. Also visible are elements of the power-
characteristics of the suspension. Specifically, by assisted rack and pinion steering subsystems. The
neglecting the compliance in the bushings, cer- given vehicle configuration has suspension control
tain modes of vibration of the suspension cannot arm bushings that are 30 times stiffer radially than
be analysed. Morman [3] had developed a more they are in the axial direction. The model developed in
complex model of the configuration, considering this paper thus limits the study to the effects of axial
the axial compliances of the lower control arm compliance in the bushing of the upper and lower
bushings. This model is able to generate a more control arms, as discussed in detail in the next section.
accurate frequency response of the system, due
to the inclusion of the dynamics of the lower con-
trol arm. The simulated dynamics of the system 2 MATHEMATICAL MODELLING OF THE
matches the published SAE reference materials DOUBLE WISHBONE SUSPENSION
quite closely. However, since the upper control
arm compliances are omitted, there is a possibility 2.1 Modelling assumptions and preliminaries
of improving the model by including this impor- The schematic of the front suspension subsystem
tant parameter. McGuire and Guenther [4] had model is shown in Fig. 2, with a complete list of
described the methodology used to model the variables and their definition given in Appendix 1.
McPherson suspension quarter-car model using The model considers the axial compliance of the
ADAMS. An ADAMS model is developed to study bushings, represented by the degrees of freedom
the effects of introducing increased longitudinal DXlca and DXuca in the equations of motion (see
compliance in a double wishbone suspension with Fig. 3). Additionally, rotational degrees of freedom
a compliant link. of the control arms auca and alca are considered. The
Numerous papers describe the suspension K&C control arms are assumed to be rigid bodies since
tests used to characterize suspension systems and there is a limit here to the low-frequency modes of
to verify suspension models. Holdmann et al. [5]
have described the use of suspension K&C tests to
characterize the suspension system of a sedan and
a sports utility vehicle, in order to validate the sim-
ulation of the suspension mechanism. The multi-
body dynamics package ADAMS was used to mod-
el the suspension subsystems. The paper also des-
cribed the application of the ADAMS model, which
was verified using the K&C test data, to evaluate
the vehicle handling characteristics.
Based on a review of these studies and prelimin-
ary studies by Cherian et al. [6, 7] and Ayglon et al.
[8] to characterize vehicle suspension configura-
tions, and an examination of the parameters of the
front suspension on the test vehicle (a 464 super
crew version of the Ford F150 2004 model), it was
decided to develop an analytical model of the
double wishbone configuration that considers the
compliance in the suspension control arm bush- Fig. 1 Configuration of the front suspension of the
ings, and also develop ADAMS models of the same F-150
vibration observed in the system, which the human the experimental set-up of the tie-rod, the assump-
body is most sensitive to (1–50 Hz). The knuckle is tion was investigated and showed that the ball joint
represented by a rigid body with six degrees of friction was fairly small in comparison to the
freedom (DOFs), and is connected to the upper and transmitted forces. This may be less relevant to a
lower control arms by frictionless spherical joints. vehicle subject to high cornering forces but is much
The knuckle is also connected to the tie-rod by a less frequent.
frictionless spherical joint. Most current vehicles Since the system of relative coordinates is used,
are equipped with a low-friction ball joint. During the knuckle lower ball joint is fixed in the lower
control arm, and the orientation is specified by a 3– development, while the complete non-linear equa-
1–3 set of spatial angles, i.e. w, h, and y. The tie-rod tions are presented in Appendix 2. The kinetic
itself is modelled as a stiff spring in order to consider energy in the suspension model, denoted by T, is
the effects of compliance in the ball joints and the composed of the sum of the kinetic energy of each of
tie-rod itself. the rigid bodies given by
knuckle can be calculated based on its degrees of Similarly, the energy in the lower control arm is a
freedom. Equating this to the instantaneous position function of the rotational and translational velocity
of the outer ball joint of the upper control arm yields of the lower control arm, obtained from the deri-
three constraint equations. These seven variables vatives of DXlca and alca as follows
2
1 d
KELCA ~ M LA DXlca ðt Þ
2 dt
2
d d
z {sin½alca ðt Þ alca ðt Þ LCA CGy {cos½alca ðt Þ alca ðt Þ LCA CGz
dt dt
2 !
d d
z cos½alca ðt Þ alca ðt Þ LCA CGy {sin½alca ðt Þ alca ðt Þ LCA CGz
dt dt
2
1 d
z alca ðt Þ Ilca xx ð3Þ
2 dt
along with three constraint equations define the The energy in the spindle is a function of the
dynamics of the 4 DOF model per each side of the rotational and translational velocities of the spindle,
suspension. which, when converted to the inertial frame, con-
The equations of motion of the system are tains the terms from the rotational and translational
obtained using the method of Lagrange multipliers displacements of the lower control arm. The kinetic
as follows. The simpler equations are presented in energy in the spindle is hence obtained as follows
line with the following text to show the model
1 d d
KESP ~ M SP DXlca ðt Þ {sin½wðt Þ wðt Þ fcos½yðt ÞSP CG x{sin½yðt ÞSP CG y g
2 dt dt
d d
zcos½wðt Þ {sin½yðt Þ yðt Þ SP CG x{cos½yðt Þ yðt Þ SP CG y
dt dt
2
d
zcos½wðt Þ wðt Þ SP CG z
dt
d d
z {sin½alca ðt Þ alca ðt Þ LCA OBJ y {sin½hðt Þ hðt Þ fsin½yðt ÞSP CG xzcos½yðt ÞSP CG y g
dt dt
d d d
zcos½hðt Þ cos½yðt Þ yðt Þ SP CG x{sin½yðt Þ yðt Þ SP CG y {cos½hðt Þ hðt Þ
dt dt dt
{sin½wðt Þfcos½yðt ÞSP CG x{sin½yðt ÞSP CG y gzcos½wðt ÞSP CG z
d
{sin½hðt Þ {cos½wðt Þ wðt Þ fcos½yðt ÞSP CG x{sin½yðt ÞSP CG y
dt
2
d d d
{sin½wðt Þ {sin½yðt Þ yðt Þ SP CG x{cos½yðt Þ yðt Þ SP CG y { sin½wðt Þ wðt Þ SP CG z
dt dt dt
d d
z cos½alca ðt Þ alca ðt Þ LCA OBJ yz cos½hðt Þ hðt Þ fsin½yðt ÞSP CG xzcos½yðt ÞSP CG y g
dt dt
d d d
z sin½hðt Þ cos½yðt Þ yðt Þ SP CG x{ sin½yðt Þ yðt Þ SP CG y { sin½hðt Þ hðt Þ
dt dt dt
ð{ sin½wðt Þfcos½yðt ÞSP CG x{ sin½yðt ÞSP CG y gz cos½wðt ÞSP CG zÞ
d
z cos½hðt Þ { cos½wðt Þ wðt Þ fcos½yðt ÞSP CG x{ sin½yðt ÞSP CG y g
dt
2 )
d d d
{ sin½wðt Þ { sin½yðt Þ yðt Þ SP CG x{ cos½yðt Þ yðt Þ SP CG y { sin½wðt Þ wðt Þ SP CG z
dt dt dt
1 d d
z cos½yðt Þ cos½wðt Þ hðt Þ z sin½yðt Þ wðt Þ
2 dt dt
d d
Ixx sp cos½yðt Þ cos½wðt Þ hðt Þ z sin½yðt Þ wðt Þ
dt dt
d d d d
zIxy sp { sin½yðt Þ cos½wðt Þ hðt Þ z cos½yðt Þ wðt Þ zIxz sp yðt Þ z sin½wðt Þ hðt Þ
dt dt dt dt
1 d d d d
z { sin½yðt Þ cos½wðt Þ hðt Þ z cos½yðt Þ wðt Þ Ixy sp cos½yðt Þ cos½wðt Þ hðt Þ z sin½yðt Þ wðt Þ
2 dt dt dt dt
d d
zIyy sp { sin½yðt Þ cos½wðt Þ hðt Þ z cos½yðt Þ wðt Þ
dt dt
d d
zIyz sp yðt Þ z sin½wðt Þ hðt Þ
dt dt
1 d d d d
z yðt Þ z sin½wðt Þ hðt Þ Ixz sp cos½yðt Þ cos½wðt Þ hðt Þ z sin½yðt Þ wðt Þ
2 dt dt dt dt
d d d d
zIyz sp { sin½yðt Þ cos½wðt Þ hðt Þ z cos½yðt Þ wðt Þ zIzz sp yðt Þ z sin½wðt Þ hðt Þ
dt dt dt dt
ð4Þ
L 1
bushing ~{ ð9Þ
2
QLCA Klca DXlca
Lq i 2
The three constraint equations obtained by equat- 3 NUMERICAL SIMULATION AND RESULTS
ing the instantaneous position of the outer ball joint of
the upper control arm to the instantaneous position of In order to verify the analytical model developed
the upper joint in the spindle are as follows: in the preceding section, the left front suspension
First constraint equation C1 was excited by a vertical force under quasi-static
conditions. To compare the kinematics of the
C1 ~DXlca zLCA OBJ x analytical and ADAMS models (see section 4), the
ADAMS model was subjected to a virtual K&C test.
zsinðwÞSP uj z{x{DXuca {UA x ð16Þ The simulations used a force that cycled at 3 Hz to
Second constraint equation C2 produce an initial motion of 40 mm jounce followed
by 40 mm of rebound motion before cycling back to
C2 ~cosðalca ÞLCA OBJ y the curb height. The wheel spindle vertical deflec-
tion, the wheel toe angles, as well as camber angles
{sinðhÞcosðwÞSP uj z{y{cosðauca ÞUA y versus force applied at the wheel spindle were
ð17Þ
plotted. The response obtained from this simulation
Constraint equation C3 is similar to what can be expected from a K&C test
with no body roll. The results of the analytical
C3 ~sinðalca ÞLCA OBJ y simulations are shown in Figs 4 to 6.
zcosðhÞcosðwÞSP uj z{z{sinðauca ÞUA y Figure 4 shows the plot of the wheel spindle
ð18Þ vertical deflection versus the applied wheel load
obtained from the simulation of the analytical model.
The set of seven ordinary differential equations
The vertical deflection has been limited to 40 mm on
obtained from evaluating equation (13) and the three
either side of the equilibrium position. At deflections
algebraic constraint equations (16), (17), and (18)
larger than this value, the non-linear effects of the
forms a system of differential algebraic equations
suspension spring start to appear, and the model does
(DAEs) that describes the dynamics of the system.
not predict the effects of suspension spring harden-
The general form of the DAEs can be given as
ing and the bump stops on the suspension being
" # impacted. In the actual suspension mechanism, and
X
7 X
3
L
€ i zCi q_ i zKi q i ~Qi z
Mi q lj cj its ADAMS model, the bump-stops in the suspension
Lq i
i~1 j~1 prevent excessive suspension travel by applying an
X
3 opposing force that is dependent on the deflection in
and cj ~0 ð19Þ the bump-stop. As seen from Fig. 4, the non-linear
j~1 effects of the suspension spring and bump-stop
cause non-linear deflection in the wheel spindle for
To convert the system of DAEs to ordinary linear load variations, although only at deflections
differential equations (ODEs), which are solvable greater than 40 mm of suspension travel. The ADAMS
using a variety of ODE solver suites in MATLAB, model is able to simulate this behaviour better as the
the three algebraic constraint equations were diff- suspension spring is modelled using the spline curve
erentiated. This converts the set of equations ob- data obtained from the manufacturer. A 10 to 27
tained from equations (13) and (16) to (18) into per cent difference can be seen in the comparison of
a set of coupled, non-linear ODEs of the follow- the models. As seen from Fig. 4, the linear approxima-
ing form, which were solved numerically using tion in the analytical model for the spring–damper
ODE15s routine, as detailed in the next section characteristics leads to a force deflection character-
istic with very minimal hysteresis. The hysteresis, as
€ gz½Cfq_ gz½Kfq g~fF g
½ M f q ð20Þ seen in the simulations, barely adds up to 2 N over one
cycle, due to the low cycling frequency. The graph also
shows an absence of any vertical lines that would
indicate unloading of the suspension at constant the analytical model shows no hysteretic character-
force. Given that the model does not account for istic and verifies that the model behaviour matches
Coulomb friction in the ball joints or elsewhere in the the characteristic behaviour of a constrained, spa-
suspension mechanism, this behaviour is expected. tial four-bar mechanism. The model also exhi-
Figure 5 shows a plot of wheel spindle vertical bits the non-linear behaviour expected owing to the
deflection versus toe change from 40 mm jounce to configuration of the suspension mechanism. The
40 mm rebound travel of the suspension from the slight difference in the simulation results can be
curb position. Since this graph shows the kinematic attributed to the difference in the way the tie-rod
characteristic of the suspension, any hysteresis in constraint is modelled. The ADAMS model con-
the graph would mean that the system could have siders the tie-rod as a rigid body with no elastic
multiple physical configurations, but, as expected, deformation, whereas the analytical model consid-
Fig. 5 Wheel spindle deflection versus toe angle change from simulation
Fig. 6 Wheel spindle deflection versus camber angle change from simulation
ers the tie-rod as a linear spring–damper element, onable magnitude, and the tie-rod and constraint
albeit of high stiffness, to account for compliance forces were also of acceptable magnitude, as shown
in the ball joints connecting the tie-rods to the on Fig. 7.
knuckle. PSD (power spectrum density) graphs are shown
Figure 6 shows the plot of the simulation results in Figs 8 and 9 and can be used to avoid unwanted
for suspension deflection versus the camber angle resonance modes such as the wheel hop mode or
change. The characteristic is noticeably non-linear, suspension yaw mode. In the case of this suspen-
as also seen in the simulations of the analytical sion, no significant vertical wheel mode was found to
model. The non-linearity in the change in camber affect the low-frequency vibrations. A slight steer
angle is much higher than the change in toe angle of response around 17–18 Hz can be seen from Fig. 9.
the suspension. The results show that the camber Using similar frequency response plots, several sus-
angle is affected by the vertical motion of the pension designs can be evaluated and the most suit-
suspension to a greater extent than the toe angle. able design can be selected.
This can be explained by the fact that the tie-rod Upon completion of the suspension model tuning,
constrains the toe angle change more than the it was concluded that the mathematical model now
camber angle change, owing to the configuration of
the suspension elements.
Using the parameter data obtained from the
ADAMS model output and the initial conditions
consistent with the loaded vehicle state, the sus-
pension model was simulated using MATLAB. The
inputs were sinusoidal forces in the radial and
tangential directions, and the outputs of interest
were the suspension states and the tie-rod forces.
The suspension states are important in the design
process so as to allow for clearances in the wheel
well. The tie-rod forces are important output para-
meters as they impact the steering feel of the vehicle
and are directly tied into the force and vibration felt
by the driver. As shown in the figures below, the
results were consistent with expectations. The sus-
pension responded in an oscillatory manner of reas- Fig. 7 Suspension constraint force output
Table 1 Spatial position of suspension linkage joints/components in the inertial frame (the tyre contact patch is
taken as the point of reference for all spatial locations)
Suspension components Symbols X (mm) Y (mm) Z (mm)
Lower control arm outer ball joint LCA_OBJ 265 412 0
CG of lower control arm LCA_CG 215 314 96
CG of upper control arm UA_CG 246 74 30
Spindle upper joint SP_uj_xyz 0 0 484
Outer ball joint on tie-rod SP_tr 131 52 93
Inner ball joint on tie-rod TR 77 88 110
Outer ball joint on upper control arm UA_xyz 268 193 0
Shock mount on chassis CHM 257 137 504
Shock mount on lower control arm Susp_mnt 220 268 23
CG of wheel spindle SP_CG 3 25 190
Wheel centre Wf 0 61 116
Upper arm midpoint between the bushings mounts x,y,z 251 99 434
Table 2 Parameters of the vehicle suspension used in the ADAMS and analytical models of the suspension
subsystem
Suspension parameters Symbols Values
Mass of upper control arm M_UA 3.4 kg
Mass of lower control arm M_LA 8.3 kg
Mass of wheel knuckle/spindle M_SP 21.8 kg
Mass of tyre/wheel assembly M_T 52 kg
Inertia of upper control arm (about the axis of rotation) Iuca-xx 1022 kg m2
Inertia of lower control arm (about the axis of rotation) Ilca-xx 9.7461022 kg m2
Inertia of wheel knuckle Isp-xx, Isp-yy,Isp-xz, 0.09, 0.15, 3.461023,
Isp-xy, Isp-zz, Isp-yz 0.012, 0.09, 4.661023 kg m2
Rotational inertia of tyre/wheel assembly It 1.2 kg m2
Nominal stiffness of suspension spring Ksp 43 300 N/m
Damping factor of shock damper Csp 2565 N/m s
Stiffness of upper control arm bushing (in fore–aft direction) Kuca 303 000 N/m
Stiffness of lower control arm bushing (in fore–aft direction) Klca 497 500 N/m
Damping factor in upper control arm bushing (in fore–aft direction) Cuca 4560 N/m s
Damping factor in lower control arm bushing (in fore–aft direction) Clca 9015 N/m s
Tie-rod unloaded length tr_length 0.3482 m
Tie-rod stiffness Ktr 650 000 N/m
Fig. 11 Out-of-phase vibration mode of the suspension control arms from ADAMS/Vibration
analysis
to fulfil its primary purpose of absorbing vertical spring force also lasts longer than as predicted by the
loads, a very small percentage of the fore–aft force is analytical model. Given the suspension configuration
actually transmitted to the chassis via the suspen- with the suspension spring connected to the lower
sion spring. This is reflected in the testing data control arms, any vibration in the control arms would
where the accelerometers recorded negligible lateral also be represented in the force levels in the sus-
acceleration while the steering rack and nibble pension spring. The radial and axial stiffness and
accelerations are much higher in magnitude. damping are lower in the lower control arms than the
Figure 14 shows that the force levels in the sus- upper control arms. This, in turn, forces the vibration
pension spring due to a longitudinal force applied in the fore–aft and lateral directions of the lower
to the wheel spindle are about 0.6 N, as obtained in control arm to take longer to be damped out, cor-
the ADAMS simulations, compared to 0.7 N for the respondingly increasing the time required for the
analytical model. The oscillation in the suspension force in the suspension spring to stabilize.
Fig. 13 Time history of force in the tie-rod due to a step input force of 220 N applied to the wheel
spindle in the fore–aft direction
Fig. 14 Time history of the suspension spring force due to a step input force of 220 N applied to
the wheel spindle in the fore–aft direction
5 KINEMATICS AND COMPLIANCE (K&C) toe) with suspension travel. Typically, the compli-
TESTING ance characteristics show force versus deflection
curves for vertical, longitudinal, or lateral loads
A kinematics and compliance (K&C) test is used applied at the wheel contact patch. Figure 15 shows
here to verify that the analytical modelling of the an instrumented vehicle on the K&C test rig.
suspension subsystem matches the characteristics Figures 16 to 18 depict the results of the K&C test
of the suspension subsystem on the vehicle. The conducted at Michelin Americas Research Corpora-
K&C test is a shop test where the vehicle chassis is tion (MARC), on the 2004, 464 F-150 truck [9]. A
suspended on rigidly fixed supports and loads are notable point is that there is significant hysteresis in
applied independently on the wheels (see Fig. 15). the compliance test data, which is not reproduced in
Equivalently, the wheels are supported on force the simulation. This can be accounted for by the fact
sensing plates and the body attitude is adjusted, that the simulation models do not consider friction
thereby inducing varying loads on the individual at the ball joints and the hysteresis losses in the
wheels. The change in wheel orientation and deflec- bushings under quasi-static simulations. The ab-
tion of the various components due to the applied sence of any vertical lines in the compliance data in
loads are recorded. The kinematics results deal with Fig. 16 implies that a Coulomb friction model is
the variation of wheel angles (camber, caster, and unlikely to explain the phenomenon accurately. A
slow relaxation effect in the viscoelastic members
that have high hysteretic losses is likely to be a
better model [10, 11].
Fig. 16 Front right wheel deflection versus wheel load from the K&C test
Fig. 17 Front right wheel deflection versus camber change from the K&C test
Fig. 18 Suspension vertical deflection versus front toe angle change from the K&C test
versus the toe angle change, and suspension deflec- ance and designers interested in the tyre dynamics
tion versus camber angle change have non-linear during vehicle manoeuvres. The suspension compli-
behaviour. This is expected given the complex ance characteristics could be better simulated using
motions in the three-dimensional constrained me- different models for the hysteresis in the bushings.
chanism that makes up the suspension subsystem. However, owing to the complexity of the current
The curves for the loading and unloading part of the analytical code, and the ability of ADAMS and the
cycle overlap, indicating that the angles are a func- multibody dynamic package to simulate the effects
tion of the vertical travel of the wheel, and not the accurately, it would be easier to implement these
load applied to get it to this position. As observed by models using multibody codes.
the comparison of the simulation and K&C test, the The current level of simulations has focused on
non-linear behaviour is reproduced by the analytical developing subsystem models and verifying them
and ADAMS models, thus confirming that the through multibody code simulations and experi-
simulated motion of the wheel is correct. mental testing. A significant step forward would be
to integrate the subsystem model with a tyre model,
a steering subsystem model, and a chassis model
7 CONCLUSION AND FUTURE WORK
to create a whole vehicle model to study the effects
of varying the system parameters on the vibration
Based on the results of the K&C tests, the quasi-static
transmission through the vehicle.
ADAMS simulation and the analytical models of the
vehicle suspension subsystem have been verified. It
was observed that the kinematics results of the ACKNOWLEDGEMENTS
suspension model matched the experimental results
very closely, even with the simplified models for the The authors would like to express their gratitude
bushing behaviour. The results of the simulation to Mr Julien Flament, Mr Britton Martin, and Mr
will prove helpful to automotive designers interested Norman Frey of Michelin Americas R&D. The
in tyre attitude variation due to suspension compli- authors gratefully acknowledge and thank Michelin
Americas R&D for the sponsorship and financial Cdashpot equivalent viscous damping of the
support of this research. The assistance provided by shock absorber
Mr David Moline at Clemson University in fabricat- Clca equivalent damping of the lower
ing and calibrating the sensors is also appreciated. control arm bushings
Csp equivalent viscous damping of the
F Authors 2009 suspension springs
Cuca equivalent damping of the upper
control arm bushings
REFERENCES
DXlca deflection of the lower control arm
1 Chen, X. and Lin, Y. Research on the dynamics of along the principal axis of the control
flexible multimode system of passenger car sus- arm bushing
pension. In Proceedings of the 1998 SAE Interna- DXuca deflection of the upper control arm
tional Congress and Exposition, Detroit, Michigan, along the principal axis of the control
April 1998. arm bushing
2 Attia, H. A. Dynamic modeling of the double
Fx tyre force acting on the wheel spindle
wishbone motor-vehicle suspension system. Eur.
J. Mechanics and Solids, 2002, 21, 167–174. in the longitudinal direction (i.e.
3 Morman, K. N. Nonlinear model formulation for global X direction)
the static and dynamic analyses of front suspen- Fz tyre force acting on the wheel spindle
sions. SAE paper 770052, 1977. in the vertical direction (i.e. global Z
4 McGuire, M. K. and Guenther, D. A. Longitudinal direction)
suspension compliance modeling with ADAMS. Ilca_xx moment of inertia of lower control
SAE paper 930764, 1993.
arm about the axis of rotation of the
5 Holdmann, P., Köhn, P., Moller, B., and Willems,
R. Suspension kinematics and compliance. SAE lower control arm
paper 80897, 1998. It moment of inertia of the tyre–wheel
6 Cherian, V., Jalili, N., and Haque, I. Development assembly
of a non-linear model of a double wishbone I_SP matrix of moment of inertia of wheel
suspension for the characterization of force trans- spindle and knuckle composed of
mission to the steering column and chassis. In individual terms Isp_xx, Isp_xy,
Proceedings of the 2004 IMECE Conference, Ana-
Isp_xz, Isp_yy, Isp_yz, and Isp_zz
heim, California, 2004.
7 Cherian, V., Jalili, N., and Haque, I. Non-linear Ixx_ua moment of inertia of the upper
modeling and theoretical development of tire non- control arm about the axis of rotation
uniformity induced tangential steering wheel vib- of the upper control arm
rations. In Proceedings of the 2005 IMECE Con- Klca equivalent stiffness of the lower
ference, Orlando, Florida, 2005. control arm bushings
8 Ayglon, V., Jalili, N., and Cherian, V. Experimental Ksp stiffness of the suspension spring
testing and validation of tangential steering wheel
Ktr stiffness of the tie-rod
vibrations due to tire non-uniformity. In Proceed-
ings of the 2005 IMECE, Orlando, Florida, 2005. Kuca equivalent stiffness of the upper
9 Frey, N. and McKibben, D. Examination of suspen- control arm bushings
sion and steering kinematics and compliance – M_LA mass of the lower control arm
2004 Ford F-150. Technical report FW544, Michelin M_SP mass of the wheel spindle and steer-
Americas R&D, 2004. ing knuckle
10 Dillinger, B. The development of analytical tire
M_UA mass of the upper control arm
models for stiffness non-uniformity, mass imbal-
ance and radial run-out. Masters thesis, Clemson qi vector of generalized coordinates
University, 2005. [DXuca, auca, DXlca, alca, h, w, y]
11 Dillinger, B. L., Jalili, N., and Haque, I. Analytical Qj vector of generalized forces
modeling and experimental verification of tire non- rtf position vector to point of action of
uniformity. Int. J. Vehicle Des., 2008, 46(1), 1–22. tyre–wheel forces on the wheel spin-
dle from the origin of the inertial
reference frame
APPENDIX 1 Wt mass of the tyre–wheel assembly
Notation
auca rotational deflection of the upper
cij elements of the Jacobian matrix control arm
alca rotational deflection of the lower [SP_cgx, vector of position of the spindle
control arm SP_cgy, centre of gravity in the coordinate
dsp net deflection of the suspension SP_cgz] frame fixed in the spindle
spring [SP_trx, vector of position of the tie-rod
dtr net deflection in the spring SP_try, mount point on the spindle in the
representing the tie-rod spatial SP_trz] coordinate frame fixed in the spindle
constraint [SP_ujx, vector of position of the upper ball
Ddashpot relative velocity between the SP_ujy, joint on the spindle in the coordinate
suspension spring shock mount on SP_ujz] frame fixed in the spindle
the lower control arm and the chassis [Susp_mntx, vector of position of the suspension
mount Susp_mnty, sprint mount on the lower control
Dspz vertical deflection of the wheel Susp_mntz] arm in the coordinate frame fixed in
spindle from curb height the lower control arm
DSPh camber angle change [UCA_CGx, vector of position of the upper
DSPQ toe angle change UCA_CGy, control arm centre of gravity in the
h second rotational displacement of UCA_CGz] local coordinate system
the coordinate frame attached to the [UA_x, vector of position of the upper
wheel spindle about the body fixed X UA_y, control arm outer ball joint in the
axis UA_z] local coordinate system
l vector of Lagrange multipliers [l1, l2,
l 3]
w first rotational displacement of the
coordinate frame attached to the
wheel spindle about the body fixed APPENDIX 2
Y axis
y third rotational displacement of the Equations of motion for the suspension subsystem
coordinate frame attached to the
wheel spindle about the body fixed Sum GF alphauca
Z axis
~{M UAg ½cosðauca ÞUCA CG y
[chx_ref, vector of spatial coordinates of the {sinðauca ÞUCA CG z
chy_ref, chassis reference point
chz_ref]
[LCA_CGx, vector of position of the lower Sum GF DXuca
LCA_CGy, control arm centre of gravity in the
d
LCA_CGz] local coordinate system ~{C uca DXuca ðt Þ {K uca DXuca
dt
[LCA_OBJx, vector of position of the lower
LCA_OBJy, control arm outer ball joint in the
LCA_OBJz] local coordinate system
{Shock spring unloaded len |f2½cosðalca ÞSusp mnt y{ sinðalca ÞSusp mnt z{CHM y
½{ sinðalca ÞSusp mnt y{ cosðalca ÞSusp mnt z
z2½sinðalca ÞSusp mnt yz cosðalca ÞSusp mnt z{CHM z
½cosðalca ÞSusp mnt y{ sinðalca ÞSusp mnt zg=
n
ðDXlca zSusp mnt x{CHM xÞ2 z½cosðalca ÞSusp mnt y{ sinðalca ÞSusp mnt z{CHM y 2
oð1=2Þ
z½sinðalca ÞSusp mnt yz cosðalca ÞSusp mnt z{CHM z2
d d
{C susp { sin½alca ðt Þ alca ðt Þ Susp mnt y{ cos½alca ðt Þ alca ðt Þ Susp mnt z
dt dt
|½{ sinðalca ÞSusp mnt y{ cosðalca ÞSusp mnt z
d d
{C susp cos½alca ðt Þ alca ðt Þ Susp mnt y{ sin½alca ðt Þ alca ðt Þ Susp mnt z
dt dt
|½cosðalca ÞSusp mnt y{ sinðalca ÞSusp mnt z{M SP g cosðalca ÞLCA OBJ y
zF z cosðalca ÞLCA OBJ y{k alpha y sinðalca ÞLCA OBJ y
1
{ Ktr f½fDXlca zLCA OBJ xz cosðwÞ½cosðyÞSP tr x{ sinðyÞSP tr y z sinðwÞSP tr z
2
{TR xg2 zðcosðalca ÞLCA OBJ yz cosðhÞ½sinðyÞSP tr xz cosðyÞSP tr y
d
Sum GF DXlca ~{C lca DXlca ðt Þ {K lcaDXlca
dt
1 n
{ K susp ðDXlca zSusp mnt x{CHM xÞ2
2
z½cosðalca ÞSusp mnt y{ sinðalca ÞSusp mnt z{CHM y 2
oð1=2Þ
z½sinðalca ÞSusp mnt yz cosðalca ÞSusp mnt z{CHM z2
1 d
EOM1~ M UA 2 { cos½auca ðt Þ auca ðt Þ d alphauca ðt ÞUCA CG y
2 dt
d
{ sin½auca ðt Þ d alphauca ðt Þ UCA CG y
dt
d
z sin½auca ðt Þ auca ðt Þ d alphauca ðt ÞUCA CG z
dt
d
{ cos½auca ðt Þ d alphauca ðt Þ UCA CG z
dt
f{ sin½auca ðt ÞUCA CG y{ cos½auca ðt ÞUCA CG zg
z2f{ sin½auca ðt Þd alphauca ðt ÞUCA CG y{ cos½auca ðt Þd alphauca ðt ÞUCA CG zg
d d
{ cos½auca ðt Þ auca ðt Þ UCA CG yz sin½auca ðt Þ auca ðt Þ UCA CG z
dt dt
d
z2 { sin½auca ðt Þ auca ðt Þ d alphauca ðt ÞUCA CG y
dt
d
z cos½auca ðt Þ d alphauca ðt Þ UCA CG y
dt
d
{ cos½auca ðt Þ auca ðt Þ d alphauca ðt ÞUCA CG z
dt
d
{ sin½auca ðt Þ d alphauca ðt Þ UCA CG z
dt
fcos½auca ðt ÞUCA CG y{ sin½auca ðt ÞUCA CG zg
z2fcos½auca ðt Þd alphauca ðt ÞUCA CG y{ sin½ausa ðt Þd alphauca ðt ÞUCA CG zg
d d
{ sin½auca ðt Þ auca ðt Þ UCA CG y{ cos½auca ðt Þ auca ðt Þ UCA CG z
dt dt
d 1
z d alphauca ðt Þ Ixx ua{ M UA
dt 2
f2½{ sinðauca Þd alphauca UCA CG y{ cosðauca Þd alphauca UCA CG z
½{ cosðauca Þd alphauca UCA CG yz sinðauca Þd alphauca UCA CG z
z2½cosðauca Þd alphauca UCA CG y{ sinðauca Þd alphauca UCA CG z
½{ sinðauca Þd alphauca UCA CG y{ cosðauca Þd alphauca UCA CG zg
{Sum GF alphauca {l2 sinðauca ÞUA yzl3 cosðauca ÞUA y
d
EOM2~M UA d DXuca ðt Þ {Sum GF DXuca zl1
dt