Multibody Syst Dyn (2009) 22: 341–365

DOI 10.1007/s11044-009-9161-7

Improved bushing models for general multibody systems

and vehicle dynamics

Jorge Ambrósio · Paulo Verissimo

Received: 11 September 2008 / Accepted: 7 May 2009 / Published online: 17 June 2009
© Springer Science+Business Media B.V. 2009

Abstract The development and computational implementation, on a multibody dynamics

environment, of a constitutive relation to model bushing elements associated with mechani-
cal joints used in the models of road and rail vehicles is presented here. These elements are
used to eliminate vibrations in vehicles, due to road irregularities, to allow small misalign-
ment of axes, to reduce noise from the transmission, or to decrease wear of the mechanical
joints. Bushings are made of a special rubber, used generally in energy dissipation, which
presents a nonlinear viscoelastic relationship between the forces and moments and their cor-
responding displacements and rotations. In the methodology proposed here a finite element
model of the bushing is developed in the framework of the finite element code ABAQUS to
obtain the constitutive relations of displacement/rotation versus force/moment for different
loading cases. The bushing is modeled in a multibody code as a nonlinear restrain that re-
lates the relative displacements between the bodies connected with the joint reaction forces,
and it is represented by a matrix constitutive relation. The basic ingredients of the multi-
body model are the same vectors and points relations used to define kinematic constraints
in any multibody formulation. One particular, and relevant, characteristic of the formulation
now presented is its ability to represent standard kinematic joints, clearance, and bushing
joints just by defining appropriate constitutive relations. Spherical, revolution, cylindrical,
and translational bushing joints are modeled, implemented, and demonstrated through the
simulation of two multibody models of a road vehicle, one with perfect kinematic joints for
the suspension sub-systems, and other with bushing joints. The tests conducted include an
obstacle avoidance maneuver and a vehicle riding over bumps. It is shown that the bush-
ing models for vehicle multibody models proposed here are accurate and computationally
efficient so that they can be included in the vehicle models leading reliable simulations.

Keywords Multibody dynamics · Vehicle suspensions · Elastometers · Joint modeling ·

Joint clearance

J. Ambrósio () · P. Verissimo

IDMEC, Instituto Superior Técnico, Lisbon Technical University, Av. Rovisco Pais, Lisboa, Portugal
e-mail: jorge@dem.ist.utl.pt
342 J. Ambrósio, P. Verissimo

1 Introduction

The road and railway vehicles must comply with a large spectrum of objectives including
noise reduction, ride enhancement, dynamic behavior, and/or handling improvement while
reducing costs. Even after the vehicle is commercialized it may be necessary to fine tune
suspensions and other functional systems. The tools and models used in the vehicles design
remain valuable for their evolution during the lifespan of the product.
In the framework of vehicle simulator implementation, aiming at real-time simulation,
the development of realistic models is of fundamental importance to understand the driver’s
behavior [1]. The analysis of transient handling response of vehicles, in a virtual environ-
ment or in design, is also dependent on the realism with which the suspension systems are
modeled [2]. The bushing elements are important in the vehicles dynamics because they
handle misalignments between the suspension components, absorb vibrations, and decrease
the transmissibility of the road irregularities to the vehicle occupants. The increase of the
under-steering tendency and vehicles a less responsive are drawbacks of their application
[3]. However, through the design of the suspension bushing joints it is possible to change
the understeering or oversteering of a vehicle making it to behave closer or further away
from neutral steering. The ride characteristics of the vehicle can be changed by modify-
ing bushing elements such as those used in engine mounts [4], and, eventually, to use their
characteristics to optimize the vehicle behavior [5].
In many modeling strategies a vehicle model is obtained by assembling different subsys-
tems, such as suspension subsystems [6]. Instead of using ideal kinematic joints, the analyst
may foresee the use of bushing joints that represent the compliance of the real joints or of
the ideal joint, which may be represented with sufficient accuracy by increasing the bushing
stiffness appropriately. The bushing joints are also commonly used in the attachments of
pneumatic springs used in road and in railway vehicles, influencing in this manner the ride
performance of the vehicle [7, 8].
In any of these cases, the numerical models of the vehicle using bushing joints not only
can, in general, be simulated with larger time steps than those with ideal joints, but also the
dynamic analysis methodologies do not require constraint stabilization, or correction, be-
cause the compliance of the joint handles any deviation from the ideal alignment. The use of
bushing joints instead of ideal kinematic joints is of particular interest in mechanical systems
with redundant constraints, such as redundant manipulators [9], due to the parallelism of the
joints. Approaches of this nature are being reported in multibody applications to robotics
in which the relative displacements are mapped into forces by using stiffness matrices of
articulated bodies [10]. Multibody systems with clearance joints, for which contact between
the bodies that share the joint exists intermittently, are known to put serious challenges to
the computational speed at which they can be analyzed [11–13]. The clearance joints can
be understood as a special case of bushing joints for which the constitutive relation allows
for relative displacements without any force being developed for a given range of relative
motion.
The multibody code DAP3D, based on the multibody methodology proposed by
Nikravesh [14], is used to support the implementation of the developments reported in this
work. In this program the kinematic joints are modeled as perfect joints, i.e., the clearance
existing in real joints or the possibility of the use of deformable elements, such as bushing
elements, are not taken into account. However, clearance in the joints can be easily accom-
modated as shown in the work by Flores et al. [11]. Elastic bushings for multibody systems,
consisting in linear springs to describe the elastic behavior of the elastomer bushing, have
been presented [15]. Since the bushing is a nonlinear elastic elastomer [16], different nonlin-
ear viscoelastic bushing providing accurate prediction of dynamic loads and displacements
Improved bushing models for general multibody systems and vehicle 343

Fig. 1 Bushing Joints in the arm

of a McPherson suspension
system

are proposed [17]. This methodology consists in the assumption that the relaxation function
can be expressed as a sum of functions that have nonlinear deformation and are expo-
nentially decreasing in time. By considering the bushings as sets of nonlinear viscoelastic
forces, which depend not only on the instantaneous bushing deformations but also on their
history, leads to heavy computational costs that are not compatible with their application to
complex vehicle models.
In this work, the concept of bushing joint is used not only to represent the restriction on
the relative motion between two bodies due to the use of elastomers but also to describe ideal
kinematic joints, without requiring the definition of kinematic constraints, and to provide a
general description to clearance joints. In the process, this work proposes a new formulation
for standard translation and revolute joints that, being based on the cylindrical joint, differ
from the standard formulations such as those proposed by Nikravesh [14]. The methodology
presented hereafter is used to model several types of nonlinear elastic bushing with stiffness
proportional damping. The constitutive equations for the bushing joints are obtained through
a detailed finite element analysis using a nonlinear finite element code. The bushing joints
are implemented in the DAP3D multibody code and demonstrated through the dynamic
simulations of two vehicle multibody models, one with perfect joints and the other with
bushing joints. The vehicle dynamics is analyzed, for both models, in two scenarios that are
relevant for ride and for handling.

2 Bushing elements in mechanical joints

Bushing elements reduce the wear of the components of a mechanism by using their energy
absorption capacity to decrease transmitted vibrations. Although this type of elements has a
much broader spectrum of applications, in this study, the use of bushings is only exemplified
in vehicle suspensions. The bushing element presented in Fig. 1, composed by elastomers
and two steel sleeves, the inner and outer ring, illustrates the type of applications that is en-
visaged. Bushings joints are modeled as force arrangements that penalize the relative motion
between the bodies connected. With different spring arrangements most joints with bushings
can be modeled being proportional damping also added to ensure the representation of the
energy dissipation. The objective of this work is to present a straightforward computational
procedure to develop the bushing models.
When using Cartesian coordinates, the equations of motion for a multibody system are
written together with the second time derivative of constraint equations as [14]
    
M Tq q̈ g
= , (1)
q 0 λ γ
344 J. Ambrósio, P. Verissimo

Fig. 2 Spherical joint: (a) Ideal joint; (b) Bushing joint

where M is the mass matrix, q̈ is the vector of the system accelerations, g is the force
vector, q is the Jacobian matrix associated to the kinematic constraints, λ is the vector of
Lagrange multipliers, which are related to the joint reaction forces, and γ is the right-hand
side of the acceleration constraint equations. The ideal kinematic constraints are implied
in the Jacobian matrix and in the acceleration constraint equations. The bushing joints are
described as forces and moments which are added to the force vector.
The solution of the forward dynamics problem, implied by (1), is achieved by using
a variable time step and variable order numerical integrator [18, 19]. Due to the fact that
the position and velocity constraint equations are not used explicitly during the integration
process, it is necessary to eliminate or maintain the violations of the constraint equations un-
der control. In this case, the initial conditions for state variables are corrected using the pro-
cedure proposed by Nikravesh [20], while the kinematic constraint violations are stabilized
using the Baumgarte stabilization method or eliminated by using a coordinate partition [21].
It should be noted that the use of bushing joints leads to a decrease of the number of
kinematic constraints required for the construction of any particular multibody model and,
consequently, to less numerical problems associated to the fulfillment of these kinematic
constraints. It is observed that, generally, when using compliance on the kinematic joints,
the size of the time steps used during the numerical integration of the equations of motion
are larger leading to faster computation times. However, for stiffer bushings, or for com-
pliances with particular nonlinear constitutive laws, this trend may not apply due to the
high-frequency contents that they bring into the multibody system response.

2.1 Spherical bushing joint

A spherical joint allows three free rotations between the bodies connected, preventing only
the relative translation displacements. With the nomenclature described in Fig. 2, the kine-
matic constraints due to a spherical joint are defined as

(s,3) = rj + sPj − ri − sPi − d. (2)

In which vector d is null for the ideal spherical joint. In this case the joint reaction forces due
to the spherical joint are included in the equations of motion of the constrained multibody
system, described by (1), as
Improved bushing models for general multibody systems and vehicle 345

fi = −(s,3)
qi λ
(s,3)
,
(3)
fj = −(s,3)
qj λ
(s,3)
= −fi ,

where the Jacobian matrix associated to the joint, (s,3)

qi , is included in the system Jacobian
matrix q , and the Lagrange multipliers for the joint, λ(s,3) , are a part of the Lagrange
multiplier vector λ.
For a nonideal spherical joint, the vector d is understood as representing the distance
between Pi and Pj as
d = rj + sPj − ri − sPi . (4)
The vector d in (4) is used to define the gap between journal and bearing in a clearance joint
[11, 12] or the deformation of the elastomer in a spherical bushing joint. In the spherical
bushing the joint reaction forces, described by (3), are replaced by forces due to the bushing
deformation over bodies i and j and are represented as

1
fi = [f (δ) + bδ̇] d,
d
(5)
fj = −fi ,

where f (δ) represents the bushing elastic force due to its state of deformation, b is the
stiffness proportional
√ damping, and δ is the bushing deformation. The length of the vector
d is given by d = dT d, while δ̇ is the time rate of change of δ. In this notation it is
allowed that a gap between the bushing and the journal exists, its size being the difference
between d and δ. The journal is in a free flight, with respect to the bearing, if it is inside
the clearance gap. When no gap exists between the two bodies involved in the joint d = δ,
which is common in applications of bushing joints, the bushing elastic force is given by the
constitutive relation calculated by using a nonlinear finite element analysis, as described in
Sect. 4 of this work.
Assuming that no gap exists in the joint, d = δ, the time derivative of δ is

dT ḋ
δ̇ = , (6)
δ

where ḋ is found from the time derivative of (4) as

ḋ = ṙj + Aj ω̃j sPj − ṙi − Ai ω̃i sPi , (7)

where ω is the angular velocity of body i. When the bushing joint is not located in the
centers of mass of the connected bodies, the transport moments are given by
 
ni = ATi s̃Pi fi ,
  (8)
nj = ATj s̃Pj fj ,

where s̃Pi is a skew-symmetric matrix made with the components of vector sPi .
The computer implementation of the spherical bushing joint requires the same set of data
necessary to define the ideal kinematic joint, i.e., the positions of points Pi and Pj that
define the joint in bodies i and j , plus the translation constitutive relation for the bushing,
f (δ), and the damping coefficient b. To represent the clearance joint, the dimension of the
346 J. Ambrósio, P. Verissimo

Fig. 3 Ideal cylindrical joint

gap must be added to the set of input data, and the constitutive relation may be obtained by
using a nonlinear finite element analysis, just as for the conventional bushing joint, or by
assuming Hertzian contact [11].

2.2 Cylindrical bushing joint

A cylindrical joint allows for one free rotation and one free translation between the bodies
connected, preventing the constrained relative translations and rotations. Using the nomen-
clature described in Fig. 3, the kinematic constraints, for the cylindrical joint, enforce the
alignment of the vectors that represent the joint axis in each body defined as

s̃ s
(c,4) = i j . (9)
s̃i d

In (9) only two independent constraint equations exist for each cross product. Such as for
the spherical joint, or any other type of kinematic constraints, the joint reaction forces due
to the cylindrical joint are included in the equations of motion of the constrained multibody
system, described by (1). Their description has a form similar to that of (2).
The cylindrical bushing penalizes the lack of alignment between the unit vectors si and
sj that define the jointby the normal component of the translational displacement, dn , rep-
resented in Fig. 4a, and by the angular displacement θ , in Fig. 4b. Based on Fig. 4, the
tangential component of vector d along the axis of the joint is given as
  
dt = dT sj ATj sj (10)

while the normal component of vector d with respect to the joint axis in body j is

dn = d − dt . (11)

The forces due to the misalignment of the axis in the normal direction are
 1
fi = f (δn ) + bδ̇n dn ,
δn
(12)
fj = −fi .
Improved bushing models for general multibody systems and vehicle 347

Fig. 4 Misalignments in cylindrical, revolution, and translational bushing joints: (a) Translational displace-
ment; (b) Rotational misalignment

In (12) the magnitude of the bushing deformation is δn = dTn dn , and the rate of deformation
in the normal and tangential directions, δ̇n and δ̇t , are

δ̇n = ḋTn dn /δn , (13)

δ̇t = ḋTt dt /δt , (14)

where the time derivative of the tangential and normal components of the misalignment
vector d are given by
  
ḋt = ḋT sj ATj sj , (15)

ḋn = ḋ − ḋt . (16)

The forces given by (12), applied in the points that define the joint, are transferred to
the centers of mass of bodies i and j . The transport moment required for the transfer is
calculated by using (8). The moment due to the angular misalignment of the joint axis, i.e.,
of vectors si and sj , which must be added to the transport moments, requires calculating
angle θ between them. Let the vector u be

u = s̃i sj . (17)

Then, using the definition of cross product, the misalignment angle in the revolute bushing
joint is
θ = arcsin(|u|), (18)
√
where |u| = uT u. The bushing misalignment moment is
 
ni = −[f (θ ) + bθ θ̇ )] ATi uM ,
  (19)
nj = [f (θ ) + bθ θ̇ )] ATj uM ,

where uM = u/|u|. The time derivative of (18) leads to

θ̇ = u̇/ 1 − (u)2 , (20)

348 J. Ambrósio, P. Verissimo

where
u̇ = uTM u̇ (21)
and

u̇ = (Aj ωj − Ai ωi )T u u (22)
ωi being the angular velocity of rigid body i.
The formulation of the cylindrical bushing joint requires the same data needed for the
ideal mechanical joint, i.e., the joint position points, Pi and Pj , the coordinates of points Qi
and Qj that defines the joint axis, plus the translation constitutive relation for the bushing,
the angular misalignment constitutive relation, f (θ ), and the damping coefficients b and
bθ . Notice that, just a described for the spherical bushing joint, a clearance cylindrical joint
can be modeled as a bushing cylindrical joint in which the normal component of the rela-
tive displacement contains a free-flying displacement plus the bushing deformation, and the
misalignment angle also includes a free flight allowance for misalignments.

2.3 Revolute bushing joint

A revolute joint only allows for one free rotation between the bodies connected, preventing
the constrained relative translations and rotations. With the nomenclature described in Fig. 3,
the kinematic constraints due to a revolute joint enforce the alignment of the vectors that
represent the cylindrical joint and prevent the sliding motion along its axis,
⎧
⎨s̃i sj
(r,5) = s̃i d , (23)
⎩ T
dt dt − δt20

where dt is given by (10), and δt0 is the length of vector dt at the start of the analysis.
The formulation for the bushing revolution joint is based on the bushing cylindrical joint,
to which a penalization of the relative displacement in the direction of the joint axis is added.
Based on Fig. 4, this force is
dt
fi = [f (δt ) + bt δ̇t ] ,
dt
(24)
fj = −fi .

The joint reaction force, described by (24), is added to the joint reaction force represented
by (12) and applied in the points defining the joint. This is equivalent to applying the resul-
tant forces to the center of mass of the rigid bodies and adding a transport moment calculated
by (8).
The input for the revolute bushing joint is the same as for the ideal revolute joint, plus the
constitutive relation defined byf (δt ) and the corresponding damping coefficient bt . Note that
in some applications there is no compliance for the sliding along the joint axis. In this case,
the revolute bushing joint is described as the cylindrical bushing joint, by the forces given
by (12) and (19), plus a kinematic constraint described by the third equation of the revolute
joint, in (23). Alternatively, an artificially high stiffness for sliding can be added, provided
that it does not add high-frequency contents to the system response that would, otherwise, be
nonstiff. Furthermore, the use of the formulation described to represent clearance revolute
joints can be understood by using the same approaches described for the cylindrical bushing
joint.
Improved bushing models for general multibody systems and vehicle 349

Fig. 5 Torsion moment in

translational bushing joint

Fig. 6 Mesh of the FE bushing

model

2.4 Translation bushing joint

A translation joint only allows for one relative translation between the bodies connected,
preventing the constrained relative translations and rotations. With the nomenclature de-
scribed in Fig. 3 and the new vectors defined in Fig. 5, the kinematic constraints due to a
translation joint enforce the alignment of the vectors that represent the cylindrical joint and
prevent the rotation about the joint axis. Assume that the unit vectors si and sj are initially
coincident, as depicted in Fig. 5. Let the unit vectors hi and hj be defined as perpendicular
to vectors si and sj , respectively, and also fixed to bodies i and j , respectively. Then, the
kinematic constraints that define the translation joint are
⎧
⎨s̃i sj
(r,5) = s̃i d , (25)
⎩ T
hi hj − h20
350 J. Ambrósio, P. Verissimo

Table 1 Material parameters for

the polynomial form of the C10 0.545816
material law C01 −0.020755
C20 0.003668
C11 0.0048571
C02 0
C30 −0.0001415
C21 0
C12 0

Table 2 Material parameters for

the Ogden constitutive law i μi αi

1 1.061898 0.428246
2 0.0578289 5.71269
3 0.0159176 −4.59726

Fig. 7 Revolute bushing joints constitutive relations for: (a) radial deformation; (b) tangential deformation;
(c) torsion; (d) constitutive relation of the spherical bushing joint

where h20 = hTi0 hj0 is defined for the initial conditions. Note that the dot product between the
unit vectors hi and hj implies the angle α between bodies i and j about the joint axis, i.e.,

α = arccos(hTi hj ). (26)

In this sense the third relation expressed in (25) means that such angle remains constant.
To avoid numerical problems, the unit vectors hi and hj must be defined, initially, in such
a way that α is not close to 0◦ or to 180◦ , i.e., the vectors hi and hj must not be close to
parallelism.
Improved bushing models for general multibody systems and vehicle 351

Fig. 8 Small family car modeled

in this work

The formulation for the bushing translation joint is based on the bushing cylindrical joint
to which a penalization of the relative rotation about the joint axis is added. Based on Fig. 5,
the penalizing moments that must be applied to bodies i and j are

ni = [f (α − α0 ) + bα̇)]si ,
(27)
nj = −[f (α − α0 ) + bα̇)]sj ,

where α̇ is calculated using (20) through (22) with θ replaced by α and with the vector u
now calculated as
u = h̃i hj . (28)
Here, it is assumed that vectors hi and hj are never parallel, which is also a condition for the
numerical stability of the kinematic conditions, given by (25) that define the ideal translation
joint.
The joint reaction force for the translation bushing joint are the same defined for the cylin-
drical bushing joint, plus the moments given by (27), which are added to those described
by (19). The input for the translation bushing joint is the same as for the ideal translation
joint, plus the constitutive relation defined byf (α) and the corresponding damping coef-
ficient b. The representation of the clearance translation joint follows the same procedure
already described for the other bushing joints.

3 Bushing elements

The models of the bushing joints require that their constitutive relations are defined. Be-
cause the bushings are rubber-type materials, their constitutive relations are nonlinear and
characterized by functions that need to be identified. For this purpose, four test cases are
conducted with the finite element (FE) program ABAQUS [22] on an FE model of each
particular revolute joint bushing element. The bushing is modeled with 1404 solid “hybrid”
elements denominated by C3D8H in the finite element code, as depicted in Fig. 6. A rigid
discrete cylindrical plate is created and connected to the nodes of the outer surface. A fixed
boundary condition is prescribed to nodes in the inner surface. By applying displacements
or rotations in the external rigid plate and measuring the reaction forces and moments in the
inner surface, the nonlinear constitutive relations for the bushings are obtained.
The FE program selected has several constitutive laws for nonlinear elastic and finite
deformation analysis, which can be chosen by the user. A polynomial form of the material
law is used here for the calculations force-displacement relations because it provides a better
352 J. Ambrósio, P. Verissimo

Fig. 9 Front suspension system of the small family car

Fig. 10 Rear suspension of the small family car

prediction the experimental deformation values [16]. The Ogden form of the material law
is used to evaluate the moments because it produces a better fit for small angles of rotation.
The parameters values used for both constitutive laws are obtained from Ref. [16], in which
Improved bushing models for general multibody systems and vehicle 353

Fig. 11 Model of the front

suspension

Fig. 12 Model of the rear

suspension

Table 3 Inertial properties and description of the system components (the right-hand side is symmetric)

Body Components description Mass Inertia (kg/m2 )

number’s (kg) ξ ξ/ηη/ζ ζ

Front 1 Chassis 1090 1000/2130/2200

suspension 2 Left Suspension Arm 1.976 0.022/0.010/0.029
3 Left Knuckle 8.897 0.158/0.155/0.048
4 Left Top Spring-Damper 3.382 0.030/0.030/0.008
5 Left Wheel 19.229 0.301/0.496/0.301
Rear 10 Left Suspension Arm 10.286 0.277/0.155/0.429
suspension 11 Left Bottom Damper 2.640 0.001/0.011/0.011
12 Left Top Damper 2.174 0.001/0.016/0.016
13 Left Wheel 25.638 0.324/0.522/0.324

a similar type of elastomer is studied. The polynomial strain energy potential is


N
U= Cij (I¯1 − 3)i (I¯2 − 3)j , (29)
i+j =1

where Cij are material parameters, and I¯1 and I¯2 are the first and second strain invariants
defined as

I¯1 = λ̄21 + λ̄22 + λ̄23 ,

(30)
I¯2 = λ̄21 λ̄22 + λ̄22 λ̄23 + λ̄23 λ̄21 .
354 J. Ambrósio, P. Verissimo

Fig. 13 Generic model for the small family vehicle

Table 4 Center of mass and Euler parameters of the left-hand side of the vehicle (the right-hand side is
symmetric)

Components code Center of mass (m) Euler parameters

(x/y/z)0 (e1 , e2 , e3 )

Frontal 1 0.0000/ 0.0000/ 0.0000 0.0000/ 0.0000/ 0.0000

suspension 2 0.9815/−0.4791/−0.2538 0.0000/ 0.0000/ 0.0000
3 1.0092/−0.6138/−0.1071 −0.0443/−0.0436/−0.0019
4 0.9470/−0.5080/ 0.2757 −0.0443/−0.0436/−0.0019
5 0.9879/−0.7042/−0.1974 0.0000/ 0.0000/ 0.0000
Rear 10 −1.2587/−0.4906/−0.1161 0.0000/−0.1736/ 0.0000
suspension 11 −1.3907/−0.5141/−0.1070 0.0000/−0.4695/ 0.0000
12 −1.2765/−0.5141/ 0.0624 0.0000/−0.4695/ 0.0000
13 −1.4821/−0.6937/−0.1974 0.0000/ 0.0000/ 0.0000

The third invariant is I¯1 = λ̄21 λ̄22 λ̄23 = 1, corresponding to the incompressibility constraint.
Table 1 presents the values used for the polynomial parameters of the bushings material law.
Improved bushing models for general multibody systems and vehicle 355

Table 5 Kinematic joints of the left-hand side of the vehicle (the right-hand side is symmetric)

Joint Joint type Body Body ξiP ηiP ζiP ξjP ηjP ζjP
Q Q Q Q Q Q
number i j ξi ηi ζi ξj ηj ζj

1 Spherical 1 4 0.9363 −0.4970 0.3980 0.0 0.0 0.1233

– – – – – –
3 Spherical 2 3 0.0140 −0.1589 −0.0307 −0.0291 −0.0086 −0.1770
– – – – – –
5 Revolution 1 2 0.8628 −0.3423 −0.2200 −0.1187 0.1368 0.0000
1.0938 −0.3423 −0.2200 0.1123 0.1368 0.0000
7 Revolution 3 5 −0.0291 −0.0323 −0.0913 0.0000 0.0502 0.0000
−0.0291 −0.0671 −0.0944 0.0000 0.0152 0.0000
9 Revolution 10 11 −0.1949 0.0065 0.0000 −0.0913 0.0300 0.0000
−0.1949 −0.0535 0.0000 −0.0913 −0.0300 0.0000
11 Revolution 1 12 −1.2181 −0.5441 0.1489 0.1043 −0.0300 0.0000
−1.2181 −0.4841 0.1489 0.1043 0.0300 0.0000
13 Revolution 1 10 −1.1466 −0.5291 −0.0753 0.1193 −0.0385 0.0000
−1.1466 −0.6356 −0.0753 0.1193 −0.1450 0.0000
15 Revolution 10 13 −0.2377 −0.2095 0.0000 0.0000 −0.0064 0.0000
−0.2377 −0.1505 0.0000 0.0000 0.0526 0.0000
17 Cylindrical 3 4 −0.0286 0.0711 0.2799 0.0000 0.0000 −0.1147
−0.0286 0.0711 0.3579 0.0000 0.0000 −0.0367
19 Cylindrical 11 12 0.0412 0.0000 0.0000 −0.1632 0.0000 0.0000
0.1087 0.0000 0.0000 −0.0957 0.0000 0.0000

Table 6 Suspension springs and dampers data for the left-hand side of the vehicle (the right-hand side is
symmetric)

ID Bodies Coordinates on bodies Spring Damping Undeformed

connected (ξi , ηi , ζi ) stiffness (N s/m) length
(i, j ) (ξj , ηj , ζj ) (N/m) L0 (m)

1 (3,4) (−0.029, 0.071, 0.213) 59190 7919 0.33

(0, 0, 0.094)
3 (11,12) (−0.091, 0, 0) – 15000 –
(0.104, 0, 0)
5 (1,10) (−0.275, −0.562, 0.455) 14530 – 0.115
(1.119, −0.071, 0)

The Ogden energy potential is written as


N
2μi  ε ε ε 
U= λ̄1i + λ̄2i + λ̄3i − 3 , (31)
i=1
αi2
356 J. Ambrósio, P. Verissimo

Table 7 Tire characteristics

Tire Radius 0.2930 (m)
Torus Radius 0.051 (m)
Radial Stiffness 200 (kN/m)
Longitudinal Stiffness 500 (kN/m)
Lateral Stiffness 150 (kN/m)
Cornering Stiffness 30 (kN/m)
Rolling Friction Coefficient 0.01
Radial Damping 0.078 (N s/m)

Fig. 14 Kinematic joints for the rear suspension: (a) model with ideal joints; (b) model with bushing joints

Fig. 15 Kinematic joints of the front suspension: (a) model with ideal joints; (b) model with bushing joints;
(c) revolute bushing joints in the suspension arm

where μi and αi are temperature dependent material parameters, λI , i = 1, 2, 3, are the

principal stretch ratios, and J = λ1 λ2 λ3 . The deviatoric principal stretch ratios are λ̄i =
J −1/3 λi .
The force-displacement constitutive functions obtained for the spherical bushing joints
and for the normal and axial displacement of the revolution bushing joints are presented
Improved bushing models for general multibody systems and vehicle 357

Fig. 16 Vehicle traveling over bumps

Fig. 17 Vertical displacement for vehicle speeds of 60 km/h, 90 km/h, and 120 km/h

in Figs. 7a and 7b. The stiffness obtained for the angular displacement of the revolution
bushing joints is presented in Fig. 7c.
The proportional stiffness damping factor, b, is assumed to be 0.01 for the normal and
tangential directions and zero for the torsion constitutive relations. The translational consti-
tutive relation for a spherical bushing joint is presented in Fig. 7d with the damping factor,
b, assumed to be null.
358 J. Ambrósio, P. Verissimo

Fig. 18 Roll acceleration of the vehicle chassis for speeds of 60 km/h, 90 km/h, and 120 km/h

4 Multibody model of a small family car

The formulation proposed in this paper is applied to the dynamic multibody model of a
small family car, presented in Fig. 8. Two multibody models of the vehicle are considered,
one with perfect kinematic joints in the suspension systems and other with realistic bushing
joints. The system components include the frontal suspension, which is of the MacPherson
type, the rear suspension, which is a Torsion Bar, plus wheels and chassis.

4.1 Suspension systems for the vehicle

The vehicle front suspension is the MacPherson mechanism presented in Fig. 9. This suspen-
sion system is characterized by a telescopic arm responsible for the wheel position control,
which provides the connection between the knuckles and the chassis of the vehicle. Due to
this characteristic, the McPherson structure is a compact suspension system, which makes it
popular in vehicles that have front engines with limited working space.
The rear suspension the vehicle is a torsion bar that is a semi-independent system. For
more compact vehicles, this system has the advantage of not requiring a stabilizer bar, but
Improved bushing models for general multibody systems and vehicle 359

Fig. 19 Deformations of bushing RB1 : (a) Radial; (b) Tangential; (c) Angular

this is not the case of the vehicle used in the applications for which the rear suspension is
shown in Fig. 10.

4.2 Multibody model of the small family car suspensions

All data for the vehicle model presented in this work results from direct measurement of
the vehicle mechanical components supported by the construction of the components of
mechanical system, such as vehicle suspensions, in the geometric modeling software Solid-
Works [23]. None of the data reported here is supplied by the vehicle manufacturer. With the
geometric model of the suspension system elements, it is possible to acquire the rigid body
data, which includes mass, inertia, positions of the centers of mass, directions of the prin-
cipal axis of inertia, and point coordinates for the kinematic joints attachments. Figures 11
and 12 depict the front and rear suspension geometric models with its components properly
labeled and their joint types and locations identified.
The assembly of the front and rear suspensions lead to the geometric model of the small
family car depicted in Fig. 13. The inertia frame XYZ and the initial position of the chassis
referential ξ ηζ are coincident.
The data obtained from the geometric model enabled the construction of the multibody
model of the small family car. The mass, inertia characteristics, and bodies description are
presented in Table 3. The initial positions of the system bodies and the Euler parameters
360 J. Ambrósio, P. Verissimo

Fig. 20 Obstacle avoidance

maneuver

Fig. 21 Vehicle steering inputs for the lane changing maneuver

associated to their orientations are presented in Table 4. The kinematics joints in the system
are presented in Tables 5, while Table 6 presents the characteristics of the spring-dampers
actuators. To represent the tire interaction with the ground, an analytical tire model with
comprehensive slip is used [24]. The data required for the tire of this model is obtained in
Ref. [25] and presented in Table 7.

4.3 Vehicle suspension with bushing elements

Two models of the vehicle are considered in the study: one with ideal kinematic joints in
the suspension systems and another using bushing joint in selected suspension elements.
The location of the bushing joints on the rear suspension system, referred to as RB3 , is
represented in Fig. 14.
The locations of the bushing joints for the front suspension are shown in Figs. 15b
and 15c. The constitutive relations used for the bushing joints model, obtained by a finite
element model analysis as described in Sect. 3, are presented in Fig. 7 for the revolution
bushing joints RB1 , RB2 , and RB3 .
Improved bushing models for general multibody systems and vehicle 361

Fig. 22 Lateral displacement for vehicle speeds of 60 km/h, 90 km/h, and 120 km/h

5 Application to ride and handling of a road vehicle

The first application scenario, represented in Fig. 16, considers the vehicle riding over ten
bumps, with heights of 0.1 m. This case excites the roll motion of the vehicle chassis, making
possible an evaluation of the suspension efficiency to reduce this chassis motion. Forward
vehicle speeds of 60, 90, and 120 km/h are used to study the case.
The dynamic behavior of the vehicle models for this scenario are characterized by its
vertical position and by the roll accelerations, depicted in Figs. 17 and 18.
The results presented in Fig. 17 for the vertical displacement of the center of mass of
the chassis show that the vertical displacement is larger for the forward vehicle velocity of
120 km/h and smaller for the velocity of 60 km/h, as expected. The bushing joints present
a relatively small influence on the vehicle dynamics for this scenario. The roll acceleration
responses, depicted in Fig. 18, also show slightly different dynamic behaviors for the two
vehicle models. However, it is observed that the model with bushing joints presents the
overall acceleration peak values smaller than those for the rigid model.
The deformations in a bushing joint, for the scenario studied for a vehicle forward ve-
locity of 90 km/h, is presented in Fig. 19 for bushing joint RB1 . The first peak shown in
any of the responses presented in Fig. 19 must be disregarded because it is associated to the
362 J. Ambrósio, P. Verissimo

Fig. 23 Lateral acceleration for vehicle speeds of 60 km/h, 90 km/h, and 120 km/h

initial conditions of the simulation only. The displacement peaks that take place between
0.25 s and 0.8 s are related to wheels traveling over the bumps. The normal deformations
reach 2.13 mm, while the tangential deformations reach 1.34 mm, and the larger angular
displacement value is 1.34◦ .
The second scenario addresses the study of the cornering ability of a vehicle. The two
vehicle multibody models are subjected to quick steering inputs allowing for the evaluation
of the lateral load transfer and to understand the effect of the bushing joints in the cornering
ability of the vehicle. This scenario is an obstacle evasion maneuver, i.e., a quick change
of lane, as showed in Fig. 20. The steering angles for the simulation of the vehicle in this
scenario, shown in Fig. 21, are characterized by quickly varying inputs leading the vehicle
to the neutral steering limit. These angles variations are found by performing several simu-
lations for each forward vehicle speed. Moreover, a traction moment of 15 N m is applied
to the front wheels of the vehicle during the initial first 0.6 seconds of the simulation to
maintain the forward speed.
In Figs. 22 and 23 the results for the lateral displacement and acceleration of the vehicle
center of mass are presented. In obstacle avoidance maneuvers the transfer of loads from
the left to the right wheels produces lateral accelerations that raise the lateral forces in the
outside tires to their maximum values, leading to oversteer or understeer, which put the
vehicle in its handling limit. The responses presented in Fig. 22 show that the vehicle lateral
Improved bushing models for general multibody systems and vehicle 363

Fig. 24 Simulation results for

the two vehicle models with a
forward speed of 90 km/h:
(a) vehicle multibody model with
ideal joints; (b) model with
bushing joints

motion is different for the models with bushing and ideal joints. The information shown in
the responses depicted by Fig. 22 and the sequence of positions of the vehicle presented in
Fig. 24 reveal that the vehicle model with bushing joints has a tendency to oversteer, which
increases with the forward speed. The loss of handling capability demonstrated by the model
with bushing joints seems to be due to the change of the suspension geometry caused by the
deformation of the bushing joints.
Figure 25 shows the bushing RB1 deformations for the right and the left suspension
arm. The results presented are for the simulation of the vehicle with a forward velocity of
90 km/h. As in the case of the scenario of a ride over bumps scenario, the first peak of
the deformation that appears in Fig. 25 is due to the initial conditions of the simulation,
and it must be disregarded. The right bushing presents the higher deformations during the
first steering input for the lane changing maneuver, at about 0.5 s, while the left bushing
has higher deformations approximately at 1.1 s when the counter-steering input takes place.
This observation agrees with the results expected for this scenario.

6 Conclusions

A general methodology for the characterization of the bushing joints in multibody systems
has been presented here. Each of the bushing joints described uses, in its formulation, the
same algebraic quantities required to built the ideal kinematic joints of the same type. In
the process of developing the bushing joints in the form presented here, the formulations for
the equivalent ideal kinematic joints have also been presented with alternative approaches to
364 J. Ambrósio, P. Verissimo

Fig. 25 Deformations of bushing RB1 : (a) radial direction; (b) tangential direction; (c) angular direction

those commonly used in the literature. The bushing joint formulations use the same dataset
required for the ideal kinematic joints plus the constitutive relations that relate the relative
displacement between the bodies connected with forces and moments between them. These
relations are obtained, for each bushing type, by performing a static analysis of the defor-
mation of these elements in a commercial nonlinear finite element program that includes
appropriate material models for the elastomers. In the process of demonstrating the use of
the bushing joints in the model of a road vehicle and of its analysis for two different scenar-
ios, it was observed that not only the computational efficiency of the vehicle models using
bushing joints is superior to that of vehicle models using perfect kinematic joints but also
that the simulations of these complex models can be handled for longer periods of time
without numerical problems associated to constraint violations. Moreover, the use of bush-
ing joints is an effective way of dealing with redundant kinematic joints, such as those used
in parallel mechanisms.

References

1. Shiib, T., Suda, Y.: Evaluation of driver’s behavior with multibody-based driving simulator. Multibody
Syst. Dyn. 17(2–3), 195–208 (2007)
Improved bushing models for general multibody systems and vehicle 365

2. Mavros, G.: On the objective assessment and quantification of the transient-handling response of a vehi-
cle. Vehicle Syst. Dyn. 45(2), 93–112 (2007)
3. Park, J., Nikravesh, P.: Effect of steering-housing rubber bushings on the handling responses of a vehicle.
SAE International Congress & Exposition, Detroit, Michigan (1997)
4. Ahn, Y.-K., Kim, Y.-C., Yang, B.-S., Ahmadian, M., Ahn, K.-K., Morishita, S.: Optimal design of an
engine mount using an enhanced genetic algorithm with simplex method. Vehicle Syst. Dyn. 43(1), 57–
81 (2005)
5. Vahid, O., Khajepour, A., Ismail, F.: In situ body mount optimization: theory and experiment. Vehicle
Syst. Dyn. 45(12), 1131–1147 (2007)
6. Kim, S.-S., Jeong, W.: Subsystem synthesis method with approximate function approach for a real-time
multibody vehicle model. Multibody Syst. Dyn. 17(2–3), 141–156 (2007)
7. Polach, P., Hajžman, M.: Design of characteristics of air-pressure-controlled hydraulic shock absorbers
in an intercity bus. Multibody Syst. Dyn. 19(1–2), 73–90 (2008)
8. Docquier, N., Fisette, P., Jeanmart, H.: Multiphysic modelling of railway vehicles equipped with pneu-
matic suspensions. Vehicle Syst. Dyn. 45(6), 505–524 (2007)
9. Kim, J.H., Yang, J., Abdel-Malek, K.: A novel formulation for determining joint constraint loads during
optimal dynamic motion of redundant manipulators in DH representation. Multibody Syst. Dyn. 19(42),
427–451 (2008)
10. Kövecses, J., Angeles, J.: The stiffness matrix in elastically articulated rigid-body systems. Multibody
Syst. Dyn. 18(2), 169–184 (2007)
11. Flores, P., Ambrósio, J., Pimenta Claro, J., Lankarani, H.: Kinematics and Dynamics of Multibody Sys-
tems with Imperfect Joints: Models and Case Studies. Springer, Dordrecht (2008)
12. Flores, P., Ambrósio, J., Claro, J., Lankarani, H., Koshy, C.: Lubricated revolute joints in rigid multibody
systems. Nonlinear Dyn. (2008). doi:10.1007/s11071-008-9399-2
13. Erkaya, S., Uzmay, İ.: A neural-genetic (NN–GA) approach for optimizing mechanisms having joints
with clearance. Multibody Syst. Dyn. 20(1), 69–84 (2008)
14. Nikravesh, P.: Computer-Aided Analysis of Mechanical Systems. Prentice-Hall, Englewood Cliffs
(1988)
15. Kang, J., Yun, J., Lee, J., Tak, T.: Elastokinematic analysis and optimization of suspension compliance
characteristics. Society of Automative Engineers (1997)
16. Kadlowec, J., Wineman, A., Hulbert, G.: Elastomer bushing response: Experiments and finite element
modeling. Acta Mech. 163, 25–38 (2003)
17. Ledesma, R., Ma, Z., Hulbert, G., Wineman, A.: A nonlinear viscoelastic bushing element in multibody
dynamics. Comput. Mech. 17, 287–296 (1996)
18. Shampine, L., Gordon, M.: Computer Solution of Ordinary Differential Equations: The Initial Value
Problem. Freeman, San Francisco (1975)
19. Gear, C.W.: Numerical solution of differential-algebraic equations. IEEE Trans. Circuit Theory CT-18,
89–95 (1982)
20. Nikravesh, P.: Initial condition correction in multibody dynamics. Multibody Syst. Dyn. 18(1), 107–115
(2007)
21. Ambrósio, J., Neto, A.: Stabilization methods for the integration of DAE in the presence of redundant
constraints. Multibody Syst. Dyn. 10, 81–105 (2003)
22. ABAQUS User’s Manual. Pawtucket, MA, Hibbitt, Karlsson & Sorensen (1999)
23. Murray, D.: Inside Solidworks. OnWord Press, Clifton Park (2005)
24. Gim, G., Nikravesh, P.E.: An analytical model of pneumatic tyres for vehicle dynamic simulations. Part
3: Validation against experimental data. Int. J. Vehicle Des. 12(2), 217–228 (1991)
25. Verissimo, P.: Improved bushing models for vehicle dynamics. M.S. Thesis, Instituto Superior Técnico,
Technical University of Lisbon, Portugal (2007)