Mechanism and Machine Theory 37 (2002) 815832

www.elsevier.com/locate/mechmt

Synthesis and analysis of the ve-link rear suspension

system used in automobiles
P.A. Simionescu *, D. Beale
Department of Mechanical Engineering, Auburn University, 202 Ross Hall, Auburn, Al 36849, USA

Abstract
The paper deals with the optimum kinematic synthesis and analysis of the ve-link independent suspension system (also known as multilink suspension, mechanism commonly symbolized 5S5S). The
synthesis goal is fullling a minimum variation of the wheel track, toe angle and camber angle during
jounce and rebound of the wheel. Two solutions obtained by synthesis are analyzed and compared to an
existing solution, and the displacement, velocity and acceleration of the wheel carrier relative to the car
body are determined, together with the variation of the momentary screw axis and the rear axle roll-center
height. Both the kinematic synthesis and the analysis are performed in a simplied, easy to program
manner, using a ctitious mechanism that has all the links dismounted from their joints.

2002 Elsevier Science Ltd. All rights reserved.
Keywords: Rigid body guidance; Suspension linkage; Constraint equations; Optimization algorithm

1. Introduction
The ve-link suspension mechanism was rst introduced by Deimler-Benz on their W201 and
W124 series under the name multilink suspension (Fig. 1a). Ever since has been successfully
implemented both in independent suspension systems and in rear axle guiding mechanisms by
many automobile manufacturers. Due to the larger number of design parameters, it has the capability of fullling better the complex kinematic and dynamic requirements imposed on suspension systems of todays automobiles. It is however much more dicult to synthesize than any
other suspension mechanism, due to its general spatial conguration. In case of multilink front
suspension, the design problem is even more complex due to the fact that the kingpin is a virtual

Corresponding author. Tel.: +1-334-844-5867; fax: +1-334-844-5865.

E-mail addresses: pasimi@eng.auburn.edu (P.A. Simionescu), dbeale@eng.auburn.edu (D. Beale).

0094-114X/02/$ - see front matter
2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 4 - 1 1 4 X ( 0 2 ) 0 0 0 3 7 - X

816

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

Fig. 1. Multilink independent suspension mechanism (a) and its kinematic diagram (b).

one [1,2] corresponding to the momentary screw axis of the wheel carrier performing the steering
motion relative to the chassis.
The rear independent wheel or axle guiding mechanism(s) are, in the sense of mechanism
theory, spatial motion generators (also known as rigid body guidance mechanisms). Research on
motion generators synthesis and analysis has been carried out on both abstract and applied
mechanisms by many researchers in the past. A general formulation of the mechanism synthesis
problem for path, function and rigid body guidance based on optimization techniques is proposed
by Aviles et al. [3]. According to the authors, a global error function to be minimized is dened as
a weighted-sum of some local error functions, previously minimized with respect to the Cartesian
coordinates of the basic points of the mechanism. The so-called basic points are the centers of
the joints and the points of the links required to generate certain paths throughout the working
range of the mechanism. Although the method is general in its formulation, the main disadvantage lays in the large number of variables required to dene the objective function, as well as in
not including the ground joint coordinates among the design parameters.
An extension of the approach of Aviles et al. to the synthesis of spatial linkages is given by
Jimenez et al. [4]. However, the main drawback of an excessive number of design variables required in formulating the synthesis problem was maintained. In the example presented of synthesizing a 5S5S suspension mechanism for only three prescribed positions of the wheel carrier,
an objective function of 64 variables was dened.
Suh [5] synthesized a double-wishbone suspension mechanism (an RSSR-SS spatial motion
generator), which can be considered a particular embodiment of the ve-link suspension [1], in a
mixed approach, with two nite and instantaneous exact positions, using displacement matrices
and constraint equations.
A combination of exact and approximate synthesis is performed by Sandor et al. [6] for the
same RSSR-SS motion generating mechanism. The authors considered part of the positions
imposed to the wheel carrier as exact positions and solved the corresponding set of equations. The
free choices in these equations were further considered design variables in an objective function,
penalized with the conditions of avoiding branching, achieving correct sequence of prescribed
positions and observing the shortest and longest links to be within prescribed limits in the remaining imposed positions.

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

817

The method detailed in the present paper for kinematic synthesis of the 5S5S rigid body
guidance mechanism considers the guided body (the wheel carrier) released from its joints and
moving in successive positions along the ideal trajectory. The synthesis problem thus becomes that
of nding the joint disposition for which the distances between the homologous released joints
(the pair joints that in the real mechanism are connected by binary links) vary as little as possible.
Essentially this is a numerical implementation of the nite-position spatial theory of kinematic
synthesis the object of which is determining those points which lie on special loci: spheres, cylinders, circles, lines, etc. [79].
The same approach of considering the wheel carrier released from its joints was applied for
displacement analysis of the same mechanism. The interested reader can nd this procedure directly applicable to solving the direct kinematic problem of a variety of parallel mechanisms of the
GoughStewart type.

2. Synthesis problem formulation

The requirements upon the motion of the rear wheel that can be transposed into kinematic
conditions when synthesizing the suspension mechanism are [10]:
minimum toe angle variation during compression and rebound;
avoid excessive outward camber thrust on corners;
avoid excessive sideways thrust and consequent rear end steering impulses on single wheel
bump or rebounds;
supplementary, the suspension elements must ensure a minimum intrusion into the passengers
and luggage accommodation, a condition that can be translated into constraints imposed to the
possible disposition of the ball joints on the chassis and on the wheel carrier.
The eect of the compliance of the rear wheel suspension upon the car ride behavior is important and in the nal design must necessarily be considered by performing a dynamic analysis
using a multibody simulation software. However, in order to simplify both the kinematic synthesis
and analysis procedures, it is common in the early stages of design to assume that the joints have
neither clearances nor elasticities, and the vehicle chassis and suspension elements are rigid. When
equipped with compliant joints, it is to be expected that a rigid joint suspension that exhibit good
kinematic characteristics, will continue to perform satisfactory (provided that the stiness rates of
the joints are properly selected).
Taking the rst three above-mentioned conditions imposed to a suspension system, it can be
considered that the ideal wheel movement along its operation travel must be close to a vertical
translation relative to the car body. This is in accordance with Raghavans ndings [11] that for
straight-line motion of the car, the motion of the wheel relative to the road should exhibit zero toe
and camber change, and that track width should be maintained constant. The same author
concluded that while turning, the wheels should remain at zero camber or should camber into the
turn if possible, and the track should stay constant.
In order to formulate the synthesis problem, all the ve links are removed from their joints (or
assumed of variable length), thus allowing the wheel carrier to be displaced in successive positions

818

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

along any trajectory. If the distance between the homologous joints varies very little in these
successive positions, the real mechanism with the ve links jointed back in place will guide the
wheel very close to the imposed path.
The above considerations are the basis for formulating the synthesis of the ve-link mechanism
as an optimization problem, i.e. of nding the minimum of the following objective function of 30
variables (Fig. 1b):
0
; z0Bi i1...5
F xAi ; yAi ; zAi ; x0Bi ; yBi

5 X
n h
X
i1

li  Ai Bi j

i2

j1

with j 1 . . . n intermediate positions of the wheel carrier evenly spaced on the prescribed trajectory.
In the followings this imposed trajectory will be a simple vertical translation of the wheel carrier
i.e. xN and yN are kept constant for zNj varying between a lower zN min and an upper zN max limit of
point N attached to the wheel carrier. One should not expect that the mechanism obtained by
synthesis to exactly generate this pure vertical motion. As will be seen later, the kinematic behavior of the synthesized mechanism strongly depends on the values chosen for zN min and zN max
(which should not necessarily be the upper and lower limits of wheel travel during jounce and
rebound, nor even belong to the actual motion range of the wheel carrier).
The reference lengths of the links noted li i 1; 5 in relation (1) are determined as the distances between the joints Ai and Bi for the wheel in its initial position, corresponding to the car
averagely loaded and in rest. The variable distances Ai Bi j between the ve homologous joints Ai
and Bi in a current position j of the wheel carrier is given by
Ai Bi j

q
xAij  xBij 2 yAij  yBij 2 zAij  zBij 2

where the coordinates x, y and z must be specied relative to the same reference frame, preferable
the xed reference frame Oxyz. Because the disposition of the ball-joint centers Bi is given in the
reference frame attached to the wheel carrier Nx0 y 0 z0 , the following transformations must be applied in order to make use of Eq. (2):
2

3
2 0 3
2
3
xBi
xNj
xBij
0 5
4 yBij 5
4 yBi
4 yNj 5
0
zBij Oxyz
zBi Nx0 y 0 z0
zNj Oxyz

In the initial position, the reference frame Oxyz attached to the chassis and the frame Nx0 y 0 z0
attached to the wheel carrier are considered parallel. Knowing the coordinates (xN0 , yN0 , zN0 ) of
the origin of Nx0 y 0 z0 frame relative to the chassis reference frame, the coordinates of the same point
N relative to Oxyz reference frame will be (xN0 , yN0 , zNj ) for a current prescribed position j, where
zNj zN min DzNj with DzNj jzN max  zN min =n.
The limitations upon the possible locations of the ball joints on the chassis and wheel carrier
can be prescribed as side constraints of the form:

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

819

xAi min 6 xAi 6 xAi max

yAi min 6 yAi 6 yAi max

i 1 . . . 5

i 1 . . . 5

zAi min 6 zAi 6 zAi max

and
x0Bi min 6 x0Bi 6 x0Bi max
0
0
0
yBi
min 6 yBi 6 yBi max

z0Bi min

6 z0Bi

6 z0Bi max

and must necessarily be imposed in order to avoid convergence to unpractical solutions with links
excessively long.
The objective function (1) together with the constraints (4) and (5) can be minimized using a
proper optimization subroutine. Of the maximum number of design variables (30 in total
irrespective of the number of intermediate positions n of the wheel carrier), some of the ball-joint
centers can be imposed xed values and the number of design variables further reduced.
In theory it is possible to prescribe a trajectory to the wheel carrier that can be exactly generated
by a real mechanism (case in which the global minima of the objective function F will be zero). In
practice however, there will always be a departure between the prescribed motion and the actual
motion of the real mechanism. Therefore a kinematic analysis is required in order to determine the
actual behavior of the suspension mechanism obtained by synthesis.

3. Kinematic analysis of the ve-link suspension mechanism

The analysis of the ve-link suspension mechanism has been tackled by a number of researchers
in the past. Lee et al. [1] derived the velocity equations of the wheel carrier and applied a step-wise
linearization to solve the position problem. Mohamed and Attia [12] used the constrained
equations obtained from the condition that the ve connecting rods and the wheel carrier are rigid
bodies. Knapzyk and Dzierzec [13] considered a modied mechanism with two of the guiding
links disassembled and solved an optimization problem describing the condition that the distances
between the homologous released joints remain equal to the lengths of the disconnected members.
Following [6], Unkoo and Byeongeui [2] used 4  4 displacement and dierential-displacement
matrices and constraint equations to solve the position and velocity problem of ve-link and struttype multilink suspensions. The referred authors also determined the imaginary kingpin axis of
these suspensions systems using screw-axis theory and compared the results with those obtained
by nite-center analysis.
The same approach of considering all the ve connecting rods removed will be further considered. For successive values of the input parameter zN , the position of the point N relative to the
horizontal axis and the orientation angles of the wheel carrier will be tuned in a searching process,
until the distances between the released joints Ai and Bi become equal (within some error limits) to
the lengths of the respective links Ai Bi .

820

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

3.1. Position problem

The ve-link suspension mechanism has six degrees-of-freedom, of which ve are trivial rotations of the connecting links around their own axes. Correspondingly, the position of the wheel
carrier can be specied using only one independent parameter viz the coordinate zN of the origin
of the Nx0 y 0 z0 reference frame relative to the central reference frame Oxyz. The remaining ve
parameters: coordinates xN , yN and angles a, b and c that dene the position and orientation of the
wheel carrier can be determined by solving the following equations of constraint:
xAi  xBi 2 yAi  yBi 2 zAi  zBi 2 l2i

i 1 . . . 5

describing the condition of the distance between joints Ai and Bi to remain constant during the
working range of the mechanism. In the above Eq. (6), the coordinates xBi , yBi and zBi are determined by applying the following transformation to the Nx0 y 0 z0 reference frame:
2 3
2 0 3
2 3
xBi
xN
xBi
0 5
4 yBi 5
Rbac 4 yBi
4 yN 5
7
z0Bi Nx0 y 0 z0
zBi Oxyz
zN Oxyz
where Rbac is the transformation matrix that express the successive rotation of the wheel carrier
relative to Oxyz by the pitch angle b, yaw angle a and roll angle c [14]:
2
3
ca
cb
sa
ca
sb
8
Rbac Rc;x Ra;z Rb;y 4 sa
cb
cc sb
sc ca
cc sa
sb
cc cb
sc 5
sa
cb
sc sb
cc ca
sc sa
sb
sc cb
cc
In the above equation Ra;z , Rb;y and Rc;x are the basic rotation matrices while ca cos a,
sa sin a and so forth.
For a given value of the independent parameter zN , the system of Eq. (6) in the unknowns a, b,
c, xN and yN can be very conveniently solved by minimizing the following objective function:
F0 a; b; c; xN ; yN

5
X
2
2
2
xAi  xBi yAi  yBi zAi  zBi

i1

In order to facilitate convergence, the starting point when minimizing F0 can be taken the position
of the wheel carrier (the same xN , yN and orientation angles a, b, c) imposed during synthesis for
the same zNj . Once the displacement problem of the wheel carrier is solved, the diagram of the
wheel track, recessional wheel motion, camber and toe angle alteration can be generated.
3.2. Linear velocity and acceleration analysis
The velocities of points Bi on the wheel carrier can be determined by dierentiating once with
respect to time the equations of constraint (6). The number of unknowns thus emerging is 15, and
therefore 10 more equations must be added, like the time derivatives of following equations:
2

xBj  xBk yBj  yBk zBj  zBk const

j 1 . . . 4 and k j 1 . . . 5

10

Table 1
The coecients of the linear system of equations used to determine the linear velocity of points Bi (i 1 . . . 5)
y_ B1

z_ B1

x_ B2

y_ B2

z_ B2

x_ B3

y_ B3

z_ B3

x_ B4

y_ B4

z_ B4

x_ B5

y_ B5

z_ B5

x_ N

y_ N

xA1  xB1
0

yA1  yB1
0

zA1  zB1
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
0

0
yA2 
yB2
0

0
zA2 
zB2

0
xA2 
xB2
0

zA3 
zB3
0

yA3 
yB3
0

xA3 
xB3
0

zA4 
zB4
0

yA4 
yB4
0

xA4 
xB4
0

zB1  zB2

zB1  zB3

zB2 
zB1
0

yB1  yB3

yB2 
yB1
0

xB1  xB3

xB2 
xB1
0

zA5 
zB5
0

yB1  yB2

yA5 
yB5
0

xB1  xB2

xA5 
xB5
0

zB1  zB4

zB3 
zB1
0

yB1  yB4

yB3 
yB1
0

xB1  xB4

xB3 
xB1
0

zB1  zB5

zB4 
zB1
0

yB1  yB5

yB4 
yB1
0

xB1  xB5

xB4 
xB1
0

xB3 
xB2
0

yB3 
yB2
0

zB3 
zB2
0

zB4 
zB2
0

yB4 
yB2
0

xB4 
xB2
0

zB5 
zB2
0

zB1  zN

yB5 
yB2
0

yB1  yN

xB5 
xB2
0

xB1  xN

zB2 
zB3
zB2 
zB4
zB2 
zB5
0

yB2 
yB3
yB2 
yB4
yB2 
yB5
0

xB2 
xB3
xB2 
xB4
xB2 
xB5
0

zB5 
zB1
0

yB5 
yB1
0

xB5 
xB1
0

yB2 
yN
0

zB2 
zN

xB2 
xN
0

zB3 
zN
0

yB3 
yN
0

xB3 
xN
0

zB4 
zN
0

yB4 
yN
0

xB4 
xN
0

xB5 
xN

yB5 
yN

zB5 
zN

xN 
xB1
xN 
xB2
xN 
xB3
xN 
xB4
xN 
xB5

yN 
yB1
yN 
yB2
yN 
yB3
yN 
yB4
yN 
yB5

zB1 
zN _zN
zB2 
zN _zN
zB3 
zN _zN
zB4 
zN _zN
zB5 
zN _zN

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

x_ B1

821

822

xB1

yB1

zB1

xB2

yB2

zB2

xB3

yB3

zB3

xB4

yB4

zB4

xB5

yB5

zB5

x_ N

y_ N

*
0
0
0
0
*
*
*
*
0
0
0
*
0
0
0
0

*
0
0
0
0
*
*
*
*
0
0
0
*
0
0
0
0

*
0
0
0
0
*
*
*
*
0
0
0
*
0
0
0
0

0
*
0
0
0
*
0
0
0
*
*
*
0
*
0
0
0

0
*
0
0
0
*
0
0
0
*
*
*
0
*
0
0
0

0
*
0
0
0
*
0
0
0
*
*
*
0
*
0
0
0

0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0
0

0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0
0

0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0
0

0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0

0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0

0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*
0

0
0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*

0
0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*

0
0
0
0
*
0
0
0
*
0
0
*
0
0
0
0
*

0
0
0
0
0
0
0
0
0
0
0
0
*
*
*
*
*

0
0
0
0
0
0
0
0
0
0
0
0
*
*
*
*
*

2
x_ 2B1 y_ B1
z_ 2B1
2
2
x_ B2 y_ B2 z_ 2B2
2
x_ 2B3 y_ B3
z_ 2B3
2
_x2B4 y_ B4
z_ 2B4
2
x_ 2B5 y_ B5
z_ 2B5
_xB1  x_ B2 2  y_ B1  y_ B2 2  _zB1  z_ B2 2
_xB1  x_ B3 2  y_ B1  y_ B3 2  _zB1  z_ B3 2
_xB1  x_ B4 2  y_ B1  y_ B4 2  _zB1  z_ B4 2
_xB1  x_ B5 2  y_ B1  y_ B5 2  _zB1  z_ B5 2
_xB2  x_ B3 2  y_ B2  y_ B3 2  _zB2  z_ B3 2
_xB2  x_ B4 2  y_ B2  y_ B4 2  _zB2  z_ B4 2
_xB2  x_ B5 2  y_ B2  y_ B5 2  _zB2  z_ B5 2
zN zB1  zN  _xB1  x_ N 2  y_ B1  y_ N 2  _zB1  z_ N 2
zN zB2  zN  _xB2  x_ N 2  y_ B2  y_ N 2  _zB2  z_ N 2
zN zB3  zN  _xB3  x_ N 2  y_ B3  y_ N 2  _zB3  z_ N 2
zN zB4  zN  _xB4  x_ N 2  y_ B4  y_ N 2  _zB4  z_ N 2
zN zB5  zN  _xB5  x_ N 2  y_ B5  y_ N 2  _zB5  z_ N 2

The stars in the table designate coecients identical to the corresponding ones in Table 1.

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

Table 2
The coecients of the linear system of equations used to determine the linear accelerations of points Bi (i 1 . . . 5)a

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

823

and
xBi  xN 2 yBi  yN 2 zBi  zN 2 const i 1 . . . 5

11

describing the condition that the wheel carrier is a rigid body. By dierentiation these equations
once with respect to time, a new independent parameter z_ N will emerge, which, the same as zN
must be specied as input during the numerical analysis. The coecients of the system of linear
equations in the 17 unknowns x_ Bi , y_ Bi , z_ Bi (i 1 . . . 5), x_ N and y_ N used for velocity analysis are
summarized in Table 1.
By dierentiating with respect to time the equations used to solve the velocity problem, a
second system of linear equations in the unknowns xBi , yBi , zBi (i 1 . . . 5), xN and yN will be
further obtained, the coecients of which are given in Table 2. In this case zN , z_ N and zN will be
the independent parameters that must be specied as inputs.
3.3. Angular velocity and acceleration analysis
The components of the angular-velocity vector (xx , xy , xz ) relative to the xed reference frame
Oxyz can be determined using the following equation known from the rigid body kinematics:
2 3 2 3 2 3 2
3
xx
xBi  xN
x_ Bi
x_ N
4 y_ Bi 5 4 y_ N 5 4 xy 5  4 yBi  yN 5
12
z_ Bi
z_ N
xz
zBi  zN
written for any two dierent points of the wheel carrier for which the linear velocity are known.
The expressions of xx , xy and xz as derived from relation (12) are given in Eq.(A.1) in Appendix
A.
The components of the angular-acceleration vector (ex , ey , ez ) can be determined writing the
following equation, the same for two dierent points of the wheel carrier:
2 3 2 3 2 3 2
3 2 3 02 3 2
31
xBi
xN
ex
xBi  xN
xx
xx
xBi  xN
4 yBi 5 4 yN 5 4 ex 5  4 yBi  yN 5 4 xx 5  @4 xx 5  4 yBi  yN 5A
13
zBi
zN
ex
zBi  zN
xx
xx
zBi  zN
The expressions of ex , ey and ez derived through analytical manipulations of relation (13) are
summarized in Eq.(A.2) in Appendix A. Alternatively, the components of the angular acceleration
can be determined by dierentiating once with respect to time the components of the angular
velocity:
ex x_ x ; ey x_ y

and ez x_ z

14

The results of velocity and acceleration analysis will be further used in determining the location of
the instantaneous screw axis of the wheel carrier moving relative to the chassis. Position, velocity
and acceleration problems are also stages required in solving the dynamic problem of the suspension system. According to Hiller [15], of the total CPU time needed to simulate the response of
a ve-link suspension to a road input, almost 70% was required for solving the kinematics of the
system. The method described above for solving the position problem it is likely to reduce this
amount of time, since requires solving a system of only ve nonlinear equations.

824

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

3.4. Instantaneous screw axis

Considering the instantaneous motion of the ve-link suspension, the wheel carrier motion
relative to the car body is a screw motion of the circle-point-surface xed to the wheel carrier with
respect to the center-point-surface xed to the car body [16]. The common tangent of these two
surfaces is the instantaneous screw axis of the spatial motion (see Fig. 2), and corresponds to the
points of minimum velocity of the wheel carrier relative to the car body. Therefore, the parameters
positioning the momentary screw axis can very well be determined by formulating a minimization
problem.
A dierent approach is to solve the system of equations expressing the condition the linear
velocity (_x, y_ , z_ ) of a point (x; y; z) attached to the wheel carrier is parallel to the angular-velocity
vector (xx , xy , xz ):
x_ =xx y_ =xy z_ =xz

15

Fig. 2. Center-point-surface and circle-point-surface of a ve-link independent suspension (solution 2 in paragraph 4)

in perspective view (a) and top view (b).

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

825

Based on Eq. (12), the above equalities becomes:

x_ N xy z  zN  xz y  yN y_ N xz x  xN  xx z  zN

xx
xy
_xN xy z  zN  xz y  yN z_ N xz x  xN  xx z  zN

xx
xz

16

The resulting expressions of the parametric equation of the momentary screw axis xy and zy
are given in Eq.(A.6) in Appendix A.
Determining of the screw axis due to steering input is of much signicant importance in the case
of the ve-link suspension mechanism used for guiding the front wheels, which has 2 DOF corresponding to the steering input. When only the steering input is active, the resulting momentary
screw axis (which can be calculated following a similar approach) will be the virtual kingpin of the
wheel during the steering motion [2].
3.5. Suspension roll center
Each suspension has a roll center dened as the point in the vertical plane through the wheel
centers at which lateral forces may be applied to the sprung mass without producing suspension
roll [17,18]. According to Reimpell and Stoll [19] there is a direct correlation between the wheel
track variation and the roll-center height hR . According to the same authors, this correlation is
also conicting, in that a high roll center (which is desirable for a favorable car body attitude
during cornering) implies a larger track alteration. The suspension roll center can be approximately determine by nite-center analysis as the intersection between the normal to the trajectory
of the path center point S projected on the vertical plane Oxz and the cars longitudinal plane Oyz
(Fig. 3). The following formula:

Fig. 3. Schematic for calculating the roll-center height of the rear axle.

826

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

x2S zNj  x2S zNj1 z2S zNj  z2Sj zNj1

hR zNj 0:5
zS zNj  zSj zNj

17

has been derived for calculating the roll-center height relative to the chassis reference frame. The
height of the roll-center measured from the ground will be
hR zNj h0R zNj  x2S zNj

18

In the above equations zNj and zNj1 are two successive positions of the wheel center, suciently
close one to the other to allow a tangent-chord approximation along the trajectory of the path
center point.
4. Numerical results
Based on the procedure described above, the synthesis of a ve-link rear wheel independent
suspension system was performed. The numerical data corresponding to the Mercedes-190 multilink suspension available [13] was used in dening the allowable positions of the ball-joint
centers (Table 3).
The origin of the Nx0 y 0 z0 coordinate system in the reference position was xN0 705 mm, yN0 0
and zN0 302 mm, while the wheel radius was R 314 mm.
Two intervals of the wheel carrier vertical travel have been considered in the objective function
F. The rst numerical solution recorded (Table 4) was obtained for the reference frame Nx0 y 0 z0
translating verticaly between zN min 50 mm and zN max 100 mm. The second solution (Table
Table 3
Searching domains of the design variables xAi , yAi , zAi , xBi , yBi , zBi (i 1 . . . 5)
190 6 xA1 6 220
481 6 xA2 6 511
389 6 xA3 6 419
422 6 xA4 6 452
341 6 xA5 6 371
53 6 x0B1 6  33
83 6 x0B2 6  63
49 6 x0B3 6  29
53 6 x0B4 6  33
83 6 x0B5 6  63

87 6 yA1 6 117
336 6 yA2 6  306
224 6 yA3 6  194
224 6 yA4 6  194
10 6 yA5 6 20
0
33 6 yB1
6 53
0
54 6 yB2
6  34
0
151 6 yB3
6  131
0
88 6 yB4
6  68
0
5 6 yB5
6 15

216 6 zA1 6 246

236 6 zA2 6 266
281 6 zA3 6 311
387 6 zA4 6 417
401 6 zA5 6 431
104 6 z0B1 6  84
149 6 z0B2 6  129
43 6 z0B3 6  23
87 6 z0B4 6 105
115 6 z0B5 6 135

Table 4
Numerical solution obtained for 50 mm 6 zN 6 100 mm in the objective function F (variant 1)
xAi
yAi
zAi
x0Bi
0
yBi
0
zBi
li

i1

i2

i3

i4

i5

190.436
87.591
238.816
33.737
43.949
90.997
483.584

482.605
317.292
236.036
63.000
36.344
129.034
329.115

401.068
210.635
289.298
31.195
135.524
43.000
284.509

422.000
198.545
410.077
46.577
78.843
87.000
265.835

344.310
3.447
430.258
67.136
3.353
115.002
293.853

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

827

Table 5
Numerical solution obtained for 3000 mm 6 zN 6 3100 mm in the objective function F (variant 2)
xAi
yAi
zAi
x0Bi
0
yBi
0
zBi
li

i1

i2

i3

i4

i5

203.760
111.186
243.559
33.000
37.798
100.101
475.783

493.038
330.546
261.996
69.814
34.000
148.828
346.393

390.597
197.256
308.938
29.608
133.477
30.041
294.182

422.066
211.100
392.424
35.783
82.386
87.057
278.679

348.449
20.000
426.300
72.934
15.000
115.226
283.807

5) was obtained for zN min 3000 mm and zN max 3100 mm. This unusual domain of zN facilitated obtaining a kinematic solution that ensures a higher location of the suspension roll center
relative to the ground. According to [19], both rear and front suspension roll centers should be as
high as possible and at approximately the same height. However, limitations imposed to the wheel
track alteration restrict choosing a rear suspension that ensures a roll center located to high.
The two solutions obtained by synthesis noted 1 and 2, were compared with an existing solution
noted 0 also available in [13]. The wheel track variation:
DSzN xS zN0  xS zN

19

is visibly improved in case of solution 1. Also improved is the camber angle alteration Dd as
compared to the existing solution 0 (Fig. 4).
In Fig. 5 is given the diagram of the recessional motion of the wheel as the variation of yS
coordinate of the center path S.

Fig. 4. Wheel track alteration (a) and recessional wheel motion (b) during jounce and rebound for an initial solution 0,
and the two solutions obtained by synthesis, 1 and 2.

828

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

Fig. 5. Camber alteration DdDzN (a) and toe angle alteration DuDzN (b) of the wheel relative to the chassis during
jounce and rebound, for the same numerical variants in Fig. 4.

DY zN yS zN0  yS zN

20

This parameter describes the fore and aft motion of the wheel during jounce and rebound.
However, since it occurs along the direction of car travel, is has a smaller eect upon the car
dynamics than the wheel track alteration.
The camber angle variation Dd was determined as the projection of the angle between the axes
Oz and Nz0 on the vertical transverse plane (Fig. 6a).
Similarly, the toe angle alteration Du shown in Fig. 7b was determined as the angle between the
axes Ox and Nx0 projected on the horizontal plane. In this case, for 150 mm 6 DzN 6 150 mm the
toe angle of solution 1 is slightly larger than that of the existing solution 0, being however
compensated by the understeer eect of track widening during jounce.
For illustrative purposes, the diagrams of the magnitude of the angular velocity x and angular
acceleration e of the wheel carrier have been plotted (Fig. 6) for z_ N 1:0 m/s and zN 0 using
Eqs.(A.1) and (A.2) in Appendix A.
The results of the kinematic analysis have been also used in the 3D visualization of the motion
of the mechanism. Fig. 7 shows superimposed positions of the suspension mechanism solution 1,
corresponding to zN0 and zN0  150 mm, viewed from the front (a) and from above (b).
The circle-point-surface and the center-point-surface in Fig. 3 were produced for solution 2.
They have been generated as ruled surfaces of the momentary screw axis relative to the chassis

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

829

Fig. 6. Variation of the angular velocity x (a) and angular acceleration e (b) for z_ N 1:0 m/s and zN 0 for the same
variants in Fig. 4.

Fig. 7. Superimposed positions of the suspension mechanism solution 1, corresponding to DzN 0 and DzN 150
mm, viewed from the rear (a) and from above (b).

(the circle-point-surface) and to the wheel carrier (the center-point-surface). The inclined position
of the screw axis relative to cars longitudinal axis it is due to the wheel carrier rotation around its
own axis, which for solution 2 corresponds to a maximum angle c of 16.2 occurring for
zN0 150 mm.
Finally, the plot in Fig. 8 of the alteration of the roll-center height with DzN have been produced. As compared to the existing solution, both variant 1 and 2 have a favorable smaller drop
of the roll center under load. According to [19], in case of the real vehicle with compliant suspension, the roll center will be higher than for the simplied mechanism with rigid joints.

830

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

Fig. 8. Variation of the suspension roll-center height, measured relative to the car reference frame (a) and relative to the
ground (b). In the reference position (DzN 0), hR 138:6 mm for variant 0, hR 73:2 mm for variant 1 and
hR 150:4 mm for variant 2.

5. Conclusions
A classic rigid body guidance mechanism synthesis problem was presented, that of designing a
ve-link independent rear suspension system under the condition of ensuring a proper motion of
the wheel carrier. Also given were complete kinematic analysis equations that allow determining
the wheel recession, wheel track, toe angle, camber angle and roll-center height variation together
with the linear and angular velocities and accelerations of the wheel carrier of a given ve-link
suspension system. Two variants obtained by synthesis were analyzed and compared to an existing solution of a Mercedes-190 suspension system. Though the characteristics of the same
mechanisms equipped with compliant joints will dier, the good behavior of the rigid joint
mechanisms obtain by synthesis are likely to be preserved.
Both the synthesis and the analysis procedures advanced in the paper can be extended to designing and simulating other suspension systems. For example the RSSR-SS double-wishbone
suspension can be synthesized in the same manner. Multilink suspensions used for front wheels of
passenger cars (that have a second DOF needed for wheel steering) can also be synthesized following a similar approach, as well as the 5S5S mechanisms used in guiding rigid axles. In this
case however, the eect of joint elasticities must necessary be assessed using a multibody simulation software, since they have an essential contribution to the combined translationrotation
motion of the real axle.
Acknowledgements
The authors would like to thank Dr. Madhu Raghavan as well as to the anonymous reviewers
for their comments and suggestions.

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

831

Appendix A
Considering two points Bj and Bk (j 6 k), the components of the angular velocity of the wheel
carrier are
xx

P 1
Dxj
Dxk P 2
Dyj
Dxk P 3
Dxj
Dzj
Dxj
Dzj
Dyk  Dyj
Dzj
Dxk

xy Dyk
xx  P 3=Dxk

A:1

xz Dzj
xy  P 1=Dyj
while the components of the angular acceleration are
ex

Q1
Dxj
Dxk Q2
Dyj
Dxk Q3
Dxj
Dzj
Dxj
Dzj
Dyk  Dyj
Dzj
Dxk

ey Dyk
ex  Q3=Dxk

A:2

ez Dzj
ey  Q1=Dyj
with
Dxj xBj  xN ;

Dxk xBk  xN

Dyj yBj  yN ;

Dyk yBk  yN

Dzj zBj  zN ;

Dzk zBk  zN

P 1 x_ Bj  x_ N ; P 2 y_ Bj  y_ N ; P 3 z_ Bk  z_ N ;


Q1 x2y x2z
Dxj  xx xy
Dyj xz
Dzj xBj  xN


Q2 x2x x2z
Dyj  xy xx
Dxj xz
Dzj yBj  yN


Q3 x2x x2y
Dzk  xz xx
Dxk xy
Dyk zBk  zN
The parametric equations of the screw axis of the wheel carrier are
h
i.

xy T1
xy
xz T2 x2x x2y
xx
xy
x2


zy  T1
xy T2
xz xy
x2

A:3
A:4

A:5

A:6

where
T1 xy
x_ N  xx
y_ N xx
xz
xN xy
xz yN  yC  zN x2x x2y
T2 xx
z_ N  xz
x_ N xy
xz
zN xx
xy
xN yC  yN x2x x2z

A:7

References
[1] D.M.A. Lee, D.M. Pascoe, W.H. El Maragy, An analysis of multilink independent suspension systems,
International Journal of Vehicle Design 14 (1993) 4458.
[2] L. Unkoo, A. Byeongeui, Method to analyze the imaginary kingpin axis in multi-link type suspension systems, SAE
paper 930262, 1993, pp. 1126.

832

P.A. Simionescu, D. Beale / Mechanism and Machine Theory 37 (2002) 815832

[3] R. Aviles, M.B. Ajuria, J. Garcia de Jalon, A fairly general method for optimum synthesis of planar mechanisms,
Mechanism and Machine Theory 20 (1985) 321328.
[4] J.M. Jimenez, G. Alvarez, J. Cardenal, J. Cuadrado, A simple and general method for kinematic synthesis of
spatial mechanisms, Mechanism and Machine Theory 32 (1997) 323341.
[5] C.H. Suh, Synthesis and analysis of suspension mechanisms with use of displacement matrices, SAE paper 890098,
1989, pp. 189200.
[6] G.N. Sandor, L.J. Xu, S.P. Yang, Computer aided synthesis of two-closed loop RSSR-SS spatial motion
generators with branching and sequence constraints, Mechanism and Machine Theory 21 (1986) 345350.
[7] B. Roth, The kinematics of motion through nitely separated positions, Journal of Applied Mechanics 34 (1967)
591598.
[8] B. Roth, Finite-position theory applied to mechanism synthesis, Journal of Applied Mechanics 34 (1967) 599605.
[9] Y.L. Sarkisyan, K.C. Gupta, B. Roth, Spatial least square approximation of motion, IFToMM International
Symposium on Linkages and Computer Design Methods, Bucharest, Romania B (1973) 512521.
[10] D. Bastow, Aspects of car rear suspension, Proceedings of the Institution of Mechanical Engineers 190 (53) (1976)
611626.
[11] M. Raghavan, Suspension kinematic structure for passive control of vehicle attitude, International Journal of
Vehicle Design 12 (1991) 525547.
[12] M.G. Mohamed, H.A. Attia, Numerical kinematic analysis of the motor-vehicle multilink suspension system,
ASME Design Engineering Technical Conferences 82 (1995) 883889.
[13] J. Knapczyk, S. Dzierzek, Displacement and force analysis of ve-rod suspension with exible joints, Journal of
Mechanical Design 117 (1995) 532538.
[14] C.H. Suh, C.W. Radclie, Kinematics and Mechanisms Design, Wiley, 1978.
[15] M. Hiller, Five-link suspension, in: W. Kort
um, R.S. Sharp (Eds.), Multibody Computer Codes in Vehicle System
Dynamics, Swets & Zeitlinger, Amsterdam, 1993, pp. 254262.
[16] M. Hiller, Five-point wheel suspension, in: J. Angeles, A. Kecskemethy (Eds.), Kinematics and Dynamics of MultiBody Systems, Springer, 1995, pp. 177188.
[17] T.D. Gillespie, Fundamentals of Vehicle Dynamics, SAE, Warrendale, PA, 1992.
[18] J.C. Dixon, Tires, Suspension and Handling, SAE, Warrendale, PA, 1996.
[19] J. Reimpell, H. Stoll, The Automotive Chassis. Engineering Principles, SAE, Warrendale, PA, 1996.