See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/249357751

Semiactive Vibration Control of Train Suspension Systems via

Magnetorheological Dampers

Article in Journal of Intelligent Material Systems and Structures · March 2003

DOI: 10.1177/1045389X03014003004

CITATIONS READS

138 1,186

2 authors:

Wei-Hsin Liao Dao-Hu Wang

The Chinese University of Hong Kong Sun Yat-Sen University
421 PUBLICATIONS 13,153 CITATIONS 43 PUBLICATIONS 1,856 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Wei-Hsin Liao on 17 February 2015.

The user has requested enhancement of the downloaded file.

 Journal of Intelligent Material Systems and
Structures
http://jim.sagepub.com

Semiactive Vibration Control of Train Suspension Systems via Magnetorheological Dampers

W. H. Liao and D. H. Wang
Journal of Intelligent Material Systems and Structures 2003; 14; 161
DOI: 10.1177/1045389X03014003004

The online version of this article can be found at:

http://jim.sagepub.com/cgi/content/abstract/14/3/161

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Intelligent Material Systems and Structures can be found at:

Email Alerts: http://jim.sagepub.com/cgi/alerts

Subscriptions: http://jim.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.co.uk/journalsPermissions.nav

Citations http://jim.sagepub.com/cgi/content/refs/14/3/161

Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009

 Semiactive Vibration Control of Train Suspension
Systems via Magnetorheological Dampers
W. H. LIAO* AND D. H. WANG
Smart Materials and Structures Laboratory, Department of Automation and
Computer-Aided Engineering, The Chinese University of Hong Kong, N. T., Hong Kong

ABSTRACT: This paper is aimed to show the feasibility for improving the ride quality of
railway vehicles with semiactive secondary suspension systems using magnetorheological
(MR) dampers. A nine degree-of-freedom railway vehicle model, which includes a car body,
two trucks and four wheelsets, is proposed to cope with vertical, pitch and roll motions of the
car body and trucks. The governing equations of the railway vehicle suspension systems
integrated with MR dampers are developed. To illustrate the feasibility and effectiveness of the
controlled MR dampers on railway vehicle suspension systems, the LQG control law using
the acceleration feedback is adopted as the system controller, in which the state variables
are estimated from the measurable accelerations with the Kalman estimator. In order to make
the MR dampers track the optimal damping forces, a damper controller to command the
voltage to the current drivers for the MR dampers is proposed. The acceleration responses
of the car body of the train vehicle with semiactive secondary suspension system integrated
with MR dampers are evaluated under random and periodical track irregularities. This
semiactive controlled system is also compared to the conventional passive suspension system
using viscous dampers without MR dampers, and the secondary suspension system integrated
with MR dampers in passive on and passive off modes. The simulation results show that the
vibration control of the train suspension system with semiactive controlled MR dampers
is feasible and effective.

Key Words: train suspension, magnetorheological damper, vibration control, semiactive

control.

INTRODUCTION Therefore, several researchers (Sasaki et al., 1994;

Shimamune and Tanifuji, 1995) have proposed and
The development of high-speed railway vehicles has investigated active suspension technology for railway
been a great interest of many countries because high- vehicles, which utilize oil cylinders and pneumatic
speed trains have been proven as an efficient actuators. The active suspension provides high control
and economical transportation means while minimizing performance over wide frequency range, but it requires
air pollution. However, the high speed of the train high power and sophisticated control implementation.
would cause significant car body vibrations, which Furthermore, the active control suspension system
induce the following problems: the ride stability, the would import the mechanical power into the system,
ride quality, and the cost of track maintenance. Thus the so the stability of the control system needs to be consi-
vibration control of the car body is needed to improve dered. In recent years, semiactive suspension systems
the ride comfort and safety of a train. Various kinds that utilize controllable devices based on smart fluids
of railway vehicle suspensions linking the bogies and have drawn the attention of many researchers. The
the car bodies have been designed to cushion riders essential characteristics of the smart fluids are their
from vibrations. In general, the suspension systems abilities to reversibly change from a free-flowing fluid
used in railway vehicles can be categorized as passive, to a semisolid with controllable yield strength in
active, and semiactive types. A passive railway milliseconds when exposed to an electric or magnetic
vehicle suspension employing springs and pneumatic field (Sims et al., 1999). Two fluids that are viable
or oil dampers has some advantages such as design contenders for the development of controllable dampers
simplicity and cost effectiveness. However the perfor- are electrorheological (ER) and magnetorheological
mance on the wide frequency range could be limited. (MR) fluids (Wereley and Pang, 1998).
Peel et al. (1996) have developed a mathematical
model of a controllable vibration damper employing ER
*Author to whom correspondence should be addressed.
E-mail: [email protected]
fluids intended for the application to suspension systems

JOURNAL OF INTELLIGENT MATERIAL SYSTEMS AND STRUCTURES, Vol. 14—March 2003 161
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
1045-389X/03/03 0161–12 $10.00/0 DOI: 10.1177/104538903033283
ß 2003 Sage Publications
162 W. H. LIAO AND D. H. WANG

of ground vehicles. The modeling technique was dampers (in lateral direction, denoted as ‘‘PD’’ in
illustrated in an application for controlling the lateral Figure 1), which are referred to as the secondary
dynamics of a modern rail vehicle. However, it is well suspension system. For the semiactive train suspension
known that ER fluids are excited by high electric fields. system in this study, only four MR dampers are
To produce sufficient levels of field strength requires included by replacing the vertical viscous dampers
high voltage, which restrains its potential applications between the car body and two trucks. The schematic
due to the safety problems. On the other hand, MR configuration of the semiactive control system for the
fluids are excited by a magnetic field, which can be railway vehicle is shown in Figure 2. In this system,
generated by a low voltage source. In addition, MR four MR dampers, which are used to control the
fluids generate significantly larger dynamic force levels vertical, pitch and roll vibrations, are vertically placed
than ER fluids and operate over wide temperature on the left and right sides of each truck (denoted as
ranges. More recently, the semiactive dampers using MRDszlr, MRDszll, MRDsztr, and MRDsztl respectively
MR fluids are developed and applied to control the in Figure 2). In order to realize the feedback control, the
vibration of automobiles and heavy trucks by some accelerations of the car body and trucks are measured
researchers (Choi et al., 2000; Simon and Ahmadian, with accelerometers, which are installed in vertical
2001). In this paper, a semiactive secondary suspension and lateral directions as shown in Figure 2.
system with MR dampers for a full-size railway vehicle
is modeled and the system performance is evaluated.
Force
Transducer
Car Body
V
RAILWAY VEHICLE SUSPENSION SYSTEMS
Accelerometer

MRD sztl MRD szll

In this study, a full-size train vehicle model has been
System Controller
formulated for a railway vehicle running on a straight MRD sztr MRD szlr
track. A nine degree-of-freedom analytical model ud
is considered. This model, which is shown in Figure 1, Trailing Truck Leading Truck MR Damper

consists of a car body, two truck frames, and four Damper Damper Damper Damper
Controller Controller Controller Controller
wheelsets. The wheelsets and truck frames are connected
by a primary suspension system that consists of springs
and viscous dampers. The car body and truck frames
are connected by springs and MR dampers (in vertical Figure 2. Schematic of semiactive control system for railway
direction, denoted as ‘‘MRD’’ in Figure 1)/viscous vehicle.

Figure 1. Nine degree-of-freedom train vehicle model with MR dampers.

Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
 Semiactive Vibration Control of Train Suspension Systems 163

Analytical Model of Railway Vehicle where F with different subscripts in Equations (1)–(9)
represent the suspension forces produced by the
The governing equations of motion for the railway secondary and primary suspensions and the definitions
vehicle with suspension systems can be derived using of F in Equations (1)–(9) are listed in APPENDIX II.
Newton’s Laws and the nomenclatures used in the deve- fszlr , fszll , fsztr , and fsztl represent the damping forces
lopment of the formulation are defined in APPENDIX I. produced by the corresponding MR dampers MRDszlr,
The governing equations for the car body (vertical zc , MRDszll, MRDsztr, and MRDsztl, which are illustrated
pitch ’c , and roll c ) can be expressed as in Figure 2. In the subscripts of F and f, the first letter
(‘‘s’’ or ‘‘p’’) is used to represent the secondary and
mc z€c ¼ Fszlr þ Fszll þ Fsztr þ Fsztl þ fszlr þ fszll þ fsztr þ fsztl primary suspensions respectively (‘‘s’’ – secondary
ð1Þ suspension, ‘‘p’’ – primary suspension). The second
letter of the subscripts (‘‘x’’, ‘‘y’’, or ‘‘z’’) represent the
Icy ’€ c ¼ ðFsxlr þ Fsxll Þhcs þ ðFsxtr þ Fsxtl Þhcs  ðFszlr þ Fszll Þl directions of the suspension and damping forces. The
þ ðFsztr þ Fsztl Þl  ðfszlr þ fszll Þl þ ðfsztr þ fsztl Þl third letter (‘‘l ’’, ‘‘t ’’, ‘‘1–4’’ ) of the subscripts is used to
ð2Þ identify the trucks (‘‘l’’ – the leading truck, ‘‘t ’’ – the
Icx €c ¼ ðFsylr þ Fsyll Þhcs  ðFsytr þ Fsytl Þhcs trailing truck) and the wheelsets ( ‘‘1–4’’ represent the
wheelsets 1–4 respectively as marked in Figure 1). The
þ ðFszlr þ Fsztr Þds  ðFszll þ Fsztl Þds last letter of the subscripts (‘‘r’’ or ‘‘l’’) is used to identify
þ ðfszlr þ fsztr Þds  ðfszll þ fsztl Þds ð3Þ the side of suspension system for the corresponding
forces (‘‘r’’ – the parts located on the right side, ‘‘l’’ – the
The governing equations for the leading truck (vertical parts located on the left side of the car body, the trucks,
zt1 , pitch ’t1 , and roll t1 ) can be expressed as follows and the wheelsets). It should be noted that the secondary
vertical damping csz is zero for the system with MR
mt z€t1 ¼ ðFszlr þ Fszll Þ þ ðFpz1r þ Fpz2r Þ þ ðFpz1l þ Fpz2l Þ dampers since the corresponding viscous dampers have
been replaced by the MR dampers. However, for the
 ðfszlr þ fszll Þ ð4Þ
purpose of comparisons, the model developed here can
also be used for analyzing the conventional passive
Ity ’€ t1 ¼ ðFsxlr þ Fsxll Þhts þ ðFpx1r þ Fpx1l Þhtp system, which the secondary vertical viscous dampers
þ ðFpx2r þ Fpx2l Þhtp  ðFpz1r þ Fpz1l Þb instead of MR dampers are used. In this case, those
damping forces fszlr , fszll , fsztr and fsztl corresponding to
þ ðFpz2r þ Fpz2l Þb ð5Þ
MR dampers are zero.
Itx €t1 ¼ ðFsylr þ Fsyll Þhts  ðFszlr  Fszll Þds MR Damper Model
 ðFpy1r þ Fpy2r Þhtp  ðFpy1l þ Fpy2l Þhtp
The MR damper model proposed by Spencer et al.
þ ðFpz1r þ Fpz2r Þdp  ðFpz1l þ Fpz2l Þdp as shown in Figure 3 is adopted in this study. The
 ðfszlr  fszll Þds ð6Þ phenomenological model is governed by the following
equations (Spencer Jr. et al., 1997)
The governing equations for the trailing truck (vertical f ¼ c1 y_ þ k1 ðx  x0 Þ ð10Þ
zt2 , pitch ’t2 , and roll t2 ) can be expressed as follows
z_ ¼ jx_  y_ jjzjn1 z  ðx_  y_ Þjzjn þ Aðx_  y_ Þ ð11Þ
mt z€t2 ¼ ðFsztr þ Fsztl Þ þ ðFpz3r þ Fpz3l Þ þ ðFpz4r þ Fpz4l Þ
1
 ð fsztr þ fsztl Þ ð7Þ y_ ¼ ½z þ c0 x_ þ k0 ðx  yÞ ð12Þ
c 0 þ c1
Ity ’€ t2 ¼ ðFsxtr þ Fsxtl Þhts þ ðFpx3r þ Fpx3l Þhtp
y x


þ ðFpx4r þ Fpx4l Þhtp  Fpz3r þ Fpz3l b Bouc − Wen
þ ðFpz4r þ Fpz4l Þb ð8Þ c1
k0 f
Itx €t2 ¼ ðFsytr þ Fsytl Þhts  ðFsztr  Fsztl Þds c0
 ðFpy3r þ Fpy4r Þhtp  ðFpy3l þ Fpy4l Þhtp k1
þ ðFpz3r þ Fpz4r Þdp  ðFpz3l þ Fpz4l Þdp
 ðfsztr  fsztl Þds ð9Þ Figure 3. Mechanical model for MR damper.
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
164 W. H. LIAO AND D. H. WANG

Table 1. Parameters for MR damper model (Spencer Jr. et al., 1997).

Parameter Value Parameter Value Parameter Value

c0a 2.10 103 N s/m a 1.40 104 N/m x0 0.143 m

c0b 3.50 102 N s/m V b 6.95 104 N/m V  190 s1
k0 4.69 103 N/m  3.63 106 m2 k1 500 N/m
c1a 2.83 104 N s/m  3.63 106 m2 n 2
c1b 2.95 102 N s/m V A 301

 ¼ ðuÞ ¼ a þ b u; c1 ¼ c1 ðuÞ ¼ c1a þ c1b u; where

c0 ¼ c0 ðuÞ ¼ c0a þ c0b u ð13a–c) " # " #
0 I 18 18
0
u_ ¼ ðu  vÞ ð14Þ A¼ , A2< ;B ¼ ,
M1 K M1 C M1 Fu

where the damping force f of the MR damper is given " #

18 4
0
by Equation (10), z is an evolutionary variable to B2< ;G ¼ , G 2 <18 16
;
represent the hysteresis loop that is given by Equation M1 Fw
 
(11), v is the command voltage sent to the current Co ¼ M1 K M1 C , Co 2 <9 18
;
driver for the MR damper. In this model, there are
a total of 14 parameters ðc0a , c0b , k0 , c1a , c1b , k1 , x0 , D ¼ M1 Fu , D 2 <9 4 ; H ¼ M1 Fw , H 2 <9 16
;
a , b , , , A, n, Þ which definitions can be referred
to APPENDIX I, to characterize the MR damper. In y is the output vector; v is the sensor noise vector.
this paper, the MR dampers that produced by Lord
Corporation are used (more descriptions can be found
in Spencer Jr. et al., 1997) and the model parameters are SEMIACTIVE CONTROLLER DESIGN
shown in Table 1.
In order to evaluate the effectiveness of the semiactive
State Equations suspension system with MR dampers, the
Linear Quadratic Gaussian (LQG) control algorithm is
Let q ¼ ½ zc c c zt1 t1 t1 zt2 t2 t2 T , employed. The performance index is chosen as
the governing Equations (1)–(9) of the railway vehicle
with the MR dampers can be written in the matrix Z tf
1
form as J ¼ lim E ½xT ðtÞQxðtÞ þ ud ðtÞT Rud ðtÞdt ð17Þ
tf !1 tf 0

M€q þ C_q þ Kq ¼ Fu u þ Fw w ð15Þ

where Q and R are symmetric semipositive-definite
and positive-definite matrices. ud ðtÞ is the vector of
where M, C, and K are the mass, damping and stiffness desired damping forces of the MR dampers, and ud ðtÞ ¼
matrices respectively, M 2 <9 9 , C 2 <9 9 and K 2 d
½ fszlr d
fszll d
fsztr d T . The control law that minimizes
fsztl
<9 9 u is the vector of controlled damping forces Equation (17) is given by
produced by MR dampers, and u ¼ ½ fszlr fszll fsztr
fsztl T w is the disturbance vector determined by track xðtÞ
ud ðtÞ ¼ K^ ð18Þ
irregularities. Let w1 ¼ ½ z1r z1l z2r z2l z3r z3l z4r z4l T
and w2 ¼ ½ z_1r z_1l z_2r z_2l z_3r z_3l z_4r z_4l T , then w ¼ ½ww12 .
where K ¼ R1 BT S, S is determined by SBR1 BT S
The elements zir and zil (i ¼ 1 4) are the vertical dis-
SA  AT S ¼ Q:^
xðtÞ is obtained from the following
placement disturbances acted on the wheels of the train
Kalman estimator (Siouris, 1996)
vehicle, which are shown in Figure 1. Fu and Fw are the
coefficient matrices of the damping force and the
disturbance vectors. x^_ ðtÞ ¼ A^
xðtÞ þ BuðtÞ þ Kf ½yðtÞ  Co x^ ðtÞ  DuðtÞ ð19Þ

By defining the state vector as x ¼ qq_ , the governing
Equation (15) can be rewritten in the state-space form where Kf ¼ Sf CTo R1f , Sf is determined by Sf Co Rf
T 1
T T
as follows Co Sf  ASf  Sf A ¼ GQf G . The noise covariance
matrices Qf ¼ EðwwT Þ and Rf ¼ EðvvT Þ are symmetric,
( semipositive-definite and positive-definite respectively.
x_ ¼ Ax þ BuðtÞ þ GwðtÞ
ð16Þ The desired damping forces vector ud ðtÞ of the MR
y ¼ Co x þ DuðtÞ þ HwðtÞ þ v dampers can be obtained from the system controller
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
 Semiactive Vibration Control of Train Suspension Systems 165

(Equations (18) and (19)). However, these forces cannot

be commanded directly to the MR dampers that are
semiactive devices, thus the following damper controller
is proposed

Vmax X d
v¼ sgn ½ f  ð1  kiÞ f  f þ 1 ð20Þ
2N 0  i  N  1

where sgnð Þ is the signum function; f d and f are the

desired damping force and the actual damping force
of the MR damper, respectively; N is a positive integer
and 0  i  N  1; k is a small constant; Vmax is the
maximum input voltage to the current driver for the MR
damper. To determine the command voltage according
to Equation (20), N times of comparisons between
the desired and the actual damping forces are needed.
F or th e it h c o m p a r i s o n , i f jf d j  jð1  kiÞf j >
0, f d and f have the same sign as well, then the
contribution of this comparison to the command
voltage will be vi ¼Vmax =2N 2; otherwise the
contribution to the command voltage will be
vi ¼ Vmax =2N 0. All the comparisons for ði ¼ 0, Figure 4. Damping forces and command voltage for MR damper
controller.
1, . . . , N  1Þ are carried out simultaneously and the
command voltage is the summation of the contribution
of NP comparisons to the command voltage, that is
SIMULATION RESULTS AND DISCUSSIONS
v ¼ 0  i  N  1 vi . So N determines the number of
comparisons between the desired and the actual damp-
In the simulation, the elements of the weighting matrix
ing forces in order to set the command voltage. In this
Qði, jÞ ¼ 0 for i ¼ 1 18, j ¼ 1 18 except
study, N ¼ 6, k ¼ 5:00 104 , and Vmax ¼ 12V are
Qð1, 1Þ ¼ 8 108 , Qð2, 2Þ ¼ 8 108 , Qð3, 3Þ ¼ 8 108 ,
used.
Qð13, 13Þ ¼ 100, Qð16, 16Þ ¼ 100 and the weighting
To illustrate the effectiveness of the MR damper
matrices R ¼ 0:01I4 4 , Qf ¼ 20I16 16 , Rf ¼ I9 9 (I is
controller (Equation (20)), an example is considered
the identity matrix, and their subscripts represent the
here. The displacement input across the damper is
corresponding dimensions). The parameters of a railway
given by
vehicle are given in Table 2 and the parameters of the
  MR damper model are shown in Table 1. It is assumed
x ¼ 2 sin 40t  ð21Þ that the wheels of the vehicle system follow the rails
3
perfectly. Two track irregularities are considered here:
and the desired damping force is set by one is the random track irregularity and the other is the
periodical track irregularity.
The power spectrum densities (PSD) of vertical, pitch
f d ¼ 3600 sinð40tÞ ð22Þ and roll accelerations of the car body of the train vehicle
under random track irregularities are illustrated in
Then the command Equation (20) is used to let the MR Figure 5. The RMS accelerations of the car body are
damper track the desired damping force determined shown in Table 3. The ‘‘Passive’’ case represents the
by Equation (22). Figure 4 shows the time history of conventional passive system using viscous dampers
the controlled damping force and the corresponding without MR dampers for the secondary suspension
command voltage, while the damping forces of other system. Using four MR dampers by replacing the
cases are also shown for the purpose of comparison. vertical viscous dampers for the secondary suspension
It can be seen that the damping forces in passive on system, there are three cases. When the command
mode (constant command voltage, v ¼ 12V) and passive voltage to the current drivers for the MR dampers is
off mode (constant command voltage, v ¼ 0V) have zero or constant 12 V, they are referred as ‘‘Passive Off ’’
significant deviations from the desired damping force. and ‘‘Passive On’’, respectively. For the ‘‘Semiactive
On the other hand, the controlled damping force tracks Control’’ case, the MR dampers are operated in
well the desired one by applying voltage input as semiactive control mode via the system and damper
Equation (20). controllers.
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
166 W. H. LIAO AND D. H. WANG

Table 2. Parameters for railway vehicle model.

Parameter Value Parameter Value Parameter Value

mc 3.96 104 kg mt 3.25 103 kg V 55.56 m/s

Icx 8.85 104 kg m2 Icy 2.46 106 kg m2 Icz 2.505 106 kg m2
Itx 3.06 103 kg m2 Ity 3.02 103 kg m2 Itz 4.27 103 kg m2
kpx 4.00 106 N/m kpy 3.25 106 N/m kpz 7.00 105 N/m
cpx 0 cpy 0 cpz 1.50 104 N s/m
ksx 1.5 105 N/m ksy 1.50 105 N/m ksz 2.90 105 N/m
csx 0 csy 5.00 104 N s/m csz * 8.00 104 N s/m
hts 0.217 m hcs 1.207 m htp  0.452 m
hwp 0.180 m l 9.00 m a 0.7465 m
ds 1.00 m b 1.25 m dp 1.00 m
*The value of csz is used for the conventional passive secondary suspension system in order to compare
with the semiactive secondary suspension. For the secondary suspension integrated with MR dampers,
csz ¼ 0.

Figure 5. Acceleration responses of car body under random track irregularities: (a) Vertical acceleration ðz€ c Þ; (b) Pitch acceleration ð’€ c Þ; (c) Roll
acceleration ð€c Þ.

To evaluate the ride quality, four positions in the passenger points of the train vehicle are given by
passenger compartment are chosen. The first two posi-
tions are in the front car body at ðxd , yd , zd Þ and
ðxd ,  yd , zd Þ, and the other two positions are in the z€fr ðxd , yd , zd Þ ¼ z€c  xd ’€ c þ yd €c ð23aÞ
rear car body at ðxd , yd , zd Þ and ðxd ,  yd , zd Þ, where
are shown in Figure 1. The accelerations at these four z€fl ðxd , yd , zd Þ ¼ z€c  xd ’€ c  yd €c ð23bÞ
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
 Semiactive Vibration Control of Train Suspension Systems 167

z€rr ðxd , yd , zd Þ ¼ z€c þ xd ’€ c þ yd €c ð23cÞ 12 V applied to the MR dampers), the car body vertical
and pitch accelerations lie between the ‘‘Passive’’ and
z€rl ðxd , yd , zd Þ ¼ z€c þ xd ’€ c  yd €c ð23dÞ ‘‘Passive Off ’’ cases for most of the frequencies shown.
While the MR dampers are operated in semiactive
Figures 6 and 7 show the PSD and time histories of control mode, it can be seen that the car body
the acceleration responses z€fr and z€rl at two passenger accelerations are significantly reduced compared to
points in the car body, where xd ¼ 9 m,yd ¼ 0.75 m, and those three passive cases.
zd ¼  0.2 m. The z€fr is the acceleration on the right side From Figure 5(c), it is illustrated that the secondary
of the front car body, and the z€rl is the acceleration on suspension system integrated with MR dampers
the left side of the rear car body. The RMS accelerations (no matter whatever modes are employed) is especially
at four passenger points in the car body are also given in effective for reducing the roll vibration of the
Table 3. car body compared to the conventional passive suspen-
Observing Figure 5(a) and (b), those passive and sion system. The results can also be quantified
semiactive cases can be compared. For the first peak from Table 3. While the vertically installed MR dampers
frequency responses of the car body vertical and pitch are effective for controlling the vertical vibration of
accelerations, the conventional passive secondary sus- the car body (about 29% improvement in terms
pension system has better isolation than the ‘‘Passive of vibration reduction for ‘‘Semiactive Control’’ vs.
Off’’ system, in which the MR dampers are not activated ‘‘Passive On’’), it is even more significant for the
(0 V). On the other hand, for attenuating the frequency controlled MR dampers for reducing the pitch and roll
responses other than the first resonance, the secondary vibrations (41 and 57%, respectively).
suspension system integrated with MR dampers in the From Figures 6 and 7, the accelerations at the
passive off mode is superior to the conventional passive passenger points are also significantly attenuated
suspension system. For the ‘‘Passive On’’ case (constant through the semiactive secondary suspension system

Table 3. Car body RMS acceleration under random excitation.

Reduction (%)

Passive Passive Off Passive On Semiactive Semiactive Semiactive Semiactive

( 103) ( 103) ( 103) Control ( 103) Control Control Control
versus versus versus
Passive Passive Off Passive On

Vertical acceleration (m/s2) 6.9 5.8 3.1 2.2 68.12 62.07 29.03
Pitch acceleration (rad/s2) 1.3 1.3 0.566 0.336 74.15 74.15 41.07
Roll acceleration (rad/s2) 2.1 0.462 0.388 0.167 91.62 63.88 56.96
Acceleration at Passenger z€ fr 13.9 10.7 5.0 3.0 78.42 71.96 40.00
Points (m/s2) z€ fl 13.9 10.7 5.0 3.0 78.42 71.96 40.00
z€ rr 14.2 14.8 5.9 3.9 72.54 73.65 33.90
z€ rl 14.2 14.8 5.9 3.9 72.54 73.65 33.90

Figure 6. Acceleration response z€ fr at (xd ,yd ,zd ) under random track irregularities: (a) PSD; (b) Time history.
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
168 W. H. LIAO AND D. H. WANG

Figure 7. Acceleration response z€ rl at (xd ,yd ,zd ) under random track irregularities: (a) PSD; (b) Time history.

whole. When the MR damper is employed in passive

on or passive off modes, the MR damping force has
more derivation from the desired damping force. Even
through the controlled damping force cannot perfectly
follow the desired damping force, the semiactive control
is still superior to the ‘‘Passive On’’ and ‘‘Passive Off ’’
cases with the MR dampers.
If the vertical excitation of the front wheelset (the
wheelset 1 as indicated in Figure 1) of the leading truck
is represented by z1l and z1r , the periodical track
irregularities can be expressed as
2 3
" # 4A 1 1 1
6 cos x  cos 2x þ cos 3x 7
z1r ðtÞ 6 3 15 35 7
¼6  7
z1l ðtÞ 4 4A 1 1 1 5
Figure 8. Damping forces and command voltage for MRDszlr under cos x  cos 2x þ cos 3x
random track irregularities. 3 15 35
ð24Þ
compared to those through the conventional passive
suspension system when the train vehicle experiences the where A is the scalar factor of the periodical irregula-
random track irregularities. It is clearly shown that the rities of the track,  ð ¼ 2 =L, unit: rad/m) is the
reduction of the accelerations at the passenger points spatial frequency, L is the spatial length (the rail length),
through semiactive control is also improved compared and x ¼ Vt. In this paper, A ¼ 25.4 and L ¼ 25 m. In
to those with constant 12 voltage (Passive On) and 0 addition, the vertical excitations to other wheelsets
voltage (Passive Off) to the MR dampers, when the z2r z4r , z2l z4l are described by
secondary suspension system is integrated with MR " # " # " # " #
dampers. The performance can also be verified from z2r ðtÞ z1r ðt  1Þ z3r ðtÞ z1r ðt  2Þ
Table 3. It should be noted that the vertical accelera- ¼ ; ¼ ;
z2l ðtÞ z1l ðt  1Þ z3l ðtÞ z1l ðt  2Þ
tions at passenger points (34% or 40% reductions for
‘‘Semiactive Control’’ versus ‘‘Passive On’’) can be " # " #
z4r ðtÞ z1r ðt  3Þ
reduced more than that of the car body (29%) due to ¼ ð25Þ
the coupling effects among the vertical, pitch and roll z4l ðtÞ z1l ðt  3Þ
motions.
In Figure 8, the damping forces and the correspond- where 1 ¼ V=2b, 2 ¼ V=2l, and 3 ¼ V=ð2ðb þ lÞÞ.
ing command voltage for the MR damper MRDszlr From Equations (24) and (25), the disturbance vector
under random track irregularities are plotted. It can be w in Equation (15) can be determined.
seen that the controlled damping force of the MR Figure 9 shows the time responses of the car
damper can follow the desired damping force on the body accelerations under the given periodical track
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
 Semiactive Vibration Control of Train Suspension Systems 169

Figure 9. Time responses of car body acceleration under periodical track irregularities: (a) Vertical acceleration; (b) Pitch acceleration.

be more significant as shown in Figure 5(c) under the

random excitation.
The damping forces and the corresponding command
voltage for the MR damper MRDszlr under periodical
track irregularities are also shown in Figure 10. It can
be seen again that the controlled damping force of the
MR damper can follow the desired damping force
on the whole under periodical track irregularities. When
the MR damper is employed in passive on or passive
off modes, the damping force of the MR damper has
more deviation from the desired damping force.
Therefore, it can be concluded again that the semiactive
control is superior to the passive on and passive off
modes of the MR damper under periodical track
irregularities, even the controlled damping force
cannot perfectly follow the desired damping force.
Figure 10. Damping forces and command voltage for MRDszlr under
periodical track irregularities.

CONCLUSION
irregularities. Almost the same levels of the car body
accelerations are obtained for the systems with conven- In this paper, a semiactive secondary train suspension
tional passive viscous dampers (‘‘Passive’’) and ‘‘Passive system with MR dampers has been investigated
Off’’ MR dampers. Under the periodical excitations, the by considering a full-size railway vehicle, which includes
accelerations of the car body when MR dampers are three vibration motions (vertical, pitch and roll) of
employed in passive on mode are larger than those in the car body and trucks. The governing equations of a
passive off mode. This shows that the increase in the nine degree-of-freedom railway vehicle model integrated
damping forces is not always beneficial to the vibration with MR dampers are developed. To illustrate the
attenuation when the MR dampers are integrated feasibility and effectiveness of controlled MR dampers
into the system. On the other hand, the controlled on railway vehicle suspension systems, the LQG control
accelerations of the car body in the vertical and pitch using the acceleration feedback is adopted as the system
motions are greatly reduced (about 50%) compared to controller, in which the state variables are estimated
the passive on case. from the measurable accelerations with the Kalman
However, the roll accelerations of the car body are estimator. In order to let the MR dampers track the
found to be very small compared to the pitch ones. It optimal damping forces, a signum function based con-
is because the given periodical track irregularities do trol algorithm to command the voltage to the current
not include cross level track irregularities. While drivers for MR dampers is proposed (damper control-
asynchronous track irregularities are considered, the ler). The acceleration responses of the car body and four
control effects on the roll motion of the car body would passenger points of the train vehicle with semiactive
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
170 W. H. LIAO AND D. H. WANG

secondary MR suspension system and with conventional ksx ¼ secondary longitudinal stiffness
passive secondary suspension system using viscous ksy ¼ secondary lateral stiffness
dampers under random and periodical track irregula- ksz ¼ secondary vertical stiffness
rities are evaluated and compared with each other. For csx ¼ secondary longitudinal damping
the secondary suspension system integrated with MR csy ¼ secondary lateral damping
dampers, the other two cases, in which the MR dampers csz ¼ secondary vertical damping
are employed in passive on or passive off modes, are l ¼ half of truck center pin spacing
also investigated. b ¼ half of wheelbase
The simulation results show that the semiactive dp ¼ half of primary suspension spacing (lateral)
secondary suspension system can significantly attenuate a ¼ half of wheelset contact distance
the vibrations of the car body and is particularly ds ¼ half of secondary suspension spacing (lateral)
effective for reducing the roll vibration of the car body hts ¼ vertical distance from truck frame center of
of the railway vehicle under random track irregularities. gravity to secondary suspension
The results also show that the increase of damping hcs ¼ vertical distance from car body center of gravity to
forces of MR dampers is not always beneficial to secondary suspension
attenuate vibration of the vehicle car body. When MR htp ¼ vertical distance from truck frame center of
dampers are integrated into the suspension system of the gravity to primary suspension
vehicle, the optimal damping forces produced by the hwp ¼ vertical distance from wheelset center of gravity to
MR dampers can be obtained through suitable system primary suspension
and damper controllers. In summary, the results of this V ¼ velocity of railway vehicle
study show that the vibration control of the train
suspension system with semiactive controlled MR Rail track irregularities
dampers is feasible and effective. However, this paper
only presents some preliminary research results that z1r ¼ vertical disturbance acted on the right wheel
verify the feasibility of applying MR dampers to train of wheelset 1
suspension systems, more thorough research work is z1l ¼ vertical disturbance acted on the left wheel of
under investigation. wheelset 1
z2r ¼ vertical disturbance acted on the right wheel
of wheelset 2
ACKNOWLEDGMENTS z2l ¼ vertical disturbance acted on the left wheel of
wheelset 2
The work described in this paper was fully supported z3r ¼ vertical disturbance acted on the right wheel
by a grant from Research Grants Council of Hong Kong of wheelset 3
Special Administrative Region, China (Project No. z3l ¼ vertical disturbance acted on the left wheel of
CUHK4216/01E) and CUHK Postdoctoral Fellowship wheelset 3
Scheme (01/ERG/04). z4r ¼ vertical disturbance acted on the right wheel
of wheelset 4
z4l ¼ vertical disturbance acted on the left wheel
APPENDIX I: NOMENCLATURE of wheelset 4

Vehicle system MR Damper

mc ¼ mass of car body c0 ¼ viscous damping observed at large velocities

Icx ¼ roll moment of inertia of car body c0a , c0b ¼ constants that determine c0
Icy ¼ pitch moment of inertia of car body k0 ¼ stiffness at large velocities
Icz ¼ yaw moment of inertia of car body c1 ¼ viscous damping observed at low velocities
mt ¼ mass of truck c1a , c1b ¼ constants that determine c1
Itx ¼ roll moment of inertia of truck k1 ¼ accumulator stiffness
Ity ¼ pitch moment of inertia of truck x0 ¼ initial displacement of spring k1 associated
Itz ¼ yaw moment of inertia of truck with nominal damper force due to accumu-
kpx ¼ primary longitudinal stiffness lator
kpy ¼ primary lateral stiffness  ¼ scaling value for the Bouc-Wen model
kpz ¼ primary vertical stiffness a , b ¼ constants that determine 
cpx ¼ primary longitudinal damping , , A, n ¼ parameters that determine the hysteresis
cpy ¼ primary lateral damping loop in Bouc-Wen model
cpz ¼ primary vertical damping  ¼ constant to govern the first-order filter
Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009
 Semiactive Vibration Control of Train Suspension Systems 171

8  
APPENDIX II: DEFINITIONS OF F IN > dp
>
> Fpz1r ¼ kpz zt1  b’t1 þ dp t1  z1r
EQUATIONS (1)–(9) >
> a
>
>
>
>  
>
> dp
>
> _
 cpz z_t1  b’_ t1 þ dp t1  z_1r ;
8 >
> a
Fsxlr ¼ ksx ðhcs ’c þ hts ’t1 Þ  csx ðhcs ’_ c þ hts ’_ t1 Þ; >
>
>
> >
>  
>
> >
> dp
>
> >
>
< Fsxll ¼ ksx ðhcs ’c þ hts ’t1 Þ  csx ðhcs ’_ c þ hts ’_ t1 Þ; > Fpz1l
> ¼ kpz zt1  b’t1  dp t1  z1l
>
> a
>
>
>
> >
>  
>
> Fsxtr ¼ ksx ðhcs ’c þ hts ’t2 Þ  csx ðhcs ’_ c þ hts ’_ t2 Þ; >
> dp
>
>
:
>
>
>  cpz z_t1  b’_ t1  dp _t1  z_1l ;
>
> a
Fsxtl ¼ ksx ðhcs ’c þ hts ’t2 Þ  csx ðhcs ’_ c þ hts ’_ t2 Þ >
>  
>
>
>
> dp
8 >
> Fpz2r ¼ kpz zt1 þ b’t1 þ dp t1  z2r
> Fsylr ¼ ksy ðhcs c þ hts t1 Þ þ csy ðhcs _c þ hts _t1 Þ; >
> a
>
> >
>
>
> >
>  
>
> >
> dp
< Fsyll ¼ ksy ðhcs c þ hts t1 Þ þ csy ðhcs _c þ hts _t1 Þ; >
> _
 cpz z_t1 þ b’_ t1 þ dp t1  z_2r ;
>
> a
>
>
>
> >
>  
>
> Fsytr ¼ ksy ðhcs c þ hts t2 Þ þ csy ðhcs _c þ hts _t2 Þ; >
> dp
>
> >
>
>
: >
> Fpz2l ¼ kpz zt1 þ b’t1  dp t1  z2l
>
> a
Fsytl ¼ ksy ðhcs c þ hts t2 Þ þ csy ðhcs _c þ hts _t2 Þ >
>
>
>  
>
> dp
8 >
> _
 cpz z_t1 þ b’_ t1  dp t1  z_2l ;
> Fszlr ¼ ksz ½zc  l’c  zt1 þ ds ðc  t1 Þ < a
>
>
>
>  
>
> >
> dp
>
>
>  csz ½z_c  l ’_ c  z_t1 þ ds ð_c  _t1 Þ; >
> F ¼ kpz zt2  b’t2 þ dp t2  z3r
> >
> pz3r a
>
> >
>
>
> Fszll ¼ ksz ½zc  l’c  zt1  ds ðc  t1 Þ >
>  
>
> >
>
>
> >
> dp
>
> >
>  cpz z_t2  b’_ t2 þ dp _t2  z_3r ;
>
<  csz ½z_c  l ’_ c  z_t1  ds ð_c  _t1 Þ; >
> a
>
>
>
>  
> >
> dp
>
> Fsztr ¼ ksz ½zc þ l’c  zt2 þ ds ðc  t2 Þ >
> Fpz3l ¼ kpz zt2  b’t2  dp t2  z3l
>
> >
> a
>
> >
>
>
> >
>  
>
>  csz ½z_c þ l ’_ c  z_t2 þ ds ð_c  _t2 Þ; >
> dp
>
> >
> _
 cpz z_t2  b’_ t2  dp t2  z_3l ;
>
> >
>
>
> Fsztl ¼ ksz ½zc þ l’c  zt2  ds ðc  t2 Þ >
> a
>
> >
>  
>
> >
>
>
: >
> dp
 csz ½z_c þ l ’_ c  z_t2  ds ð_c  _t2 Þ >
> Fpz4r ¼ kpz zt2 þ b’t2 þ dp t2  z4r
>
> a
>
>
8 >
>  
Fpx1r ¼ kpx htp ’t1  cpx htp ’_ t1 ; >
> dp
>
> >
>  cpz z_t2 þ b’_ t2 þ dp _t2  z_4r ;
>
> >
> a
>
> >
>
>
> Fpx1l ¼ kpx htp ’t1  cpx htp ’_ t1 ; >
>  
>
> >
> dp
>
> >
>
>
> >
> Fpz4l ¼ kpz zt2 þ b’t2  dp t2  z4l
>
> Fpx2r ¼ kpx htp ’t1  cpx htp ’_ t1 ; >
> a
>
> >
>
>
> >
>  
>
< Fpx2l ¼ kpx htp ’t1  cpx htp ’_ t1 ; >
> dp
:  cpz z_t2 þ b’_ t2  dp _t2  z_4l
a
>
> Fpx3r ¼ kpx htp ’t2  cpx htp ’_ t2 ;
>
>
>
>
>
>
> ¼ kpx htp ’t2  cpx htp ’_ t2 ;
> Fpx3l
>
>
>
>
>
> REFERENCES
>
> Fpx4r ¼ kpx htp ’t2  cpx htp ’_ t2 ;
>
>
>
:
Fpx4l ¼ kpx htp ’t2  cpx htp ’_ t2 Choi, S. B., Nam, M. H. and Lee, B. K. 2000. ‘‘Vibration Control of a
MR Seat Damper for Commercial Vehicles’’, Journal of Intelligent
8 Material Systems and Structures, 11(12): 936–944.
>
> Fpy1r ¼ kpy htp t1 þ cpy htp _t1 ; Fpy1l ¼ kpy htp t1 þ cpy htp _t1 ; Peel, D. J., Stanway, R. and Bullough, W. A. 1996. ‘‘Dynamic
>
>
>
> Modeling of an ER Vibration Damper for Vehicle Suspension
>
< Fpy2r ¼ kpy htp t1 þ cpy htp _t1 ; Fpy2l ¼ kpy htp t1 þ cpy htp _t1 ; Applications’’, Smart Materials and Structures, 5(5): 591–606.
Sasaki, K., Kamoshita, S. and Enomoto, M. 1994. ‘‘A Design and
>
> Bench Test of Multi-Modal Active Suspension of Railway
>
> Fpy3r ¼ kpy htp t2 þ cpy htp _t2 ; Fpy3l ¼ kpy htp t2 þ cpy htp _t2 ; Vehicle’’, In: Proceedings of the 20th International Conference on
>
>
>
: Industrial Electronics, Control and Instrumentation (IECON’94),
Fpy4r ¼ kpy htp t2 þ cpy htp _t2 ; Fpy4l ¼ kpy htp t2 þ cpy htp _t2 Italy, pp. 2011–2016.

Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009

172 W. H. LIAO AND D. H. WANG

Shimamune, R. and Tanifuji, K. 1995. ‘‘Application of Oil-Hydraulic Siouris, G. M. 1996. An Engineering Approach to Optimal Control and
Actuator for Active Suspension of Railway Vehicle: Experimental Estimation Theory, John Wiley & Sons, Inc., New York, USA.
Study’’, In: Proceedings of the 34th Society of Instrument and Spencer Jr, B. F., Dyke, S. J., Sain, M. K. and Carlson, J. D. 1997.
Control Engineers Annual Conference (SICE’95), Japan, pp. 1335– ‘‘Phenomenological Model for Magneto-Rheological Dampers’’,
1340. ASCE Journal of Engineering Mechanics, 123(3): 230–238.
Simon, D. and Ahmadian, M. 2001. ‘‘Vehicle Evaluation of the Wereley, N. M. and Pang, L. 1998. ‘‘Nondimensional Analysis of
Performance of Magneto-Rheological Dampers for Heavy Truck Semi-Active Electrorheological and Magnetorheological Dampers
Suspensions’’, ASME Journal of Vibration and Acoustics, 123(7): Using Approximate Parallel Plate Models’’, Smart Materials and
365–375. Structures, 7(5): 732–743.
Sims, N. D., Stanway, R. and Johnson, A.R. 1999. ‘‘Vibration Control
Using Smart Fluids: A State-of-the-Art Review’’, The Shock and
Vibration Digest, 31(3): 195–203.

Downloaded from http://jim.sagepub.com at Shanghai Libary on November 2, 2009

View publication stats