Rail Vehicle Active Suspensions

for Lateral Ride and Stability

G. W. Celniker Improvement
Research Assistant, Department
of Mechanical Engineering. The potential performance benefits of simple, low power active elements in the
lateral secondary, between the truck and carbody is investigated analytically by
parametric studies utilizing a fifteen degree of freedom lateral dynamic model
J. K. Hedrick subject to alignment and cross-level inputs. It is shown that significant im-
Associate Professor of provements in ride quality can be obtained at current operating speeds by using less
Mechanical Engineering. than 3 kw per truck by sensing lateral carbody accelerations and utilizing lateral
force actuators between the truck and carbody. It is also shown that in order to
Massachusetts Institute of Technology, improve the truck critical speed and/or to improve the ride quality at very high
Cambridge, Mass. 02139 speeds requires control of the truck kinematic mode. This can be achieved, at the
expense of increased power requirements, by sensing absolute truck lateral and yaw
velocities and utilizing combined lateral force actuators to exert a yaw moment on
the truck. It is concluded that significant ride quality improvements can be achieved
by low power, small bandwidth actuators while stabilty improvements requires
increased power and larger bandwidth systems.

Introduction
The performance limits of conventional passenger rail A recent state-of-the-art review paper [8] discussed the
vehicles using typical passive suspensions is rapidly becoming basic theoretical advantages of active suspensions as well as
well known [1, 2]. The primary suspension is largely deter- the work currently underway in several countries. This paper
mined by the lateral stability/curve negotiation trade-off [2], presents the results of an analytical study whose objective was
while the secondary suspension is largely determined by the to evaluate the potential performance improvements offered
ride quality/suspension stroke trade-off [1]. In order to by a lateral active suspension and to determine the most ef-
improve the vehicles' performance beyond these limits many fective and efficient means of achieving these improvements.
countries are now evaluating the use of active suspensions.
Great Britain [3], Japan [4J, and Germany [5] currently have Lateral Suspension Considerations
analytical and experimental research programs aimed at
determining the performance benefits, maintenance Passive Suspensions. The primary purposes of lateral rail
requirements and economic trade-offs of active suspensions. vehicle suspensions are to provide, 1. "tracking" of in-
Active suspensions, by definition, require a power source tentional inputs such as curves and switches, 2. isolation of
(e.g. electric, hydraulic, and pneumatic), sensors (e.g. ac- the truck and carbody from unintentional disturbances such
celerometers, relative displacement transducers), and ac- as random track irregularities and rail joints and, 3. damping
tuators (e.g., hydraulic, pneumatic, electro-mechanical, and in all modes, but in particular the kinematic or "hunting"
magnetic). The first active suspensions to be employed in mode. In conventional "passive" passenger trucks these tasks
revenue service were "tilt body" devices [6, 7] which bank the are divided between the primary suspension (which, in
carbody an appropriate amount on curves to improve lateral general, are rubber donuts or chevrons between the journal
ride quality. Many countries have evaluated prototype tilt bearing and the truck frame) and the secondary suspension
body devices and these systems are expected to see widespread (which can be provided by hydraulic dampers, anchor rods
application. The use of active suspensions for ride quality and with rubber bushings, centerplates, central bearings, rubber
lateral stability augmentation is currently being evaluated [3, bump stops, and the shear properties of the vertical
4, 5] by many countries and is the topic of this paper. suspension). As mentioned in the introduction, many studies
This paper concentrates on the use of active suspensions to have been done on the optimization of both primary and
improve lateral ride quality and stability. In general this is a secondary passive suspensions [1, 2, 3, 9] and the limiting
more difficult problem than vertical ride quality improvement trade-offs are well known. Stability improvement by con-
due to the more restricted clearances and the absence of ventional passive suspensions is limited by the curving per-
parallel air-spring in the lateral direction. formance, i.e., those parameter changes that improve stability
have an adverse effect on the vehicles' curving performance
[1, 2]. Ride quality improvements by passive means are
Contributed by the Dynamic Systems and Control Division for publication in
the JOURNAL OF DYNAMIC SYSTEMS, MEASUREMENT, AND CONTROL. Manuscript
generally limited by the available suspensions travel [1, 9],
received by the Dynamic Systems and Control Division, June 22, 1981. typically AMTRAKs' North East corridor vehicles are

100/Vol. 104, MARCH 1982 Transactions of the ASME

Copyright © 1982 by ASME
Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab
 Accelerometer

/dt

™%
Ym

K< b
-&~
• Force Actuator, FA

TTTT7T7tTTTTTTT7
Fig. 1 Active suspension

Fig. 3 Acceleration transfer function as a function of absolute

t = -707- velocity feedback gain

^ '•^passive s\
W„y g

Fig. 2 Acceleration transfer function for passive and active

suspensions

Fig. 4 Acceleration transfer function as a function of acceleration

restricted to ±1.5 in. of lateral travel between the carbody feedback gain
and truck before bump stops are contacted. Thus further
improvements by passive means are extremely difficult.
Active Suspensions. References [8, 9] summarize the
fundamental performance improvements that the use of
sensors and actuators can provide, the most important being
that active suspension forces can be functions of absolute
quantities as opposed to passive suspension forces which can
only be functions of local relative quantities. Figure 1 is a
schematic of a single degree of freedom system with an active • Lateral Actuator
force which is proportional to absolute acceleration and Fig. 5 Actuator location for lateral active suspension
velocity, i.e.,
FA — -~ Cvym — Cay„ (1) Active Rail Vehicle Lateral Suspensions
Figure 2 shows the basic trade-off in choosing the damping
ratio of a passive damper, high values attenuate the resonant A number of previous analytical studies [11-14] have
peak near the natural frequency but provide greater am- looked at applying active suspensions to improve the lateral
plification at high frequencies. performance of rail vehicles. In this paper a more complete
Figure 3 shows the effect of increasing the amount of model (e.g. the use of a flexible truck model) and a more
absolute velocity feedback, clearly resonance control is complete control study is made in order to make more realistic
achieved without sacrificing higher frequency performance. performance and power calculations.
Figure 4 shows the effect of increasing absolute acceleration Actuator Location. It is assumed in this paper that the force
feedback, both the natural frequency and the acceleration actuators are located between the truck and carbody as shown
magnitude are decreased. Increasing the gain Ca has precisely in Fig. 5. The pair of lateral actuators can be operated in
the same dynamic effect as increasing the vehicles' mass phase to produce lateral forces or out of phase to produce yaw
without the static disadvantages, e.g., increased fuel con- torques, thus the only forces and torques considered in this
sumption. paper are:
Figure 2 shows an "optimal" [10] one-degree-of-freedom
suspension response using a passive spring and an active FAC = —FAT=FA1 +FA2 (1)
damper, i.e. in Fig. 1, b = Ca = 0. TAc = ~TAT=l(FAi -FAl) (2)

Journal of Dynamic Systems, Measurement, and Control MARCH 1982, Vol. 104/101
Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab
 where FAc is the total active force acting on the carbody from
a single truck, FAj is the reaction force on the truck, FA and
FAl are the forces from the two actuators and / is the distance
between them. Secondary Hfl
!
Suspension
Vehicle Model. A well documented fifteen degree of
—vertical primary
freedom linear lateral model [1] with random track alignment Primary
Suspension tP
and cross-level inputs was used in this paper. Figure 6 is a
schematic diagram that illustrates the modeling nomen-
clature. The degrees of freedom include wheelset lateral and
i L>^=%
yaw, truck lateral and yaw and carbody lateral, yaw, and roll.
(a) Modeling of vertical suspensio.

Track Input Model. The vehicle was assumed to be

disturbed by random lateral alignment and cross-level inputs
as described in reference [1]. The following analytical spectral
density forms have been shown [1] to be representative of Wheelset 1

typical North East Corridor track. Truck 1

0.00159(0.251)2 Wheelset 2
5 0 (Q) = (3)
O 2 (fi 2 +(0.251) 2 ))
0.00159(0.251)2
S c (0) = (4)
(n 2 +(0.251) 2 )(fl 2 +(0.133) 2 )
Wheelset 3
The numbers in equations (3) and (4) have been chosen to be
representative of class 6 track and Sa is the alignment input Truck 2

(in 2 /rad/ft), Sc is the cross-level input (in 2 /rad(ft), andflis Wheelset 4

the spatial wave number (rad/ft).
Performance Indices
(b) Modeling of l a t e r a l and longitudinal
Ride Quality. The lateral ride quality was evaluated by 1/3 suspension

octave band analysis of the lateral acceleration spectral Fig. 6 Vehicle model
density according to the ISO/2631 criteria [15]. In particular,
the model output was compared to the one-hour reduced of actuators per car. The average power which must be
comfort boundary. The available room for the lateral supplied to a typical hydraulic or pneumatic actuator can be
suspension deflection was taken to be 1.5 in., thus the approximated by [12]:
maximum rms secondary stroke was set at one third of this PAV = F.V (6)
value (0.5 in.).
All signals in the analysis were assumed to be Gaussian, where F is the peak force and Kis the rectified average piston
thus even though the 3 a value of the secondary stroke length velocity. The actuator force and piston velocity were each
was set at 1.5 in., this value will still be exceeded a finite assumed to be Gaussian signals, thus the peak force can be
amount of the time. In order to approximate how often the approximated by:
"bump stops" would be contacted, the frequency of ex- F=3oF (7)
ceeding 1.5 inches was calculated by the expression for
Gaussian signals due to Rice [16], i.e., where aF is the rms actuator force. The rectified average
piston velocity can be expressed as [12]:
1 E[x2] CL V=\/27*ov (8)
Fc= — (5)
2
2TT E[X ] 2E[x2 where av is the rms piston velocity. The average power is then,
where Fc is the frequency (Hz) of exceeding the clearance PAy = 3<rFy/2/vaB (9)
value, CL, and x, and x are the stroke and stroke velocity.
The rms power about the mean can be expressed as [12]:
Stability. Several parameters have been used in the
literature to represent the lateral stability characteristics of a
Pim=J(T+U?joFou (10)
rail vehicle. Traditionally the critical speed, Vcr, is used and is where:
defined to be the first forward speed of the vehicle at which E[Fv]
zero damping occurs in one of the vehicles' dynamic modes. PA-'E^Eiv2]
Alternatively the damping in the least damped mode at the
operating speed has been used where the operating speed is The peak power deviation from the average value was
generally defined as a safety factor below the critical speed. assumed to be three times P r m s .
Reference [1] has shown that both of these definitions are
essentially equivalent for conventional rail passenger vehicles. Parametric Studies
In this paper we have used the critical speed, Vcr, as our
stability parameter and have defined the operating speed as The baseline conventional passive vehicle was chosen to be
the maximum speed the vehicle can travel at which the ride an optimized design [2]. Table 1 lists the parameters for the
quality and stroke lengths are still within the specified optimized vehicle whose parameters are typical of an AM-
bounds. TRAK intercity passenger vehicle. The passive vehicle has a
critical speed of about 200 mph and can satisfy the ISO one
Power Consumption. A very important consideration in the hour reduced comfort boundary with less than a 0.5 in. rms
evaluation of any active system is the amount of power secondary stroke on class 6 track for speeds up to 90 mph.
consumed and the peak force required. Estimates of the total The baseline vehicles' stability/curving trade-off has been
power consumed per car were made by considering the forces optimized so that it can negotiate, at balance speed, a constant
and velocities of one actuator and multiplying by the number 1.3 degree curve without flanging [2]. The active suspension

102/Vol. 104, MARCH 1982 Transactions of the ASME

Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab

 Table 1 Baseline lateral vehicle dynamics model parameters
Wheel Rail Parameters Icy carbody pitch moment of inertia 1.765X10" slug-ft2
on wheelset roll coefficient 0.0625 (2.391 X10 6 kg-m 2 )
80 centered contact angle 0.0625 Icx carbody roll moment of inertia 76,340 slug-ft2
A contact angle difference coefficient 4.5 (103,500 kg-m 2 )
/i i lateral creep force coefficient 2.24xl0 6 lb Mc carbody mass 2,860 slugs
(9.98xl0 6 N) (41,740 kg)
/12 lateral/spin creep coefficient 17525 ft-lb Primary Suspension Parameters
(23760N-m) kPy primary lateral stiffness (4 per truck) 500,000 lb/ft
/22 spin creep coefficient 375 ft2-lb (7,296,958 N/m)
(154.97 N-m 2 ) kpx primary longitudinal stiffness 1,000,000 lb/ft
/33 longitudinal creep force coefficient 2.520 x 106 lb (4 per truck) (14,593,917 N/m)
Wheel Parameters lb
2000
a half of track gauge 2.35 ft ft/s
(0.716 m)
W load per wheelset 30.883 lbs (29.187 )
(137.152 N)
0.1 V m/s/
X wheel conicity lb
120. slugs primary longitudinal damping (kpx/250) 4000
Mw wheelset mass (1751. kg) ft/sec
561 slug-ft2 NX
Iwx wheelset roll moment of inertia (761 kg-m 2 ) 58.375
112 slugs-ft2 m/s/
/„,_,, wheelset pitch moment of inertia (152 kg-m 2 ) Secondary Suspension Parameters
561 slugs-ft2
Iwz wheelset yaw moment of inertia (761 kg-m 2 ) ksy secondary lateral stiffness (2 per truck) 24000 lb/ft
1.5 ft (350,254 N/m)
R0 centered wheel rolling radius (.4572 m) kz secondary vertical stiffness (2 per truck) 22200 lb/ ft
Arspsi secondary yaw stiffness (1 per truck) 1,000,000 ft-lb/rad
Truck Parameters (1,355,818 N-m/rad)
dp 1 /2 of primary springs lateral spacing 2.0 ft lb
csy secondary lateral damping 1200
(.06096 m) (2 per truck) ft/s
b 1/2 of wheelbase 4.25 ft
(1.295 m) 17,512
h,p vertical distance from truck e.g. 0.35 ft - )
to primary lateral springs (.1067 m) m/s/
h,s vertical distance from truck e.g. 1.68 ft lb
to secondary lateral springs (.6511m) secondary vertical damping 2000-
(2 per truck) ft/s
I,x truck frame moment of inertia 2058 slug-ft2
(2790 kg-m 2 )
Ily truck frame pitch moment of inertia 434 slug-ft2 (29,187 — )
(588 kg-m 2 ) V n/s/
Ilz truck frame yaw moment of inertia 2486 slug-ft2 ft-lb
(3371 kg-m 2 ) cspsj secondary yaw damping 16000-
M, truck frame mass 250 slugs (1 per truck) rad/s
(3649 kg) / N-m \
21,693 )
Carbody Parameters V rad/s /
ds 1 /2 of secondary spring lateral spacing 3.75 ft Passenger Locations for Ride Quality
(1.163 m) Front distance from jc = 29.75ft z = 3.75 ft
hcs vertical distance from carbody e.g. 2.94 carbody e.g. (9.06 m) (1.14 m)
to secondary lateral springs (.8961 m) Rear x= -29.75 ft s = 3.75 ft
ls 1 /2 of truck center pin spacing 29.75 ft
(9.068 m) ( - 9 . 0 6 m) (1.14 m)
lc carbody total length 25.91 m Performance Capabilities of Baseline Vehicle
w weight of carbody 92,006 lbs
(409.263 N) Critical speed 197 mph
Icz carbody yaw moment of inertia 1.764 x l O 6 slugs-ft2 Operating speed 90 mph
(2.341 x l O 6 kg-m 2 ) Curving 1.3 degree curve without flange contact

to be defined in the following sections will only use rate obtained by integrating and high pass filtering the ac-
feedback terms thus the addition of the active components celeration measurements to eliminate drift errors and low
will not change the vehicles' steady-state curving per- frequency curving inputs. It was also decided to limit the
formance. accelerometers to truck and carbody locations since axle
Active Control of Hunting. It is well known that the mounted accelerometers could lead to severe maintenance and
damping in the kinematic or "hunting" mode decreases with reliability problems.
increasing forward speed. Figure 7 shows the damping ratio In order to determine an effective control scheme several
of several lateral vehicle modes as a function of forward speed feedback laws were considered where the lateral force and
for the baseline passive vehicle. Modes 1 and 2 are kinematic yaw torque were allowed to be linear functions of truck and
modes with mode 1 (in phase wheelset lateral) going unstable carbody linear and angular velocities and accelerations in-
at 197 mph. Modes 3, 4, and 5 are suspension modes and are cluding lateral, yaw, and roll terms. Numerous parametric
not affected by speed. studies, which are outlined in reference [17], were performed
As stated earlier the proposed actuator configuration is to determine the most effective and simplest form to control
shown in Fig. 5 and can produce the lateral force and torque the kinematic mode.
given by equations (1) and (2). For reliability and ease of It was found that the lateral force actuators could be used
maintenance it was decided to limit the sensors to state-of-the- to influence the kinematic mode but that they also had a
art, rugged accelerometers, thus our control laws will consider strong influence on carbody motions and led to ride quality
only velocity and acceleration signals. Velocity terms can be problems. This occurs due to the large yaw torque that the

Journal of Dynamic Systems, Measurement, and Control MARCH 1982, Vol. 104/103

Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab

 front and rear lateral actuators can exert on the carbody since
the moment arm between them is quite large. A pure torque
on each truck also influences the kinematic mode without the
harmful effect on the carbody modes, thus the stability
controller chosen, had the form:
TA=k\y,+k2^, (11)
Figure 8 shows how the damping in the kinematic mode is
changed by each gain. Note that increasing the yaw rate gain,
k2, always results in more damping in the kinematic mode
while too high a value of the lateral rate gain, kt, can lead to C r i t i c a l Speed
instability.
Since these control laws are stabilizing the wheelsets by 120 140
forces and torques applied through the primary suspension it Spiked, V, mph
is important to determine how changes in the primary
Fig. 7 Change in modal damping with increases in speed for baseline
suspension influence the results. vehicle
As was expected it was found that the control effectiveness
decreased as the primary suspension became softer. Of the
two primary stiffnesses, softening the longitudinal stiffness
had a much stronger negative effect than the lateral stiffness.
The baseline design had a primary longitudinal stiffness of
kpx = 106 lb/ft. It was found that below values of kpx = 5 x
105 lb/ft the active suspension had very little influence on the
kinematic mode.

Active Control of Ride Quality. The forced response of the

vehicle to random track alignment and cross-level inputs (as
defined by equations (3) and (4)) was computed using 10 100 k.
frequency domain spectral density analysis [1]. Figure 9
shows the lateral acceleration spectral density of the baseline a) Gain k.,=10,000

passive vehicle at 100 and 185 mph. At 100 mph most of the
rms acceleration is in the 1 Hz region which is controlled by Fig. 8 Changes in modal damping caused by active truck feedback at
the secondary lateral suspension. At 185 mph significant 185 mph
power appears both in the 1.0 Hz region and also in the 5 Hz
region which is the kinematic frequency. From Fig. 9 it can be
concluded that improving the ride quality of the vehicle at
speeds a significant factor below the critical speed requires
improvement in the lower frequency region (less than 2 Hz)
while improvement at speeds close to the critical speed will
require reductions in the spectral density out to higher
frequencies (7 or 8 Hz). The stability controller defined in the
previous section is effective in controlling the kinematic mode
thus the next section will look at the use of active control to
improve the lower frequency carbody motions. It should be
mentioned that passive means to improve the ride quality were
ineffective since any softening of the lateral suspension
resulted in rms secondary strokes greater than 0.5 in. and any
increases in the damping resulted in greater rms accelerations.
It was concluded in the previous section that the lateral
force actuators had a significant influence on carbody motion
while the torque actuators did not, thus the lateral active
Frequency 1Hz]
forces were used to alter the low frequency carbody motion.
Numerous control schemes were analyzed and are presented Fig. 9 Acceleration spectral density of the front passenger location
on the baseline vehicle at 100 and 185 mph
in reference [17], these included measuring all truck and
carbody velocities and accelerations. The two major types
considered were a "modal" controller [3] and a "local"
controller [4]. The modal controller needs enough FA=k3yd + k4yd (12)
measurements to be able to define carbody lateral, yaw, and where ycl and yd are the local lateral carbody velocity and
roll modes and to coordinate the active forces to in- acceleration. In this paper the sensors were assumed to be
dependently control each carbody mode. The local controller placed on the carbody over the truck e.g., thus:
uses only a lateral accelerometer above each truck e.g. and
feeds this signal to the lateral force actuators. The major (13)
advantage of the modal controller is its greater versatility in
controlling the vehicle, the major advantage of the local where ls, and Hcs are defined in Fig. 6, and yc,4/c, and (j>c are
controller is its simplicity. The parametric studies presented in the lateral, yaw and roll of the carbody.
reference [17] show that the local controller was nearly as A parametric study was made to determine the appropriate
good as the more general modal controller and thus will be gains, k}, and &4, for improving the ride quality at 100 mph,
presented in this paper. Table 2 shows the performance of the controlled and un-
The following control law was assumed for the local ride controlled vehicle. The rms acceleration has been reduced by a
controller: factor of three at a cost of an average power consumption of

104/Vol. 104, MARCH 1982 Transactions of the ASME

Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab

 Table 2 Ride quality of controlled and uncontrolled vehicle io~"
at 100 mph
k3 = 12,000 lb-s/ft, Ar4 = 1000 lb-s 2 /ft o- Passive
A - Active
Fore rms Fore rms secondary Average power per — ISO Standard
Acceleration stroke car »
[g] [in.] [kw] g 10-
passive 0.074 0.50 0.
active 0.025 0.41 6.4 o o

o A A
Table 3 Vehicle performance at 185 mph*
Fore rms Fore rms secondary Average power per
Acceleration stroke car
[g] [in.] [kw]
passive .23 1.0 0.
active .078 .45 64

•The control gains selected were

k{= 200,000 lb-s, •• 20,000 ft-lb-s Frequency fHz]
*2 =
2
* 3 = 12,000 lb-s/ft k4 == 1000 lb-s /ft Fig. 10 1/3 octave band total rms lateral accelerations for the un-
controlled and selected ride controlled vehicle at 100 mph (front
passenger location)
Table 4 Expected bumpstop contact frequency
Speed Contact frequency
[mph] [bumps/minute]
Passive 100 .210
Ride control 100 .115
uncontrolled
only
controlled
Dual control 100 .007 . ISO standard
Passive 185 8.50
Dual control 185 .34

6.4 kw per car and a peak actuator force of 2700 lb. Figure 10
shows the 1/3 octave band accelerations for the controlled
and uncontrolled vehicle. It is clear that the ride controller is
acting primarily in the 1-2 Hz frequency band.
A sensitivity study was made to determine the performance
changes of the ride controller to variations in the vehicle
parameters. It was found the controller was fairly insensitive
to parameter changes, e.g. a 20 percent change in carbody Frequency [Hz]

masses and inertias produced less than a 7 percent change in Fig. 11 1/3 octave band total rms accelerations for the uncontrolled
the rms acceleration. and the combined kinematic and ride controlled vehicle at 185 mph
From Fig. 9 it is clear that in order to improve the ride (front passenger location)
quality of the vehicle at 185 mph both the ride and kinematic
controllers will be required. Thus both the controllers defined
by equations (11) and (12) were used at 185 mph. Table 3 evaluated by parametric studies using a fifteen degree-of-
summarizes the performance of the passive and controlled freedom lateral model. An optimized conventional passenger
vehicle. Clearly the ride controller by itself would be inef- vehicle was used as the baseline vehicle. The two significant
fectual at solving the problem, whereas the combined ride and lateral modes that need to be controlled are the lateral car-
kinematic controller or "dual" controller is able to meet the body mode ( = 1 Hz) and the truck kinematic mode whose
performance specifications with an average power con- frequency is dependent on the vehicles' forward speed.
sumption of about 64 kw per car. Figure 11 shows the 1/3 It was shown that a very efficient means of controlling the
octave band accelerations for the uncontrolled and dual carbody lateral mode is to use a "ride controller" of the form,
controlled vehicle, clearly the ride is improved at both the PA = kT,yci + ka,yd, where FA is a lateral force produced by
carbody lateral natural frequency (1 Hz) and the kinematic the in phase motion of a pair of lateral actuators between each
frequency (5 Hz). truck and the carbody and ych and yd are local carbody
lateral velocity and acceleration. It is shown that at speeds
The average "bumpstop" frequency of contact was
well below the vehicles' critical speed that this controller can
computed by equation (5). The nonlinear behavior of the
significantly improve the ride quality. It is also shown that
bumpstop has not been included in this analysis, thus
fairly low bandwidth actuators (less than 3 Hz) consuming
equation (5) is only a very approximate indicator of this
small amounts of power (1.5 kw/actuator) are quite suc-
behavior. Table 4 summarizes the effective bumpstop
cessful at speeds up to 100 mph on class 6 track, thus
frequencies at 100 and 185 mph for the proposed con-
pneumatic actuators might be adequate.
figurations. As can be seen the dual controller is very effective
at reducing the frequency of contact while the ride controller In order to increase the vehicles' critical speed or to obtain
is only moderately effective. reasonable ride quality at speeds close to its critical speed, it is
necessary to control the trucks' kinematic mode. It is shown
that an efficient means of achieving kinematic mode control is
Conclusions to use a controller of the form, TA = kly, + k2 4/t where rA is
The potential stability and ride quality improvements a torque produced by the out of phase motion of the pair of
offered by simple lateral active suspensions have been lateral actuators on each truck and y, and \f/, are the truck

Journal of Dynamic Systems, Measurement, and Control MARCH 1982, Vol. 104/105

Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab

 lateral and yaw velocities. It is shown that this form of active 3 Goodall, R. M., Williams, R. A., and Lawton, A., "Practical Ap-
plications of Actively Controlled Suspensions to Railway Vehicles," presented
controller is very effective at controlling the kinematic mode at the 1979 ASME Winter Annual Meeting, New York, 1979.
without affecting the carbody lateral mode. The combined 4 Jindai, K., et al., "Fundamental Study on Semi-Actively Controlled
ride and kinematic controllers can be used to provide ac- Pneumatic Servo Suspensions for Rail Cars," presented at the 1981 Joint
ceptable ride quality at speeds close to the critical speed at the IEEE/ASME Railroad Conference, Atlanta, Ga., May 1981.
expense of larger bandwidth actuators (=7 Hz) and increased 5 Mauer, L., "Improvement of Railcar Vehicle Performance-State-of-the-
Art Report," U.S. German Cooperation Project (MAN-MIT) EDS-022, Oct.
power consumption (= 15 kw per actuator at 185 mph). 16, 1979.
There were a number of idealizations made in this paper 6 Santanera, O., "FS and Renfe Try Out Tilt-Body Prototype Trains,"
that will have to be addressed in future work. First, a linear Railway Gazette International, Mar. 1974.
7 Babcock, D., and Newman, M., "The Advanced Passenger Train,"
dynamic analysis was made thus, the effects of flange contact, Railway Division, 1976.
creep force saturation and suspension nonlinearities were not 8 Hedrick, J. K., "Railway Vehicle Active Suspensions," State-of-the-Art
included. These effects are not anticipated to be large for well Papers, 7th IA VSD/IUTAM Symposium, Vehicle System Dynamics, Sep-
maintained passenger vehicles operating on reasonably tember 1981.
smooth track. Second, the actuators were assumed to have 9 Cox, J. J., Hedrick, J. K., and Cooperrider, N. K., "Optimization of Rail
Vehicle Operating Speed with Practical Constraints," ASME JOURNAL OF
infinite bandwidth, the effects of finite bandwidth pneumatic, DYNAMIC SYSTEMS, MEASUREMENT AND CONTROL, Vol. 100, No. 4, Dec. 1974.
hydraulic and/or electric actuators are currently being in- 10 Hedrick, J. K., and Wormley, D. N., "Active Suspensions for Ground
vestigated. It is anticipated that bandwidth limitations will not Transport Vehicles-A State-of-the-Art Review," Mechanics of Transportation
considerably influence the ride controller but could be im- Systems, AMD-Vol. 15, ASME, 1975, pp. 21-40.
11 Jeff'coat, R. L., and Wormley, D. N., "Improvement of Rail Vehicle
portant for the kinematic mode controller. Third the response Lateral Dynamics, Performance Through Active Control," Vehicle System
to curves was covered in a quasi-static sense but the dynamic Dynamics, Vol. 4, No. 2-3, 1975, pp. 169-173.
effects of curve entry/exit on suspension stroke length were 12 Sinha, P. K., Wormley, D. N., and Hedrick, J. K., "Rail Passenger
not investigated and will be the subject of future studies. Vehicle Lateral Dynamic Performance Improvement Through Active Con-
trol," ASME JOURNAL OF DYNAMIC SYSTEMS, MEASUREMENT AND CONTROL,
Vol. 100, Dec. 1978, pp. 270-283.
Acknowledgments 13 Sarma, G. N., and Kozin, F., "An Active Suspension System Design for
the Lateral Dynamics of a High Speed Wheel-Rail System," ASME JOURNAL OF
This work was supported by the U.S. D.O.T., Office of DYNAMIC SYSTEMS, MEASUREMENT AND CONTROL, Vol. 93, No. 4, Dec. 1971.
University Research under contract DTRS5680-C-00018. 14 White, R. C , and Cooperrider, N. K., "Guideway-Suspension Trade-
offs in Rail Vehicle Systems," ASME JOURNAL OF DYNAMIC SYSTEMS,
References MEASUREMENT AND CONTROL, Vol. 103, September 1981, pp. 237-244.
15 "Guide for the Evaluation of Human Exposure to Whole-Body
1 Hedrick, J. K., et al., "Performance Limits of Rail Passenger Vehicles: Vibration," l.S.O. 2631, International Organization for Standardization, 1977.
Conventional, Radial and Innovative Trucks," Final Report, U.S. DOT 16 Rice, S. O., "Mathematical Analyses of Random Noise," Bell Systems
Contract DOT-OS-70052, Sept. 1980. Tech., Vol. 23, 1944, pp. 282-332. Vol. 24, 1945, 45-156, reprinted in N. Wax,
2 Horak, D., Bell, C. E., and Hedrick, J. K., "A Comparison of the ' 'Selected Papers on Noise and Stochastic Processes,'' Dover, New York, 1954.
Stability and Curving Performance of Radial and Conventional Rail Vehicle 17 Celniker, G. W., "The Application of Active Control to Rail Vehicles,"
Trucks," to be published in ASME JOURNAL OF DYNAMIC SYSTEMS, M.S. thesis, Department of Mechanical Engineering, M.I.T., Cambridge, MA
MEASUREMENT AND CONTROL, Vol. 103, September 1981, pp. 191-200. 02139, USA, Jan. 1981.

106/Vol. 104, MARCH 1982 Transactions of the ASME

Downloaded From: http://dynamicsystems.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jdsmaa/26070/ on 05/08/2017 Terms of Use: http://www.asme.org/ab