Active-suspension Design for a Special Road-Rail

Vehicle Based on Vehicle-Track Coupled Model

Using Genetic Algorithm
Roohollah Talebitooti
School of Mechanical Engineering, Iran University of Science and Technology, Tehran, Iran.

Reza Bayat
School of Automotive Engineering, Iran University of Science and Technology, Tehran, Iran.

(Received 8 April 2015; accepted 13 April 2016)

In this paper, a PID controller for a special road-rail vehicle is developed. The dynamic model is derived to
properly consider the vehicle and track vibrations. This model contains the effects of track elasticity which makes
it more reliable and precious compared to traditional models. The vehicle model contains a two dimensional 10
DOF vehicle model, and the track model consists of 40 sleepers. In addition, the effect of the sleeper number on
car-body vibration is investigated. The results show that the present vehicle-track coupled model is more efficient
in controller design compared to traditional ones. Finally, tuning of controller gains is performed with the aid of a
genetic algorithm, in order to achieve a well-organized active suspension.

NOMENCLATURE trip, using active suspension systems in these vehicles has be-
come one of the most effective technologies in the transporta-
C1 Damping coefficient of primary suspension.
tion industry.
C2 Damping coefficient of secondary suspension.
CF Cost function in genetic algorithm. So far, there exist mainly two theories on railway vehicle dy-
e Control error. namics. One is the traditional theory of railway vehicle dynam-
F Force matrix of system. ics,1 and the other is the vehicle-track coupled theory.2 The
Fi , Fj Shear forces exerted on nodes i and j. traditional theory of railway vehicle dynamics usually focuses
Ib Half-mass moment of inertia of bogie. on the railway vehicle itself without considerations of the dy-
Ic Half-mass moment of inertia of car-body. namic behavior of the track. In the vehicle-track coupled the-
K1 Stiffness coefficient of primary suspension. ory, the track is treated as an elastic structure. Vibrations of the
K2 Stiffness coefficient of secondary suspension. vehicle can be transmitted to the track via the wheel-rail con-
l1 Semi-longitudinal distance between bogies. tact and excite vibrations of the elastic track structure, which
l2 Semi-longitudinal distance between can in reverse influence the vibrations of the vehicle. There-
wheelsets in bogie. fore, the vibrations of the vehicle and the track are essentially
M Mass matrix. coupled with each other.2
Mb Half-mass of bogie. For various research purposes, different types of models
Mc Half-mass of car-body. have been presented. To study the feasibility for improving
Mi , Mj Moments about z-axis exerted ride quality using magnetorheological dampers, a nine DOF
on nodes i and j. model of railway vehicles was developed, including vertical,
Mw Half-mass of wheelset. pitch and roll motions of car-body and trucks.3 Also, a 17 DOF
MO Maximum overshoot of controller. model of a semi-active suspension system was used to improve
N Total number of rail-sleeper supporting the ride quality on train.4 In a study done by Sezer and Atalay,5
points. a 54 DOF model was established to design a fuzzy controller
NJ Number of rail nodes. to reduce the vertical, lateral, and angular vibration of a rail
q Time-variant state vector. vehicle containing a body, three bogies, and six axles. Track
u Actuator force. models presented in the aforementioned papers were consid-
V Vehicle forward speed. ered rigid.
Zb Displacement of bogie. Investigating safety limits against derailment, several mod-
Zc Displacement of car-body. els have been proposed to consider the track vibration. A cou-
Zr Displacement of rail. pling model of vertical and lateral vehicle-track interaction was
Zw Displacement of wheelset. proposed by Zhai et al.6 in which the vehicle subsystem was
θb Pitch angle of bogie. demonstrated as a multi-body system with 37 DOF, and the
θc Pitch angle of car-body. track substructure was modeled as a discretely supported sys-
tem of elastic beam. In addition, the vibration of the train as
1. INTRODUCTION well as the track was discussed, where the rail was modeled
based on the finite element method.7 Another similar work
Rail transportation is one of the most common forms of was done by Uzzal et al.,8 in which the dynamic model was
transportation in the world. In order to provide a comfortable presented with the aid of partial differential equations. More-
International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018 (pp. 9 15) https://doi.org/10.20855/ijav.2018.23.11063 9
R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

Figure 1. Beam elements for the rail structure.

over, in order to review the vibration of a vehicle and track to

a single rail irregularity, Zakeri et al.9 presented general equa-
tions of a vehicle-track coupled system in a matrix form using
finite element method. Furthermore, a road-rail fire-fighting
vehicle was considered to demonstrate the effectiveness of the
vehicle-track coupled model.11
As mentioned in the above literatures, most studies have ex- Figure 2. Vehicle-track coupled model.
amined the vibration of railway vehicles, and the track is as-
sumed as a rigid body that may lead to significant errors in the is placed at the center of the track. Finally, the corresponding
mathematical models; therefore, it is inaccurate. In some of vibration equation of the vehicle-track coupled model shown
the literatures reviewed above, the vehicle-track coupled model in Fig. (2) can be written in standard form as:
has been used. However, the sheer stress component between
the ballast masses is completely ignored. Although both of M q̈ + C q̇ + Kq = F. (1)
these assumptions have been considered in a few of the stud-
The matrices M , K, and C are assembled according to (2) as
ies, no active suspension is involved there.
follows:
In this paper, an active suspension system is applied to a  
vehicle-track coupled model in order to control the vertical Carbody 0 0 0
and pitch vibrations of the car-body. This model includes a  0 Rail R/S 0  ; (2)
10 DOF model for the vehicle and a track, including a rail, T

 0 R/S Sleeper S/B 
a sleeper, and a ballast, as well as consideration of the sheer 0 0 S/B T Ballast
stress components of the ballast. A new code is developed in
MATLAB to study the vibration response of the system. More- where the sub-matrices R/S, S/B represent the interaction
over, an active PID controller is designed to reduce the vibra- between rails and sleepers, sleepers and ballast respectively.
tion. Tuning the corresponding parameters of the controller, a In Fig. (2), it is apparent that dynamic contact force between
genetic algorithm code is established. The results show that wheel and rail is assumed as non-linear equation.9
using the active controller reduces the acceleration of the car-
body by more than 60%. Furthermore, the effects of the vehi- 3. NUMERICAL PROCEDURE
cle speed on efficiency of the active controller indicate that the
controller performance is gradually reduced by increasing the A numerical code is developed in MATLAB to investigate
vehicle speed. the vibration of vehicle-track interaction. It should be noted
that the specifications of the vehicle and track are the same as
2. MODELLING what is used by Li et al.10 Matrices of the mass, stiffness and
damping are square, and all have 10 + 2N J + 2N elements,
The vehicle system model used in this study consists of a where N J is the number of rail nodes on which an external
car-body supported by two bogies, each of which has two ax- force is exerted, and N is the number of sleepers. So, N J is N
els. Without considering the yaw and roll motions of the car- plus four points that wheels touch rail. Figure (3) shows the ef-
body and bogies, the total DOF of the model is 10: the vertical fect of these parameters on the vibration of the car-body. Also,
displacement of the car-body, the pitch angle of the car-body it can be depicted that by increasing the number of sleepers and
with respect to its center of mass, the vertical displacements of nodes, the acceleration response is getting more improved into
bogies, the pitch angles of bogies with respect to their center of real one. It also illustrates that by increasing the number of
mass, and the vertical displacements of wheel sets. In this pa- sleepers and nodes, the dominant period of response increases.
per, vehicle equations proposed by Li et al.10 were used which Since displacement diagrams for the higher number of sleepers
are deduced according to the Newton’s laws of motion. and nodes have less concavity, it can be concluded that increas-
Based on the model presented by Li et al.,10 it is assumed ing the number of sleepers and nodes results in the decrease of
that the track includes a rail mass, sleepers, and ballast masses. the car-body acceleration. Decreasing the car-body accelera-
With the aid of finite element method, the rail model is estab- tion can be expressed by the fact that by adding the length of
lished with applying the Euler-Bernoulli beam element type the rail, the effect of the boundary condition is diminished, and
presented by Liu11 (Fig. (1)). Other track equations related to the track is more reasonable in conditions.
sleepers and ballast masses are utilized from the track equa- Since overloads on the wheel are predicted when the rigid
tions presented by Li et al.10 track (traditional model) is considered, further acceleration
The vehicle and track coupled model is shown in Fig. (2). It (concavity of response) on the car-body is observed. In order
is proved in Section 3 that the rail is long enough to neglect the to take away from the traditional model, the number of sleepers
boundary effects, so every deflection in both ends of the rail is improved in the presented model. Therefore, the track model
is assumed to be zero. It should be mentioned that the vehicle becomes softer, and the applied force between wheel and track

10 International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018

 R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

Figure 5. Comparison between theresponse of a 10 DOF model and vehicle-

Figure 3. The effect of sleeper number in car-body vibration. track coupled model with a high stiffness track.

Figure 6. Wheel-rail contact force of current model and the model presented
Figure 4. Surface irregularity.
by Uzzal et al.8

approaches the actual one. In other words, the overestimated tact force between the wheel and track falls, due to onset of
acceleration on the car-body is removed by omitting the over- contacting the wheel flat irregularity with the track. Then, the
loads in the present model. As mentioned above, choosing the contact force increases because the wheel suddenly impacts the
values of N and N J as 40 and 44, respectively, leads to a reli- track. The dominant period of oscillation of the contact force
able model. is approximately 10 milliseconds. However, a slight difference
between two models was observed 30 milliseconds later, due
4. VERIFICATION OF THE MODEL to the differences between finite element method and partial
differential solution. This difference influences computation
The present model is verified here with the aid of two view- of the damping matrix. In other words, it slightly affects the
points. Firstly, the body response of the present model to an ir- wheel-track damping force. Nevertheless, this has a negligible
regularity shown in Fig. (6) is obtained assuming a great num- effect on car-body vibration.
ber for the stiffness of the track. Then, the response is com- Another comparison has been done using the 10 DOF vehi-
pared with that of the model reported by Li et al.10 As depicted cle model used by Sun and Dhanasekar,12 as well as the exper-
in Fig. (7), an excellent agreement is easily observable. imental work performed by Newton and Clark.13 In the work
Secondly, the present model is compared with the models done by Sun and Dhanasekar,12 a track model with four-layer
presented by Uzzal et al.,8 and Sun and Dhanasekar.12 In the is considered, and the Hertz contact coefficient is deduced from
work done by Uzzal et al.,8 dynamic analysis of a railway vehi- Johnson.15 However, in present work, the recent model of con-
cle moving on an elastic track is performed. A five DOF model tact phenomenon presented by Zhai2 is developed. It can be
is assumed for the vehicle that contains vertical displacement observed from Fig. (7) that when the wheel flat irregularity
of the vehicle car-body, bogie‘s pitch and vertical displace- touches the track, the contact force reduces to zero. This means
ment, and vertical motion in two axles. The track model is the wheel and the rail separate for a while. When they meet
similar to the one previously depicted in Fig. (2). Using the ve- again, an enormous peak force is induced between them. This
hicle and track characteristics of that model into present model, phenomenon is reasonably predicted in these three methods.
the results of this comparison are illustrated in Fig. (6). Both numerical models show some disagreements with the ex-
As shown in this figure, there is a good agreement between perimental data. This is due to the discrepancy that occurred as
these two models. In other words, in both models, the con- a result of disagreement between the real and calculated damp-
International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018 11
R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

Figure 8. Comparison between the car-body responses of a 10 DOF model

and vehicle-track coupled model.
Figure 7. Wheel-rail contact force of current model and the model presented
by Sun and Dhanasekar,12 and Newton and Clark.13

ing values of track components. In addition, the results of Sun

and Dhanasekar12 give the maximum force much higher than
that of the experimental model,13 whereas the present work
gives the maximum force much more reliable.
These two comparisons confirm the reliability of the present
model in simulating the vehicle- track coupled dynamics, es-
pecially for designing the controller, in which the reduction of
maximum contact force is of high importance.

5. COMPARING VEHICLE-TRACK COUPLED

MODEL WITH 10 DOF MODEL REGARD- Figure 9. Schematic of present secondary active suspension system.
ING VEHICLE BODY RESPONSE
proves the car-body dynamic response and provides a better
This section discusses the importance of modelling accu- isolation to the track irregularities compared to using only pas-
racy in controller designing. At first, in order to illustrate the sive springs and dampers. Actuators can be replaced with the
importance of considering track vibrations, car-body response passive suspensions, and the suspension behavior will be com-
of the 10 DOF model (traditional model) was compared to the pletely controlled via active means. In practice, however, it
response of the vehicle-track coupled model (present model). is more beneficial to use actuators in conjunction with passive
Then, in the next section, suitable PID controller gains are cal- components.15
culated based on the vehicle-track coupled model using the ge- Figure (9) shows a schematic of active components. In or-
netic algorithm. der to ease mathematic equations, the front and rear actuators‘
Applying the irregularity shown in the Fig. (4) to the mod- forces were broken into two components u1 as vertical force
els, vertical displacement of the car- body for the traditional and u2 as pitch moment.
and the present model was obtained, as illustrated in Fig. (8). In this model, the actuator forces are expressed as:
Significant differences are observed comparing the results of
these two models. This is due to the fact that in the traditional u1 + u2 = uf ; (3)
model, the stiffness of the track is not considered. In present
work, as the elasticity of the track is fully considered, it can u1 − u2 = ur ; (4)
be imagined that the vehicle travels on a softer bed. Appar-
where ur and uf are the rear and front actuators‘ forces.
ently, applying spring elements to a vibration system in series-
form causes decreasing stiffness; consequently, oscillations de- Because of using actuators in the suspension system, the
crease, and thus displacements increase. Moreover, the ampli- vertical force and moment exerted on the car-body and bogies
tude of the acceleration of car-body (concavity of displacement should be revised. Equations (5) and (6) are the revised form of
graph) decreases. In other words, it can be seen that present the car-body and bogies equation proposed by Li et al.10 with
model differs from the 10 DOF model in a significant quan- consideration of the actuators‘ forces. Equations of vertical
tity; therefore, the results of the present model is of a higher displacement and pitch motion of car-body are as follows:
accuracy to design PID controller for the road-rail vehicle.
Mc Z̈c + 2C2 Żc − 2K2 Zc − C2 Żb1 − K2 Zb1

6. ACTIVE SUSPENSION SYSTEM −C2 Żb2 − K2 Zb2 = 2u1 ; (5)

6.1. Active Suspension System Modeling

Ic θ̈c − 2C2 l12 θ̇c + 2K2 l12 θc − C2 l1 Żb1 − K2 l12 Zb1
Using active secondary suspension is an efficient way to re-
duce car-body vibrations.14 Secondary active suspension im- +C2 l1 Żb2 + K2 l1 Zb2 = 2u2 l1 ; (6)

12 International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018

 R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

where M c is the half-mass of the car-body, Ic is the half- Table 1. Genetic algorithm specifications.
mass moment of inertia of the car-body, K and C represent Parameter Characteristic value
Number of variables 3
the stiffness coefficient and damping coefficient of suspension
Population Size 20
elements, and the subscripts 1 and 2 are used to identify pri- Crossover fraction For u1 For u2
mary and secondary suspension elements. 0.8 0.4
Respectively, the vertical displacement of the front and rear Generations 100
bogies are written as: Stall generation 100
Function tolerance 1e-10
Constraints tolerance 1e-10
Mb1 Z̈b1 + C2 + 2C1 Żb1 + K2 + 2k1 Zb1 − C2 Żc Select function Roulette
+l1 θ̇c − K2 Zc + l1 θc − C1 Żw1 − K1 Zw1 − C1 Żw2 Crossover function Scattered

−K1 Zw2 = −u1 − u2 ; (7)

Table 2. Optimized controller gains.
Kp KI KD
Mb2 Z̈b2 + C2 + 2C1 Żb2 + K2 + 2k1 Zb2 − C2 Żc u1 3535659 68314788 32983
u2 13313522 14352104 258263
−l1 θ̇c − K2 Zc − l1 θc − C1 Żw3 − K1 Zw3 − C1 Żw4
−K1 Zw4 = −u1 + u2 ; (8)
values are converged by assuming the best crossover fraction
where Mb1 and b2 are the half-mass of the bogies, Ib1 and Ib2 for minimizing cost function. Furthermore, 100 was selected
are the half-mass moment of inertia of the bogies, l1 and l2 as the number of generations because increasing this parameter
represent, respectively, the semi-longitudinal distance between does not result in a noticeable change in fitness values in either
bogies and the semi-longitudinal distance between wheelsets optimizations for vertical or angular actuators. Optimized con-
in bogies. troller gains are listed in Table 2.
Considering the above relationships, the governing equa- The responses of car-body acceleration in both active and
tions of the system can be written in a matrix form as: passive suspension systems are compared in Fig. (12). The
effect of applying the active suspension system on the acceler-
M q̈ + C q̇ + K1 = F + u; (9) ation of the car-body is clearly observable from these figures.
The reduction percentage of maximum vertical and pitch ac-
where u is the actuator force. For the PID controller u is fol- celerations of the car-body are respectively 58% and 73%. In
lowed as: addition, by assuming ±0.02 m/s2 as marginal acceleration for
Zt vertical acceleration and ±0.002 rad/s2 for pitch acceleration,
d the settling time for those accelerations are 0.053 s and 0.051 s
u(t) = Kp e(t) + KI e(t)dt + KD e(t); (10)
dt while for the passive suspension system, those were 1.3 s and
0 2.9 s. Therefore, the new presented active control can effec-
where Kp , Ki , and Kd are, respectively, proportional, integral, tively enhance the ride quality of the vehicle.
and derivative gains of the PID controller. The variable e(t) is In Fig. (13), the effect of vehicle speed is investigated for
the control error. speeds 1.2V and 0.8V as well, where V is the previous speed
amounting to 27 km/h. As illustrated, by increasing the vehicle
6.2. Obtaining the Gains of the PID speed, the maximum car-body acceleration increases, and the
Controller and Evaluating Results dominant period decreases. By increasing vehicle speed, the
acceleration of irregularity increases and its period decreases,
First, in order to find out the gains of the controller, the ir- so the acceleration of car-body rises, and the dominant period
regularity shown in Fig. (4) is applied to the system, and vi- of car-body acceleration lowers. In Fig. (14), the forces of ac-
bration of the car-body is measured. Then, the controller gains tuators for different vehicle speeds are depicted. These figures
are optimized by using the genetic algorithm to minimize the justify this phenomenon. The higher the speed, the worse the
car-body vibrations. The objective function of the genetic al- actuator can reflect. As a result, maximum acceleration in-
gorithm is to reduce integral of absolute error and maximum creases.
overshoot, written as:
The percentage of maximum acceleration reduction of the
car-body is given in Table 3. As listed, by increasing the vehi-
Z
CF = M O + e(t)dt; (11) cle speed, the efficiency of active suspension is reduced. The
loss of efficiency is due to increasing frequency and input force
where CF is cost function and M O is maximum overshoot. that increases control error, so it reduces the efficiency of active
Parameter specifications for the genetic algorithm are listed suspension. In other words, the designed controller performs
in Table 1. One of those specifications is a crossover frac- better at low speeds.
tion. A crossover fraction is the fraction of individuals in the Based on the above discussion, the designed PID controller
next generation, other than elite children, that are created by for the suspension of the road-rail fire-fighting vehicle reduces
crossover (that is, mating). The rest are generated by mutation. car-body vibration very well and improves the ride quality.
A crossover fraction of one means that all children other than
elite individuals are crossover children. A crossover fraction of Table 3. Maximum acceleration reduction (%).
zero means that all children are mutation children. Choosing Pitch Acc. Vertical Acc. Velocity
the best crossover fraction leads to better optimization, so the 0.8V 65 % 78 %
effect of the crossover fraction was investigated for both actua- V 58 % 73 %
1.2V 53 % 71 %
tors that can be seen in Fig. (10). Figure (11) shows how fitness
International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018 13
R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

a) a)

b) b)

Figure 10. The effect of crossover fraction on best fitness value for: (a) actu-
ator No.1 ands (b) actuator No.2. Figure 12. Car-body vertical and pitch acceleration.

a)
a)

b)
b)

Figure 11. Convergence of fitness value for: (a) actuator No. 1 and (b) actua-
tor No. 2. Figure 13. Car-body vertical and pitch acceleration for different speeds.

14 International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018

 R. Talebitooti, et al.: ACTIVE-SUSPENSION DESIGN FOR A SPECIAL ROAD-RAIL VEHICLE BASED ON VEHICLE-TRACK COUPLED MODEL. . .

3
a) Liao, W. H. and Wang, D. H. Semi-active vibra-
tion control of train suspension systems via mag-
netorheological dampers, Journal of Intelligent Mate-
rial Systems and Structures, 14 (3), 161–172, (2003).
https://dx.doi.org/10.1177/1045389X03014003004
4
Wang, D. H. and Liao, W. H. Semi-active suspension
systems for railway vehicles using magnetorheological
dampers. Part I: system integration and modeling, Ve-
hicle System Dynamics, 47 (11), 1305–1325, (2009).
https://dx.doi.org/10.1080/00423110802538328
5
Sezer, S. and Atalay, A. E. Dynamic modeling and
fuzzy logic control of vibrations of a railway vehi-
cle for different track irregularities, Simulation Model-
b) ing Practice and Theory, 19 (9), 1873–1894, (2011).
https://dx.doi.org/10.1016/j.simpat.2011.04.009
6
Zhai, W. M., Cai, C. B., and Guo, S. Z. Cou-
pling model of vertical and lateral vehicle-track interac-
tion, Vehicle System Dynamics, 26 (1), 61–79, (1996).
https://dx.doi.org/10.1080/00423119608969302
7
Lei, X. and Noda, N. -A. Analyses of dynamic re-
sponse of vehicle and track coupling system with
random irregularity of track vertical profile. Journal
of Sound and Vibration, 258 (1), 147–165, (2002).
https://dx.doi.org/10.1006/jsvi.2002.5107
8
Uzzal, R. U. A., Ahmed, W., and Rakheja, S.
Figure 14. The force of actuators at different speeds. Dynamic analysis of railway vehicle-track interactions
due to wheel flat with a pitch-plane vehicle, Jour-
7. CONCLUSIONS
nal of Mechanical Engineering, 39 (2), 86–94, (2008).
In this paper, with the aid of a 10 DOF vehicle model, a https://dx.doi.org/10.3329/jme.v39i2.1851
vehicle-track coupled model is derived. The governing equa- 9
Zakeri, J. A., Xia, H., and Fan, J.J. Dynamic responses of
tions of the system are modeled in Simulink, and thus car- train-track system to single rail irregularity, Latin American
body accelerations are obtained. The results indicate that the Journal of Solids and Structures, 6 (2), 89–104,(2009).
vehicle-track coupled model can affect the system‘s responses 10
and then controller design. Therefore, a PID controller is de- Li, X. W., Zhang, J. W., Lu, T. L., and Xue, L. Optimization
signed based on the vehicle-track coupled model to improve of suspension parameters based on vehicle-track coupled
the ride quality of the road-rail fire-fighting vehicle. model for a special railway vehicle, Journal of Shanghai
Jiaotong University, 46 (3), 346–351, (2012).
In particular, the results can be summarized as follow:
11
Liu, Y. Lecture notes: Introduction to Finite Element
1. Considering vehicle-track coupled model has a noticeable Method, University of Cincinnati, Cincinnati, (2003), 1st
effect on the the system response. Edition.
12
2. By increasing the number of sleepers and rail length, the Sun, Y. Q. and Dhanasekar, M. A dynamic model for
car-body displacement increases, and the amplitude of its the vertical interaction of the rail track and wagon sys-
acceleration decreases. tem, International Journal of Solids and Structures, 39
(5), 1337–1359,(2002). https://dx.doi.org/10.1016/S0020-
3. Supposing a track consisting of 40 sleepers is appropriate 7683(01)00224-4
when considering car-body vibrations. 13
Newton, S. G. and Clark, R. A. An investiga-
4. Using the finite element method shows an acceptable re- tion into the dynamic effects on the track of
sult for simulating the vehicle-track coupled model. wheel flats on railway wagon, Journal of Mechan-
ical Engineering Science, 21 (4), 87–297, (1979).
5. Using an active suspension system has a significant effect https://dx.doi.org/10.1243/JMES JOUR 1979 021 046 02
on reducing car-body vibration amounting to more than
14
60%; consequently, it improves ride quality of road-rail Ornavas, A. On active secondary suspension in rail vehicles
vehicles. This effect declines by increasing velocity. to improve ride comfort, Doctoral Thesis in Railway Tech-
nology, KTH Engineering Science, Department of Aero-
REFERENCES nautical and Vehicle Engineering, Stockholm, (2011).
15
1
Goodall, R. M. and Mei, T. X. Handbook of Railway Vehicle
Garg, V. K. and Dukkipati, R. V. Dynamics of railway vehi- Dynamics, CRS Press, Taylor & Francis Group, Florida,
cle systems, Academic Press Canada, Toronto, (1984), 1st (2006), 1st Edition.
Edition. 16
2
Johnson, K. L. Contact Mechanics, Cambridge University
Zhai, W. M. Vehicle-track coupling dynamics, Science Press, Cambridge, MA, (1985).
Press Pub., China, (2007), 3rd Edition.
International Journal of Acoustics and Vibration, Vol. 23, No. 1, 2018 15