ISSN 1 746-7233, England, UK

World Journal of Modelling and Simulation

Vol. 5 (2009) No. 2, pp. 130-138

Investigation of vertical dynamic behaviour and modelling of a typical indian

rail road vehicle through bond graph

Vivek Kumar , Vikas Rastogi ∗

Department of Mechanical Engineering, SLIET Longowal, Sangrur, 148106, Punjab, India
(Received December 2 2008, Accepted April 26 2009)

Abstract. Indian Rail transport is one of the major mode of transportation, so it must offer high comfort
level for the passengers and the staff. However, the comfort that passengers experience is a highly complex
and individual phenomenon. The improvement of passenger comfort while travelling has been the subject of
intense interest for many train manufacturers, researchers and companies around the world. Although new
techniques in manufacturing and design ensure better ride quality in railway carriages, still it is impossible
to completely eliminate track defects or various ground irregularities. The dynamic behaviour of a train also
depends on the load and the mechanical systems, such as springs, dampers, etc., which interact with the
wheels, the train body and bogies.
The present study deals with the effects of railway track imperfections on dynamic behaviour, and
investigates the effect of vehicle speed and the rail irregularity on ride comfort through numerical simulation.
The numerical simulation of the vertical dynamic behaviour of a typical railroad vehicle will be done by
modelling through bond graph technique. The model consists of 17 degree of freedom with wheel set, bogie
and car. For assessment of ride comfort Sperling ride index used by Indian Railway is calculated using filtered
RMS accelerations. The ride characteristics of the vehicle provide assessment of the dynamic behaviour of
the vehicle through analysis of accelerations at the vehicle body, whereas ride comfort assesses the influence
of vehicle dynamic behaviour on the human body. The bond graph modelling and its simulation, is performed
using SYMBOLS SHAKTI software[1] .
Keywords: sperling index, ride quality, bond graph modelling, dynamic behaviour, ride comfort

1 Introduction
Indian Rail transport is one of the major mode of transportation, so it must offer high comfort level for the
passengers and the staff. However, the comfort that passengers experience is a highly complex and individual
phenomenon. In several researches, noise and vibration have been identified as the most important factors for
high comfort. The main sources of vibration in a train are track defects in welding or rolling defects, rail joints,
etc. The nature of vibration itself is random and covers a wide frequency range[10] . The vibration accelerations
consist of six components: three translational components along the x−, y and z−axes and three rotational
accelerations round the axes x, y and z, respectively.
The improvement of passenger comfort while travelling has been the subject of intense interest for many
train manufacturers, researchers and companies all over the world. Although new techniques in manufacturing
and design ensure better ride quality in railway carriages, it is sometimes impossible to completely eliminate
track defects or various ground irregularities. The dynamic behavior of a train also depends on the load and
the mechanical systems, such as springs, dampers, etc., which interact with the wheels, the train body and
bogies.
The dynamic performance of a rail road vehicle as related to safety and comfort is evaluated in terms
of specific performance indices. The quantitative measurement of the ride quality is one of such performance
∗
Corresponding author. E-mail address: : [email protected].

Published by World Academic Press, World Academic Union

World Journal of Modelling and Simulation, Vol. 5 (2009) No. 2, pp. 130-138 131

indices. Ride quality is interpreted as the capability of the rail road suspension to maintain the motion within
the range of human comfort and Sperling’s ride index which is a measure of the ride quality and ride comfort
used by Indian Railways[7] .
Due to the unique dynamics that exist between the rail and wheel, rail vehicle dynamics are often difficult
to model accurately. This velocity-dependent dynamics justify the importance of the track input to railcar
modeling. In the physical system, the input comes from the actual track. In a model, a user-defined input is
used to predict the actual track characteristics. The user-defined input can be created analytically or can be
based on actual measurements. Measured track data are obtained by running a specialized railcar down the
track. Analytic track data are created using mathematical shapes, such as cusps, bends, sines, to represent the
track geometry[3] .
There have been several studies, which are contributed by many researchers regarding the dynamic anal-
ysis and to enhance the ride comfort while travelling. Kumar and Sujata[7] presented the numerical simulation
of the vertical dynamic behavior of a railway vehicle and calculated Sperling ride index for comfort evalu-
ation. They modeled a typical Indian rail road vehicle running on broad gauge for the analysis. Kumaran et
al.[8] discusses the dynamic response of a typical pre-stressed concrete rail track-sleeper due to wheel-track
interaction dynamics. Nielsen and Igeland[9] investigated the vertical dynamic behaviour for a railway bogie
moving on a rail which is discretely supported by sleepers resting on an elastic foundation. Effects of imper-
fections on the running surfaces of wheel and rail were studied by assigning irregularity functions to these
surfaces. Bureika and Subacius[2] investigated the dynamics of vertical interaction between a moving rigid
wheel and a flexible railway track. Gangadharan et al.[5] studied the influence of different track irregularities
on dynamic response and coupling between vertical and lateral dynamics.

2 Model for ride comfort

Track irregularities, vehicle characteristics and vehicle speed generate motion quantities that are per-
ceived by passengers. The combinations of these quantities affect the passengers’ perception of ride comfort
as shown in Fig. 1.

Fig. 1. Significant track and vehicle parameters for ride comfort

3 Modeling of rail road vehicle

To analyze the dynamic behavior of railway vehicles, usually the vehicle (and if necessary the environ-
ment) is represented as a multibody system. A multibody system consists of rigid bodies, interconnected via
mass less force elements and joints. Due to the relative motion of the system’s bodies, the force elements gen-
erate applied forces and torques. Typical examples of such force elements are springs, dampers, and actuators
combined in primary and secondary suspensions of railway vehicles.

WJMS email for subscription: [email protected]

132 V. Kumar & V. Rastogi: Investigation of vertical dynamic behaviour

3.1 Assumptions

The assumptions made in formulating the model are as follows:

• Bogie and car body component masses are rigid.
• The springs and dampers of the suspension system elements have linear characteristics.
• Friction does not exist between axle and bearing.
• The vehicle is moving with constant velocity on rigid and constant gauge.
• All wheel profiles are identical from left to right on a given axle and from axle to axle and all wheel
remains in contact with the rails.
• Straight track was assumed.
• An irregularity in vertical direction with same shape for left and right rail is considered in this model.

3.2 Rail road vehicle model

Fig. 2 illustrates the train vehicle model adopted in this study conforming to Indian railways. It consists
of a vehicle body, two bogie frames and four wheel sets. Each bogie consists of the bogie frame, the bolster,
and two wheel sets. The car body is modeled as a rigid body having a mass Mc ; and having moment of
inertia Jbx and Jcy about the transverse and longitudinal centroidal horizontal axes, respectively. Similarly,
each bogie frame is considered as a rigid body with a mass Mb (Mb1 andMb2 ) with moment of inertia Jbx
and Jby about the transverse and longitudinal centroidal horizontal axes, respectively. Each axle along with
the wheel set has a mass Mw (for four axles Mw1 ; Mw2 ; Mw3 amd Mw4 ). The spring and the shock absorber
in the primary suspension for each axle are characterized by spring stiffness Kp and damping coefficient
Cp respectively. Likewise, the secondary suspension is characterized by spring stiffness Ks and damping
coefficient Cs respectively. As the vehicle car body is assumed to be rigid, its motion may be described by the
vertical displacement (bounce or Zc ) and rotations about the transverse horizontal axis (pitch or θc ) and about
the longitudinal horizontal axis (roll or Φc ). Similarly, the movements of the two bogie units are described by
three degrees of freedom Zb ; θb and Φb , each about their centriods. Each axle set is described by two degrees
of freedom Zw ; and Φw . about their centriods. Totally, 17 degrees of freedom have been considered in this
study for the vehicle model shown in Fig. 2. The detailed parameter regarding the moment of inertia and mass
of different component is given in Tab. 1.

Table 1. Detailed parameter of rigid bodies

Moment of Inertia( Kg-m2 )
S.No Name of Rigid Bodies Mass ( Kg ) IXX IZZ
1 Car Body 33700 5.24 × 104 7.36 × 105
2 Bogie-I 3150 2.02 × 103 3.56 × 103
3
3 Bogie-II 3150 2.02 × 10 3.56 × 103
2
4 Wheelset-I 1500 7.13 × 10 -
5 Wheelset-II 1500 7.13 × 102 -
6 Wheelset-III 1500 7.13 × 102 -
7 Wheelset-IV 1500 7.13 × 102 -

3.3 Vehicle parameters

Some parameters regarding the rigid bodies are already given in Tab. 1; however, the other parameters,
which are essential for the simulation of the vehicle are presented in Tab. 2.

3.4 Bond graph model of a rail road vehicle

A typical rail road vehicle system is composed of various components such as car body, springs, dampers,
Bogies, Wheel-set, and so forth. When such dynamic systems are put together from these components, one

WJMS email for contribution: [email protected]

World Journal of Modelling and Simulation, Vol. 5 (2009) No. 2, pp. 130-138 133

Fig. 2. Physical model of railway vehicle

Table 2. Vehicle parameters

S.No. Parameter Nomenclature Values
1 Primary spring stiffness Kp 0.7 × 106N/m
2 Secondary spring stiffness Ks 0.41 × 106N/m
3 Primary damping coefficient Cp 5.88 × 103N s/m
4 Secondary damping coefficient Cs 22000N s/m
5 Half of primary suspension spacing b 1.448m
6 Half of distance between bogie centers c 7.315m
7 Half of secondary suspension lateral spacing d 0.794m
8 Semi gauge length e 0.864m
9 Half of primary suspension lateral spacing f 1.127m

must interconnect rotating and translating inertial elements with axial and rotational springs and dampers, and
also appropriately account for the kinematics of the system structure. Bond graphs are well suited for this task.
In deriving the bond graph models, the velocities in upward direction are assumed to be positive and springs
and dampers are assumed positive in compression.
The bond graph model for the rail vehicle is made by enforcing the kinematic constraints correctly. Kine-
matic constraints are always related with the flow variables[6] . To add the flow according to the constraints,
0-junctions are used. The transformers are used to convert the angular velocity into linear velocity compo-
nents, which is ωc and ωd at c and d points respectively. The complete bond graph model for rail vehicle
system is shown in Fig. 3.

4 Track inputs to rail road vehicle

The dynamic wheel loads generated by a moving train are mainly due to various wheel/track imperfec-
tions. These imperfections are considered as the primary source of dynamic track input to the railroad vehicles.
Normally, the imperfections that exist in the rail-track structure are associated with the vertical track profile,

WJMS email for subscription: [email protected]

134 V. Kumar & V. Rastogi: Investigation of vertical dynamic behaviour

Fig. 3. Bond graph model

cross level, rail joint, wheel flatness, wheel/rail surface corrugations and sometimes uneven support of the
sleepers.

In actual practice different type of periodic, aperiodic or random track irregularities may exist on the
track, but in the present study bump type of irregularity is considered as shown in Fig. 4[3] . The shape of
irregularity is assumed to be similar on left and right rail.

The bump excitation of the front wheel of leading bogie is

WJMS email for contribution: [email protected]

World Journal of Modelling and Simulation, Vol. 5 (2009) No. 2, pp. 130-138 135
 
V L
y = H × sin π × ×t for 0≤t≤ (1)
L V
L
= 0 for t> (2)
V
The bump excitation of the rear wheel of leading bogie is
  
V A2 A2 A2 + L
y = H × sin π × × t − for ≤t≤ (3)
L V V V
A2 + L
= 0 for t > (4)
V
where, A1 is the distance between the front and the rear axle.

Fig. 4. Model of track irregularity

Thus the expression for flow variables are obtained through differentiating Eqs. (1) ∼ (4) w.r.t. time. In
the present study, H is taken as 0.03m and L is taken as 1m.

5 Simulation study
The Simulator of Symbols Shakti, which is the base post-processing module of SYMBOLS Shakti, is
used for the simulation of the rail vehicle. Simulations were done under two conditions.

5.1 Preload analysis

Preliminary simulations were performed to find the static equilibrium of the vehicle and for checking the
bond graph model of the vehicle. This was usually done to confirm since the vehicle is a symmetric vehicle,
the various loads and deflections observed were equal. It was also essential to check whether the vehicle is
correctly balanced or not.

5.2 Dynamic analysis

Dynamic analysis was carried out for vehicle at different speed of 15m/s, 30m/s, 45m/s and 60m/s.
The following output parameters are evaluated:
• Vertical acceleration at the floor of the car-body centre of mass.
• Vertical acceleration at the front and the rear bogie centre pivot.
• Vertical forces for all the wheel-sets.
The acceleration response of the car-body at speed of 15m/s, 30m/s, 45m/s, 60m/s are shown in Fig. 5.
Plots show that initially, the value of acceleration is nearly equal to −9.8m/s2 , which is mainly the acceler-
ation due to gravity. Finally it goes to zero, when the vibration of the car body ceases and it stabilizes. The
acceleration is generally within acceptable range and does not show any instability.
The acceleration response of front and the rear bogie with time is presented at different velocities of
vehicle in Fig. 6. It is evident from the plots, which initially the wheels of the front bogies comes in contact
with the track irregularity and the vibration starts in the front bogie and latter these vibrations are shifted to
the rear bogie. The amplitude of the vehicle vibration also increased with vehicle speed.

WJMS email for subscription: [email protected]

136 V. Kumar & V. Rastogi: Investigation of vertical dynamic behaviour

Fig. 5. Vertical acceleration at the floor of the car-body C. G for vehicle velocity of 15m/s, 30m/s, 45m/s and 60m/s

Fig. 6. Vertical acceleration at front & rear bogie centre pivot respectively for vehicle speed of 15m/s, 30m/s, 45m/s
and 60m/s

5.3 Sperling ride index

Sperling’s ride index is defined as[4] :

v
u nf
uX
WZ = t WZ10i
10
(5)
i=1

where nf is the total number of discrete frequencies of the acceleration response of the railway vehicle iden-
tified by the FFT and WZi is the comfort index corresponding to the ith discrete frequency, given by:
 1
WZ = a2i B(fi )2 6.67

(6)

where ai denotes the amplitude of the peak acceleration response (in cm/s2 ) measured on the floor of the ith
frequency identified by the FFT and B(fi ) a weighting factor, given by:

WJMS email for contribution: [email protected]

World Journal of Modelling and Simulation, Vol. 5 (2009) No. 2, pp. 130-138 137
 21
1.911f 2 + (0.25f 2 )2

B(f ) = k (7)
(1 − 0.277f 2 )2 + (1.563f − 0.0368f 3 )2

where k = 0.737 for horizontal vibration and 0.588 for vertical vibration.

5.4 Comfort evaluation

Acceleration v/s frequency plot was generated for car body at vehicle speed of 15m/s and is shown in
Fig. 7 to calculate the Sperling ride comfort index. The FFT plot is generated for frequency range between 0
to 25 Hz, as the human beings are most sensitive in the frequency range 4 to 12.5 Hz.
Ride comfort analysis has been performed for speeds ranging from 15m/s to 60m/s. The analysis has
been performed on the system model to calculate the vertical acceleration of system. FFT output is taken to
get peak acceleration frequency component. Comfort index has been calculated through Eqs. (5) ∼ (7), which
are present in Tab. 3.

Fig. 7. Vertical acceleration at car body C.G. for vehicle speed of 15m/s, 30m/s, 45m/s and 60m/s2

Table 3. Sperling’s ride comfort index for different vehicle velocities

S. No. Vehicle Speed (m/s) Sperling Index (Wz)
1 15 2.01
2 30 1.83
3 45 1.72
4 60 1.65

6 Conclusion

Vertical dynamic analysis has been carried out for a typical Indian Railway Vehicle of AC/EMU/T (Alter-
nating Current/ Electrical Multiple Unit/ Trailer) type running on broad gauge. A 17 degree of freedom model
is used for analysis. Velocity input at all the wheel set is given by considering similar bump irregularity at both

WJMS email for subscription: [email protected]

138 V. Kumar & V. Rastogi: Investigation of vertical dynamic behaviour

right and the left rail. Vertical acceleration response at the car body C.G has been calculated in the frequency
domain. Sperling Ride indices have been found out for the above vehicle. Comparing the result obtained with
the standards set by the Indian Railways, the values are found well in the satisfactory limit. The Sperling Ride
Index values at different speeds are presented in Tab. 3. The results showed a good agreement with the Indian
Standard and validate the correctness of the model.

References

[1] Users manual of symbols shakti. Http://www.htcinfo.com/.

[2] G. Bureika, R. Subacius. Mathematical modeling of dynamic interaction between wheel-set and rail track. Journal
of Transport, 2002, XVII(2): 46–51.
[3] V. Dukkipati, J. Amyot. Computer-Aided Simulation in Railway Dynamics. Marcel Dekker, New York, 1988.
[4] J. Forstberg. Ride comfort and motion sickness in tilting trains - Human responses to motion environments in train
and simulator experiments. Master’s Thesis, KTH/ FKT/D-00/28-SE, Division of Railway Technology, Department
of Vehicle Engineering, KTH, 2000.
[5] K. Gangadharan, C. Sujatha, V. Ramamurti. Dynamic response of railroad vehicles: a frequency domain approach.
International Journal of Heavy Vehicle Systems, 2008, 5(1): 65 – 81.
[6] D. Karnopp, D. Margolis, R. Rosenberg. System dynamics, modelling and simulation of mechatronic systems.
John Wiley & Sons, 2005.
[7] H. Kumar, Sujata. Vertical dynamic analysis of a typical indian rail road vehicle. in: Proceedings of second
international conference on computational mechanics and simulation, IIT, India, 2006, 8–10.
[8] G. Kumaran, D. Menon, K. Nair. Dynamic studies of rail track sleepers in a track structure system. Journal of
Sound and Vibration, 2003, 268: 485–501.
[9] J. Nielsen, A. Igeland. Vertical dynamic interaction between train and track-influence of wheel and track imper-
fections. Journal of Sound and Vibration, 1995, 187: 825–839.
[10] D. Skarlatos, K. Karakasis, A. Trochidis. Railway fault diagnosis using a fuzzy logic metho. Applied Acoustic,
2004, 65(10): 951–966.

WJMS email for contribution: [email protected]