Air Suspension Characterization. Increasing Air Suspension Effectiveness by Means of A Variable Area Orifice PDF
Air Suspension Characterization. Increasing Air Suspension Effectiveness by Means of A Variable Area Orifice PDF
Air Suspension Characterization. Increasing Air Suspension Effectiveness by Means of A Variable Area Orifice PDF
net/publication/254220242
CITATIONS READS
23 1,282
4 authors:
Some of the authors of this publication are also working on these related projects:
POWER. Potential of waste energy recovering in thermal combustion engines. Energy and environmental implications. View project
All content following this page was uploaded by Jordi Vinolas on 21 January 2015.
ABSTRACT
The air spring is one of the components that affects vehicle comfort. This element usually makes up the
main part of the secondary suspension, which introduces both stiffness and damping between the bogie
and the car body. Therefore, a deep understanding of this element is necessary in order to study the
comfort of a vehicle, the influence of different parameters and the ways to improve it. In this work, the
effect that the air spring system has on comfort is studied.
To accomplish this, first a typical pneumatic suspension composition is briefly studied. Then, the test
bench developed to study this system is described, presenting experimental results. Correlation of the
results with some theoretical models is also addressed.
Then, the effect of the air spring system on comfort is analysed and finally, it is discussed the
improvements from introducing a variable area orifice in the pipe that joints the air spring and the surge
reservoir.
INTRODUCTION
In secondary suspensions systems of high-speed trains, inter-citys, and conmuters the classical helicoidal
spring has nearly been completely replaced by air springs. Several advantages can be exploited when air
springs are used; firstly, the suspension height can be controlled by the air spring internal pressure;
secondly, some damping is introduced in if an auxiliary air chamber is connected to the air spring. In
addition, the typical force/displacement curve of an air spring provides a natural frequency of the
suspension nearly independent of the sprung mass, and unlike helicoidal springs the air springs filters out
high frequency vibrations, thus reducing the vibration transmission form the bogie to the carbody.
Therefore, being a so frequently used component all the available multibody software codes provide the
means to model the dynamic vertical stiffness of an air spring. Different levels of complexity are
available starting obviously from the simpler which would be just a spring. A significant research effort
has been made to propose specific and more precise models of this component. Among others the
following works can be underlined: the Nishimura Model [1, 2], the vampire model [3], the GENSYS
model [4] and the Docquier et all. model [5, 6]. In spite of all these work, further research is needed to
represent the air spring behaviour accurately [7].
The aim of the work that this paper summarises is, on one side, the verification, by testing a specific air
spring, of the validity and limitations of different classical models. On the other side, concentrating on
comfort aspects the paper proposes a new idea to improve the behaviour of air spring suspensions. A
specific test bench was designed and built in order to study in depth the dynamics of an air spring which
is described just after this section. On the following sections, some test results are provided which show
in a summarized way the typical behaviour of a vertical suspension system and the effect of the
connection between air spring and auxiliary chamber. It was also considered interesting to include a
comparison of test results of a typical air spring secondary suspension with theoretical results obtained by
using classical models.
As this element usually makes up the main part of the secondary suspension, which introduces both
stiffness and damping between the bogie and the car body, the last sections of the paper concentrate on
the role of the air spring in comfort and particularly and how the modelling of this component is affecting
the results. A final section is devoted to present the idea of a variable orifice valve inside the pipe that
connects the air spring and the auxiliary chamber; some preliminary results and a sketch of the design are
presented.
Air spring
Emergency
spring
Figure 1.- Air spring and emergency spring.
Auxiliary air
chamber
Pipe
In Figure 4 the testing arrangement of the assembly (air spring, emergency spring, pipe and auxiliary air
chamber) used in a secondary suspension can be seen.
Emergency
spring
Pipe
Figure 4.- Air spring-emergency spring-pipe-auxiliary air chamber assembled on the ASTB.
Effect of amplitude
Figure 5 shows the vertical dynamic stiffness of the pneumatic system at different frequencies and for two
different amplitudes (0.5mm, 1mm).
1.8E+06
1.6E+06 A=0.5 mm
A=1.0 mm
1.4E+06
Dynamic stiffness (N/m)
1.2E+06
1.0E+06
8.0E+05
6.0E+05
4.0E+05
2.0E+05
0.0E+00
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Figure 5.- Dynamic stiffness (Pressure 4 bar, Pipe interior diameter of 38 mm)
The following considerations can be drawn from the analysis of this graph:
The pneumatic system is clearly non-linear in the medium frequency range (from 6 to 14 Hz in
this case). As can be seen, the dynamic stiffness value in these frequencies varies according to
oscillation amplitude. On the contrary, the value of said parameter at high and low frequencies is
not related to oscillation amplitude.
As frequency is reduced, the stiffness value tends to a fixed value. This value is known as
pneumatic system static stiffness.
Likewise, when frequency is increased, the system leans toward a different fixed value: the
dynamic stiffness of the system for high frequencies. As can be seen, the stiffness value for high
frequencies is greater than the stiffness value for low frequencies.
1.2E+06
1.0E+06
8.0E+05
6.0E+05
4.0E+05
2.0E+05
0.0E+00
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Figure 6.- Dynamic stiffness of the set. Effect of the pipe diameter. Pressure 4 bar, amplitude 1 mm.
As can be seen, the stiffness of the system for low and high frequencies is not affected by changing the
pipe; however, the medium frequency range is clearly affected by this change. The position and the
maximum and minimum stiffness values are modified: as the pipe diameter increases, the transition range
moves towards higher frequencies.
1.0E+06
8.0E+05
6.0E+05
4.0E+05
2.0E+05
0.0E+00
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Figure 7.- Dynamic stiffness of the set with pipe A. Effect of the pipe and chamber. Pressure 4 bar.
1,8E+06
Without calibrated orifice
1,6E+06 Calibrated orifice = 30 mm
Calibrated orifice = 25 mm
1,4E+06 Calibrated orifice = 15 mm
Dynamic stiffness (N/m)
1,2E+06
1,0E+06
8,0E+05
6,0E+05
4,0E+05
2,0E+05
0,0E+00
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Figure 8.- Dynamic stiffness of the set with pipe A. Effect of a calibrated orifice on the pipe. Pressure 4
bar, amplitude 0.5 mm.
THEORETICAL MODELLING OF A PNEUMATIC SYSTEM’S VERTICAL
BEHAVIOUR
The theoretical prediction of a pneumatic system’s behaviour has been widely studied. Among the
different models used, the following are mentioned:
Modelling by means of a single stiffness element.
Nishimura model [1, 2] (figure 9-a). In this model, K1 refers to the flexibility of the air inside the
air spring, K2 refers to the air inside the auxiliary air chamber and C relates to the resistance of
the fluid as it flows through the connection pipe.
VAMPIRE model [3](figure 9-b) (which provides the same results as the GENSYS model [4]).
In this model, the K1, K2 and C parameters have the same meaning as in the Nishimura model.
The M parameter relates to the inertia of the fluid inside the pipe.
a) b)
Figure 9.- Theoretical model for pneumatic suspension ( a) Nishimura model, b) VAMPIRE model)
As can be seen, all three models have different degrees of complexity. Figure 10 compares the predictions
provided by the different theoretical models.
5
x 10
14
K
Nishimura
12 VAMPIRE
Dynamic Stiffness (N/m)
10
2
0 5 10 15 20 25 30 35 40
Freq (Hz)
Using only the first, it is possible to use a single stiffness value (which usually coincides with the stiffness
at low frequencies).
On the other hand, when using the Nishimura model, proper representation of stiffness at low and high
frequencies is achieved. However, the results obtained during the transition are not accurate enough.
Finally, it can be seen that the results provided by the VAMPIRE or GENSYS model are accurate in the
entire frequency range.
Figure 11 shows the experimental results obtained in other cases along with their corresponding
theoretical predictions provided by the VAMPIRE or GENSYS model. As can be seen, the provided
theoretical results are very close to the experimental ones.
5
x 10
16
Without orifice Exp
14 Without orifice Teo
Orifice 30 mm Exp
12 Orifice 30 mm Teo
Dynamic Stiffness (N/m)
Orifice 25 mm Exp
10 Orifice 25 mm Teo
Orifice 15 mm Exp
Orifice 15 mm Teo
8
0
0 2 4 6 8 10 12 14 16 18 20
Freq (Hz)
As conclusion, we can affirm that the VAMPIRE or GENSYS models enable to adequately reproduce the
behaviour of the pneumatic system in the entire frequency range.
400 C=10C0
C=100C0
350
C= inf
FRF Aceleración
300
250
200
150
100
50
0
0 5 10 15 20 25
f (Hz)
The VAMPIRE model has already been used for obtaining figure 13. This figure compares the results
obtained while varying the pipe length. As can be seen, the results between cases vary largely. It can also
be observed that if the peak stiffness that appears during the transition between static and dynamic
stiffness reaches near one of the vehicle vibration modes, it may produce a significant comfort reduction.
350
M=0
M=M0
300
M=3M0
M=5M0
250
M=10M0
FRF Aceleración
200
150
100
50
0
0 5 10 15 20 25
f (Hz)
From the analysis above, it can be concluded that the characteristics of the pneumatic system affect
comfort in a large degree. Likewise, it can be concluded that the use of a model that enables to properly
represent the behaviour of the pneumatic system in the entire frequency range is required for performing
comfort predictions.
INTRODUCTION OF A VARIABLE ORIFICE VALVE
Experimental as well as theoretical results have verified that the secondary pneumatic suspension is a
non-linear system. This non-linear property results from the loss of pressure to air speed ratio when
travelling through the connection pipe.
This non-linear property makes choosing the system’s damping value difficult because the C value, which
maximizes passenger comfort depends entirely on the level of excitation. This fact can be clearly seen in
figures 14 and 15.
Figure 14 shows the effective value of vertical accelerations at two points of the car body for different
absorption values and for a specific level of track irregularities (value 1 corresponds to the maximum
value of comfort that can be achieved at one of the measuring points by only varying the C parameter for
absorption).
As can be seen in this particular case, the C parameter, which is able to achieve good levels of comfort at
the two measuring points is around 0.65·10 4.
3.5
Center
Pivot
3
Normalized Confort Parameter
2.5
1.5
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
C 4
x 10
Figure 14.-Normalized comfort parameter for different values of C. Level of oscillations L0
Figure 15 is obtained if the excitation is doubled. As can be seen, if the C obtained in the previous case is
used, the level of comfort is not the optimum (the optimum level that can be achieved is around 1.36 and
when calculating C=0.65·104 we obtain 2.35).
6
Center
5.5
Pivot
5
3.5
2.5
1.5
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
c 4
x 10
Figure 15.-Normalized comfort parameter for different values of C. Level of oscillations 2xL0
In order to solve this problem, we propose to insert a calibrated orifice inside the pipe whose opening will
be dependent on the pressure. By choosing a proper opening, the system can achieve a lineal property.
This way, the above mentioned problem may be avoided and optimum values of comfort achieved that
are independent of the rail irregularity level.
The system’s non-lineal property is due to the ratio between the pressure drop (P) in the pipe and the air
speed (v):
P Kv 2 (1)
By making a variable orifice valve that varies its opening according to pressure, a load loss (K) that is not
constant and inversely proportional to the pressure value can be obtained.
K'
K (P) (2)
P
By substituting (2) for (1), a linear expression between pressure and speeds is achieved.
In order to determine the law for orifice opening according to pressure, it can first be assumed that load
losses follow the following law:
2
Apipe
K (P) 2.78· (3)
A ( P )
orifice
By making expressions (2) and (3) equal, the variable orifice opening law can be calculated:
0.6 K (P)
Aorifice (P) (4)
Apipe K'
In order to realise this device, different solutions has been proposed. Figure 16 show one of the possible
solutions.
Figure 16.-Variable opening valve
ACKNOWLEDGMENT
REFERENCES
[1] Matsumiya S., Nishioka K., Nishimura S. and Suzuki M. (1969). On the diaphragm air spring
sumride, The Sumitomo Search, No 2, pp 86-92.
[2] Oda N., Nishimura S.(1970) Vibration of air suspension bogies and their design Bulletin of
JSME Vol 13 pp.43-50
[3] VAMPIRE USERS MANUAL, C2.4 shear elements, C2.4.2 Air spring elements
[4] Berg M. (2000)Three-dimensional airspring model with friction and orifice damping. Vehicle
System Dynamics, vol 33 suppl. Pp 528-539
[5] Docquier N. Fisette P., Jeanmarg H.(2007), Multiphisic modelling of railway vehiles equipped
with pneumatic suspensions Vehicle System dynamics, vol 45, pp. 505-524.
[6] Docquier N. Fisette P., Jeanmarg H.(2008), Model-based evaluation of railway pneumatic
suspensions Vehicle System Dynamics, v 46, n SUPPL.1, p 481-493, 2008
[7] Evans J.; Berg M. (2009) Challenges in simulation of rail vehicle dynamics Vehicle System
Dynamics Vol. 47, No. 8, August 2009, 1023–1048