Vehicle System Dynamics, 28 (1997). pp. 159- 195 0042-3 1 14/97/2802- 159$12.

00
O Swets & Zeitlinger

PantographICatenary Dynamics and Control

',
G. POETSCH J. EVANS 2 , R. MEISINGER 3, W. KORTUM 4,
',
W. BALDAUF A. VEITL and J. WALLASCHEK

SUMMARY
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

The pantograph-catenary system with its dynamic behaviour turned out to be a crucial component for
new train systems required to run at higher speeds. With the present systems, operational limitations
have to be accepted when running with several pantographs in the train set, when tilting trains are
employed, when running on low quality catenary sections or when stricter noise reduction regulations
are forcing lower noise emissions also for the pantographs. This paper gives an overview of the
methods to describe the catenary and the pantograph system dynamics. Furthermore, aspects concern-
ing the interaction between current collectors and overhead equipment, the acquisition of the model
data and the verification are presented. Finally various constructions of passive pantographs and
proposals for active control concepts are discussed.

1. INTRODUCTION

Railroad systems presently experience rapid technological advances, so that some

authors speak of a renaissance and even of a golden age for advanced railway
systems, [I]. The technological frontiers can be identified in Japan and Europe,
USA still betting on air and road traffic. Europe is of particular interest with
respect to the topic of this survey as the rising level of EU high speed links means'
frequent changing from one railway system to another, compatibility between the
pantograph and the overhead lines becoming a crucial problem. Overhead lines, a
complex arrangement of wires and cables, are disturbed by passage of pan-
tographs, an interaction of two dynamic systems take place. The understanding of
this complex interaction of moving pantograph and stationary overhead line
equipment is mandatory for a prediction capability by computer simulation, firstly
to allow assessment of current design stage, and secondly to aid developments on
existing routes where performance improvements are sought. These include in
particular increasing the speed of existing overhead line equipment (21.

I DB AG, Forschungs- und Technologiezentrum. Volckerstr. 5. 80939 Munchen, Germany. E-Mail:

[email protected], E-Mail: [email protected]
BR Research Lltd. PO Box 2. London Road, Derby DE24 8YB, UK. E-Mail: [email protected]
' Fachhochschul,: Numberg, Box 210320 90121 Numberg, Germany. Fax: *49-91 1-5880135
DLR, Institute for Robotics and System Dynamics, 82234 Wessling. Germany. E-Mail:,
[email protected],E-Mail: [email protected]
' Heinz Nixdorf Institut, Fiirstenallee 11, 33102 Paderborn, Germany. E-Mail: [email protected]
paderbom.de
 160 G . POETSCH ET AL.

Traditionally, each country in Europe had developed its own overhead line
system and has optimised its pantographs for use on it. In the near future, trains
should not stop when crossing from one system to another and change locomo-
tives. Thus there are two major challenges in the area of pantograph/catenary
interaction:
- Designing pantographs which are capable of operating satisfactorily on a range
of different overhead designs, sometimes existing even within one system.
- New train systems being required to run at higher speeds as originally planned
and therefore would contribute to a considerable competitiveness through a
higher degree of utilisation of the existing infrastructure at low costs. To gain
these effects pantographs could even be 'gold-plated' if no or only minor
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

changes to the infrastructure is required; see the parallelism to tilting trains.

From the current collection point of view, the ideal solution to higher efficiency
and operational speed would always be new and higher specifications of the
overhead system. However, the immense capital cost of installing new overhead
line equipment is difficult to justify financially, with the exception perhaps on new
lines. It is therefore a considerable incentive to obtain the optimum performance
from existin8 equipment which has been installed, there is still a challenge of
rnain;aining speeds when running off the new system onto more conventional
routes [31.

Assessment of Current Collection Quality

It is the function of the pantograph/catenary system, see Fig. I , to provide
uninterrupted and reliable transmission of the electric energy supplied via the
contact wire to the electrical driving units of traction vehicles by means of suitable
pantographs adapted to the design of the catenary. The decisive criterion for
assessing the contact quality and therefore the quality of the energy transmission is
the contact-force time response occurring between the pantograph slipper and the
contact wire. The contact force is composed essentially of a static force component
given by the application force and a dynamic component dependent on the running
speed and the vibrational behaviour of the catenary and the pantograph. Too low
contact ,forces result in contact interruptions. Technical specifications in most
countries have stated that the lost of contact between the pantograph and the

- Pantograph

Train

Fig. I . The combined system ' pantograoh/catenary'.

 PANTOGRAPH/CATENARY DYNAMICS. CONTROL 161

overhead line should be less than 1% at normal operating speeds. However, with
today's strict requirements concerning electromagnetic compatibility, the interfer-
ence caused by stray currents associated with loss of contact needs to be kept to a
minimum; therefore, a figure of 1% loss of contact is considered highly unsatisfac-
tory [3].
Arcing occurs which increases wear on the catenary and the slipper and
therefore reduces the service life of the components. If the contact forces are too
high, catenary wear increases because of increased abrasion. In extreme cases,
unacceptably high catenary lifts can occur in which the maximum permissible lift
relevant to operational safety on the support is exceeded when the pantograph
passes. This car) result in major damage to the pantograph and catenary equipment.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

High dynamic ,contact force components, which occur on following pantographs .

above all during multiple unit traction, excite the system to severe oscillations.
Because of the heavy stresses from alternating loads arising from this, the result
can be damage due to material fatigue - especially of the dropper and support
components. Since catenary equipment demands a very large volume of invest-
ment in general, it must be designed in such a manner that a long service is
guaranteed with the least possible amount of technical support
(maintenance/s;ervicing). In order to keep wear (contact wire abrasion, material
fatigue) low and to achieve satisfactory contact quality at the same time, there
must be optimum harmonisation between the pantograph and catenary. The aim of
the design is, therefore, to keep the dynamic contact force fluctuations as low as
possible so that reliable contact is achieved without interruptions even with very
low mean contact forces. Contact wire abrasion is minimised in this operating
method; when :running with multiple unit traction, there are greater lift reserves
available for following pantographs.
Recent discussion among European railways has resulted in statistical analysis
of the contact force as opposed to a loss of contact. A major disadvantage of using
loss of contact :is that a measurement of zero gives no clear indication of whether
the system is performing well within its operational speeds, or whether only a
small speed increment of parameter variation would cause the contact performance
to break down. The statistics of contact force can easily be obtained from
measured or calculated force values. Thus, the following key performance parame-
ters mdy be used [3]:
meah,contact force M,
range of contact force M 1- 3u,
statistical occurrence of loss of contact,
statistical occurrence of low contact force below specified safety margin (e.g.,
30N or 50N!,
maximum wire displacement at registration arms.

When assessing the system performance limits the sensitivity of a given

pantograph/overhead system to a range of key and critical parameters offering
overhead performance are:
 162 G.POETSCH ET AL.

contact wire tension,

catenary wire tension,
contact wire sag,
errors in setting the contact wire height and
static/dynamic uplift forces.

It should be noted that only optimisation of uplift force on the pantograph also
needs to take into account aerodynamic forces. Airflow acting on the pantograph
is, however, a very complex physical process depending on geometry, speed and
various additional varying conditions such as crosswind, airflow, etc., so that
uplifts in windy conditions can be in excess of twice those which occur at calm
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

weather.
As a result, the performance measures cited above are very crucial for design
qualification. Moreover, designs need to be very 'robust' (parameter insensitive)
versus real world fluctuating conditions.

European Individualism and Future Needs

The catenary network is a heterogeneous structure which has evolved historically
with standard catenary types with different running qualities. In Britain, electrifica-
tion has used relatively lightweight, low tension design of overhead line equipment
with considerable development going into a sophisticated pantograph in order to
maintain current collection quality at high speed. In France, good quality current
collection on TGV lines has been achieved by the use of a high tension overhead
system. In Germany, the tension used by DB for its ICE lines in midway between
that used by BR and SNCF, and a combination of a stitched overhead design and a
lightweight pantograph has been used to maintain performance.
A particular source of incompatibility between systems is whether they run with
AC or DC power. With high voltage AC systems, the amount of current required
to power the train is relatively low, and such systems are usually based upon
lightweight overhead conductor wires and light pantograph designs. The need to
avoid excessive movements of the light and often relatively low tensioned wires
means that uplift forces are generally kept as low as possible on AC systems. In
contrast, with low voltage DC system, much greater levels of current are neces-
sary. In such circumstances, continuity of contact can be better guaranteed with
higher level of uplift force. There is also a widely held belief that DC current flow
can become interrupted, even without actual contact loss, unless reasonable levels
of contact force are maintained. Accordingly, DC systems are usually operated
with much greater level of uplift force, and DC overhead equipment tends to be
designed such that it gives its optimum performance under these conditions.
Thus, DC pantographs are in general dynamically inferior to AC pans, on
account of their greater dynamic mass. But, more crucially, if operated on AC
equipment with their normal uplift force characteristics, they are liable to cause
dangerous levels of wire displacements. The operation of an AC pantograph on a
DC overhead system is a more realistic proposition, but it may not necessarily be
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 163

able to perform satisfactorily at line speed, without increasing uplift force level
beyond its norrnal specifications [3].
Together wnth the introduction of new, faster train generations, improved
catenary designs were developed which permit running at the maximum target
speed while ob:;erving the relevant directives and specifications. The development
or rather modii'ication of the pantographs employed was performed according to
the relevant cat~enaryconfigurations on the track sections scheduled for operations.
With the present systems, operational limitations have to be accepted when
running with several pantographs in the train set, when tilting trains are employed
and when running on low quality catenary sections which are only designed for
operations at low speed.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

On account of the very high investment costs, new catenary equipment is only
being installed on purely new construction lines or on track sections which have to
be upgraded for high-speed traffic. Modernisation measures on the existing
network are limited to the elimination of specific, local weak points and are only
performed if this leads to the elimination of significant operational limitations for
the flow of traffic.
What is in the way of comprehensive optimisation or modemisation of the
existing catenary infrastructure to improve the interaction between pantograph and
catenary - as is necessary to achieve the increase in efficiency aimed at - is the
immense expenditure involved. This is especially true since it is assumed that not
all train sets by any means will run on the relevant track sections at higher speeds.
It seems mo:re favourable to improve the properties of the pantographs to such
an extent, or to introduce new, efficient pantographs, that the above-mentioned
operational lim~itations can be avoided without modification measures to the
existing infrastructure and the increases in performance required can be achieved.
Since the purely passive pantographs employed at present are practically ex-
hausted with regard to the technical opportunities they offer, it is not possible to
achieve the ne:cessary improvements by further development or modification
measures with rhese.
A cost-efficient alternative might be thc employment of new, actively con-
trolled pantographs which can adapt to the various catenary properties or opera-
tional conditions in their vibration-dynamic behaviour and largely control the
contact variations. On account of the reduction in the dynamic contact variations,
the mean static: contact force can be decreased further and therefore wear and
contact wire lift can be reduced. Sufficient lift reserve is available for following
pantographs.

2. MODELLING AND SIMULATION OF THE SYSTEM DYNAMICS

The design of catenary-pantograph-systems initially was based on some principal

dynamic considerations which were obtained from simplified mathematical models
of catenary anti pantograph. These basic design models include, e.g., lumped-
 164 G. POETSCH ET AL.

mass-models [4], infinite-string-models [5] and dropper-reflection models [6].

Initiated by the demand for increasing train speed, the models used for the design
of catenary-pantograph systems became more and more complex, and, starting in
the early 70s, computer-based simulations of the catenary-pantograph system were
used for optimisation. The aim was to provide simulation tools which enable
reliable assessment of the system quality, reducing the need for costly measure-
ment and - by way of investigation of the system behaviour under the influence of
parameter variations - allow systematic optimisation of the overall design.

2.1. Catenary Dynamics

Although catenary and pantograph form a coupled dynamic system, much of the
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

research has been dealing with the modelling and simulation of the catenaries,
since the computational labour involved by numerically solving the partial differ-
ential equations (PDEs) describing their system behaviour turned out to be the
limiting factor. Particular problems arise from two system properties:
The low damping of the system results in catenary oscillations which are not
confined to a small region near the pantograph-catenary contact point. Wave
propagation takes place in the whole catenary, which therefore must be
simulated in its entirety. The low damping poses the additional problem that
errors introduced by numerical approximations will inevitably be conserved
during the following simulation process and can in sequel lead to further error
accumulation.
The existence of high-frequency processes occurring in pantograph-catenary
interaction has been confirmed by measurements of contact wire corrugation
profiles [7] as well as by measurements of contact breaks [8]. The system's
behaviour therefore requires high resolution both in space and in time for
correct numerical treatment. This results in a high computational load due to
the dimensions of the catenary.

In the following, an overview about the modelling of catenaries, outlining the

significance of different physical effects occurring in the system dynamics, will be
given. Then, some common methods for the numerical simulation of the system
dynamics will be discussed with respect to their applicability to different catenary
models.

2.1.1. Simplified Models Derived from Measurements

An interesting approach to circumvent the tedious and time-consuming computa-
tions usually arising from simulation of the catenary dynamics consists in extract-
ing information about system properties form measurement data. This has been
done, e.g., in the work [9], where a "dynamic stiffness model" of the catenary has
been created by combination of measured contact force and displacement values.
This model allows an easy assessment of the quality of pantograph modelling, but
its use is restricted to the simulation of the pantograph type used during the
measurements. Since the dynamic effects included in the catenary model are an
 PANTOGRAPH/CATENARY DYNAMICS,CONTROL

Fig. 2. Stitched wire catenary (above) and compound equipment (below).

unique result of the measurement conditions (not only pantograph, but also of
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

catenary type and train speed), the model is unlikely to be suitable for simulations
with the purpose of pantograph optimisation.

2.1.2. Modelling of the Catenary

The early development processes have created a wide variety of catenaries (see,
e.g., [lo]). Today, two basic configurations are in general use for high-speed lines:
the stitched wire catenary and the compound equipment catenary Figure 2.
From a struclural point of view, catenaries can be represented as an assembly of
simple standard elements (one-dimensional wires and droppers, registration arms,
insulators and :supports). However, the choice of appropriate models for each
element has shown to be essential for reliable simulation results. Effects consid-
ered to be of major importance are
the slackening of droppers, when subjected to compressive forces [ I I],
the effect of the static contact wire sag [12], and
the excitatioli of the pantograph due to the stagger of the contact wire.

These effects can easily be included in the simulation model in the form of case
distinctions and algebraic equations. The modelling of droppers, insulators and
registration arms can be achieved using standard methods of multibody dynamics
leading to ordinary differential equations (ODEs) which can be handled by
standard numerical methods. On the other hand, wave propagation taking place in
the wires can only be described satisfactory by PDEs which are significantly more
difficult to handle numerically than ODEs since solutions are time- and space-de-
pendent. We will therefore concentrate on the wire models here.
The most common mathematical model that can be used for the wires is the
string model, whose equation of motion is given by
PAW= -Pw+TwU+q(x, t ) , (I)

describing the wire as a homogeneous string of mass per unit length describing the
wire as a homogeneous string of mass per unit length PA, with tensile force T and
viscous damping determined by the parameter P. This model has been used in a
large number of works, e.g., [11 , I 3-23]. The advantage of the string model lies in
the simplicity o;t the PDE's solutions. If damping is low (in real systems, values
 166 G . POETSCH ET AL.

are in the range of 0.3% to 1% of critical), propagation of waves as described by

the model is almost non-dispersive. Waves travel with a constant propagation
speed c = \
Im which does not significantly depend on their wavelength.
This property can be exploited by numerical methods, resulting in very fast
computational algorithms.
Besides this model, beam models have been used to represent the catenary
wires [12,24,25]. Beam models take the bending stiffness of the wire into account
(Euler-Bernoulli beam) and are capable to include the effects of shear deformation
and rotatory inertia (Timoshenko beam). The equation of the Euler-Bernoulli beam
is given by
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

where EI denotes the bending stiffness of the wire. The effect of the bending
stiffness complicates the mathematical analysis considerably, although it is con-
fined to a small region near the pantograph-catenary contact point ([I 1,231).
Several authors regarded bending stiffness as a negligible factor [11,18,22,23].
Nevertheless, there are reasons which make the use of beam models preferable:
A comparison of computational results obtained for string and beam models
[24] shows significant differences in the solutions when the train speed
approaches the critical wave propagation speed c of the string model. This limit
is of special importance in the case of trains running on low quality catenary
sections, which are only designed for low speed operation.
Under the influence of a moving point force, the string model reveals a
discontinuity in the slope of the contact wire, which is an unrealistic representa-
tion of the physical behaviour. This discontinuity is not apparent for the
Euler-Bernoulli beam model [26].
Modelling of the dispersion of the wave propagation and the formation of wave
fronts becomes important when the pantograph approaches a dropper or a
registration arm. The string model leads to formation of wave fronts, the
high-frequent parts thereof being completely reflected at point masses [27]. The
situation of a pantograph approaching a dropper is therefore similar to the
approach of the pantograph to a support, which theoretically results in infinite
growth of the contact force peak values [28].
I t should be noted that the latter two points are only apparent in simulation
results, when high-resolution methods are used. The difference between string
model and Euler-Bernoulli beam model becomes significant for waves with a
wavelength A smaller than 2 r J / which is about 0.6 nl for the contact wire
of the German catenary Re 250. For this system, differences between Euler-
Bernoulli beam model and Timoshenko beam model are negligible for waves with
A > 5 cm, so that there seems to be no need to consider shear deformation and
rotatory inertia of the wires.
From measurements of corrugation profiles [7] and the above-mentioned con-
vergence problems imposed by properties of the string motlel, ~t can be concluded
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 167

that the bending stiffness of the contact wire plays an essential role in the
high-frequency pantograph/catenary interaction and therefore should be included
in the simulation models, when high accuracy is required. In our experience, the
effect of bending stiffness has shown to be important especially for the accurate
determination of high-frequency contact processes (corrugation, contact breaks).
However, as shown in Section 3.1, for many real applications some simplification
can be acceptable. Neglecting the bending stiffness, e.g., will hardly influence
simulation results with respect to the low-frequency behaviour. In many cases, it is
therefore possible to extract valuable information about the principal system
behaviour by use of the string .model.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

2.1.3. Numerical Treatment of the Catenary DifSerential Equations

Many different numerical methods have been used for catenary simulation. Until
recently, there has been no consensus which of these offers best performance. The
following overview will outline the fundamental properties of the methods which
have been used: restrictions imposed on the modelling as well as accuracy and
computational speed of the methods.
An important characteristic of each method is the dependency of computational
effort (W) on the resolution of the method which will be characterised by the
number N of degrees of freedom (DOF). For "global" methods like modal
analysis, N denotes the number of eigenfunctions used to approximate the system
behaviour, for "local" methods like d' Alembert's method, the Finite Difference
Method and the Finite Element Method N is equivalent to the number of space
discretization points. A criterion for estimating the efficiency of a method can be
obtained by relating the accuracy of results to the computational work, both
expressed in terms of N. As the accuracy of the results also depends on the
physical model of the system, the final discussion will be given after the following
characterisation of methods.
Modal Analysis and related methods [12,15,17,18,21] are semi-discrete in
nature, since the catenaries vibrations are represented by a superposition of a
finite number of exact or estimated eigenfunctions which are continuous in
space. This approach allows an easy treatment of the pantograph-catenary
coupling. The displacement of the contact point can be calculated without
interpolation, and the contact force can be described as a concentrated point
force. The drawbacks of modal analysis are the time-consuming pre-processing
(computation of eigenfunctions); a lack of flexibility when the structural design
of the catenary needs to be changed and the assumption of system linearity
which is violated by the slackening of droppers. The computational load of the
method depends at least quadratic' on the resolution (W N~),regardless -
whether the solution is calculated stepwise in time or obtained after transfor-
mation in the frequency domain as in [21].

Use o f the Euler-E:ernoulli beam model might even require further reductio~io f time steps for reasons
-
I
o f accuracy, leading to W N3.
 168 G. POETSCH ET AL.

D' Alembert's travelling wave method [22] is a special case of the method of
characteristics, applied to the one-dimensional wave equation. This method is
restricted to the string model, but offers exceptional performance. The method
makes use of known solution properties (nondispersive wave propagation). The
solution is given by two waves travelling in opposite directions. Performing a
time step corresponds to shifting both waves by Ax = cAt in space which can
be done by shifting a pointer on the first (resp. last) value. As only new values
resulting from wave-boundary-interactions need to be computed, the computa-
tional work per time step is independent of the spatial resolution. Increasing
the spatial resolution linearly raises the number of timesteps which results in
-
W N.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Finite Difference Method (FDM) [11,14,20,23,25,29] and Finite Element

Method ( E M ) [16,30,31] are not restricted to special physical models and
therefore both offer a high degree of flexibility. Physical effects inherent to the
PDEs are misrepresented to a certain degree by the approximate character of
the space discretization, but this can be overcome by choosing an appropriate
numerical integration scheme [32]. Both methods lead to a high computational
load when a high space resolution is used, since the stepsize used for time
integration needs to be chosen in dependence on the stepsize of space

At - -
-
results in W N3 for the Euler-Bernoulli beam model.
-
discretization. For the string model, At Ax leads to (W N2), whereas

The differences between FDM and FEM have been analysed by [32]. ODEs
resulting from FDM are explicit, and good approximations of the model behaviour
can be obtained by explicit integration. In contrast to FDM, FEM puts stronger
restrictions on the integration time step and leads to implicit systems of ODEs. T o
achieve good approximation properties, in combination with FEM, implicit inte-
gration must be used. Since use of implicit methods leads to time-consuming
computations, FDM is much preferable to FEM.
Independent of the method used, the PDEs describing the dynamical behaviour
cause most of the computational work. The ODEs arising from the modelling of
droppers and registration arms require additional mathematical treatment, which
however causes no particular problems except for modal analysis where account-
ing for non-linear system behaviour is troublesome.
With respect to computational speed, d' Alembert's method must be considered
as the most efficient method available. Unfortunately, this method is restricted to
the string model for which singular behaviour has been shown with respect to
contact force calculation.
The high computational load for the treatment of beam models with FDM or
FEM makes the use of those methods prohibitive at first sight. Still, convergence
properties have not yet been considered which will shed a different light on the
situation. Since convergence is not only dependent on model properties and the
numerical method applied, but also on the character of the problem to be analysed,
we will discuss this topic with some examples.
 PANTOGRAPH/CATENARY DYNAMICS. CONTROL 169

Free Catena'ry.Vibrations. The numerical analysis of free catenary vibrations

has been used by several authors to judge the quality of both catenary model and
numerical methods [27,30,33]. Thereby, additional effects occurring from panto-
graph-catenary coupling are excluded in order to ease the verification. In all
calculations, the string model is used, but comparison shows very good agreement
between numerically calculated and measured displacements. An analysis given in
[35]shows that
the solution is dominated by low-frequency components of f < 50 Hz. There-
fore, the string model is sufficient, and a crude discretization of Ax = 1 m may
be used.
convergence to a reference solution calculated with high resolution (Ax = 0.02
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

m) is linear, i. e. the numerical error E = II wA, - wXfll is inversely propor-

tional to N.

The analysis of free catenary vibrations allows to explain a general problem of

catenary simulation, namely the difficulties arising in (contact) force calculation.
The solution for the catenary displacement at a given point is therefore expressed
as a sum of hamonic components.
Evaluating the errors for displacement, velocity and acceleration analytically
shows that displacements converge with O(N), whereas velocities and accelera-
tions (and forces, as a consequence) cannot possibly converge. This problem is of
importance, since for convergent force calculations, a combination of model and
numerical method is needed which should give at least O(N3) convergence for the
displacements. For the Euler-Bernoulli beam model, the situation is even worse
since the highest frequencies depend quadratically on the resolution a, N2, -
which results in a need of O(NS) -convergence for the displacements, when
convergent force calculations are to be achieved.
The example of free catenary vibrations clearly reveals the problems arising in
force calculations. Although convergent results for the displacements can be
obtained, the calculated forces are not convergent. This effect does not result from
numerical inaccuracies, but from the coupling of the system components by point
forces. The application of forces to the catenary, which is discussed in the
following, will give more detailed information on this effect.
Application of a Moving Force. Principally, catenary and pantograph form a
dynamically coupled system and therefore cannot be treated separately. Neverthe-
less, modelling of the pantograph-catenary interaction can formally be divided into
two steps: Determination of the contact force and application of this force to the
catenary. Since small errors in the displacements of the contact wire lead to big
discrepancies in the contact force values, the accurate numerical treatment of the
force applicatictn is of extreme importance.
The mere application of a constant moving force to the contact wire causes a
principal problem for space-discretizing methods (dlAlembert, FDM, FEM) since
'the locus of force application does not necessarily coincide with a discretization
point. A common technique to solve this problem is to apply the force at two
 170 G. POETSCH ET AL.

adjacent discretization points in accordance with the relative distances in space

[11,23,34]. For FDM and FEM, it is shown in [35] that this procedure is
permissible (though not optimal), when high-order space discretization methods
are used in combination with small time steps.
An additional problem is introduced by the temporal variation of the moving
force. The variation of the force in combination with its motion is cause of a
Doppler effect [36] which leads to the excitation of high-frequency oscillations.
Again, the use of "small" timesteps is required. For the string model, for example
At a (Ax)/c must be chosen to resolve the high-frequency oscillations. Other-
wise, aliasing effects occur and lead to erroneous results. Hence, d' Alembert's
method is problematic here since timestep and space discretization are firmly
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

connected by the relation At = (Ax)/c.

For beam models, these problems are less pronounced, since high-order space
discretization and small timesteps must be chosen anyway for reasons of accuracy
and numerical stability [32].

2.1.4. Conclusions and Practical Considerations

The theoretical results stated above have shown benefits and drawbacks of all
numerical methods incorporated in currently available simulation programs. The
weakness of modal analysis and dlAlembert's method, however, lie in the
limitations imposed on the modelling and are impossible to be overcome.
FDM and FEM are not subjected to principal limitations and offer much more
flexibility with respect to modelling. Making allowance, e.g., for the simulation of
irregularly worn wires, is a possible extension for future programs which is
principally excluded by the other methods. With respect to accuracy, programs
using FDM and FEM are also currently to be considered best.
Although theoretical studies indicate that none of the programs currently in use
will be able to give convergent results for the contact forces, it is still possible to
extract valuable information from the numerical results. The reason is that the lack
of convergence is caused by the very-high-frequency components of the contact
force, which, luckily, only marginally influence the deflection of catenary and
pantograph and hardly contribute to the principal system behaviour. Realistic
contact force time histories can therefore be obtained by post-processing of the
calculated contact force values with low pass filters. This approach will be
discussed in Section 2.3, which treats the dynamic interaction of pantograph and
catenary.

2.2. Pantograph Dynamics

The aim of the investigations today and in the future is to improve the pantograph
design, because this system has an important influence on the run of the contact
force while interacting with the catenary. For the development of advanced
pantographs, a comprehensive understanding of the dynamical behaviour of today's
current collectors is essential.
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 171

The fundamental kinematics of passive collectors is based on three degrees of

freedom:
1. Lift up the pan head in the major range of about 2 m; this task is realised with
the scissors cclncept which turns a rotational movement of the arms in a vertical
movement of the pan head. Motions with frequencies up to 1 Hz or 2 Hz are
possible.
2. Level out displacements with middle high frequencies, up to 10 Hz; the contact
shoes have fle:xible suspensions, e.g. rubber elements (SBS 81) or translational
springs (DSA 350 S).
3. Compensate clisplacements with high frequencies; the contact strip itself is an
elastic body which can compensate displacements. The frequencies of these
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

excitations arr: very high in relation to the above mentioned motions, from 20
Hz to 50 Hz, but only with low amplitude.

To improve pantographs, the combination of all movements with their specific

dynamic behaviour must be taken into account. But also new concepts have to
accomplish these tasks, because they are founded upon the catenary construction.
Therefore, mathe:matical models have to be generated which represent the mechan-
ical system.

2.2.1. Low-order-Models of Pantographs

In the past, it .was common to use models with few translational degrees of
freedom (DOF) according to the fundamental kinematics. Real pantographs con-
tain non-linear force elements as well as non-linear kinematics, so the models with
linear DOF and often with linear force laws are only valid for one pantograph
layout at one operational point. It is not possible to take different shoe heights or
lateral contact points into account, or to simulate the concrete influence of side
forces or horizontal foot excitation. But in special situations these simplified
models are sufficient enough to check out a new draft of a control layout or a
catenary model. The variations reach from two mass models with linear force laws
to models whiclh include joint friction and bumpers and contain four and more
masses (Fig. 3). They contain the rotational movement of the strips as well as the
approximated elasticity of the upper frame [37].

frontview sideview
H. translation and ro~ationof the contact strips
rotation of the pan head

translation and rotation of the upper frame ZL

based on the elasticity
translation o f the lower frame

Fig. 3. 9 DOF simplified pantograph model.

 force ercrlallon
upper framc
\
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 4. H ~ g h s p c dpanlograph DSA 3-50 S. mcdclled In 5IMPACK

For more realistic and groun$ simulation motlels. non-linear 3D motleis heconie
absolurel y necessary.
The validity of these simple two- or four- mnss nioclels I S questionable, since in
such a model the mass needed for use in the calculat~onsis not rhe 'effective' or
'apparent' dynamic mass. This f i ~ u r e(a function of frequency and speed) is
somewhat greater rhan the weighed weigh1 of the collrctor head but less than the
weiyhed weight of frame components depending on the comple.~kinematics o l the
mechanism [?I.

2.2.2. 3D-Model @f Panro~raphr

In future. new design concepts or new ideas ro Improve usual concepts. will and
m u s ~he tesred by using dynamic simulations ~ncluileclin tlic desryn process. T h i s
p r ~ e d u r eenables kinematics and dynamic analys~sfrom the first design sketches
to the fully worked out construction.
For an effective interaction between CAD consrructlon. FTM analysis ancl MBS
simulation, it is necessary to combine h e w CrlE.~oolsfor settiny up a MBS
morlel consitlering the design data in a correct way.
As a part of a project of the DB AG, Adrranz ant1 the DI,R. a full 3D-model of
lhc hiphspeed pantograph DSA 350 S hat! to hc W I up using ~ h j smoclelling
strategy shown in Fig. 5.
The central dn[ahase is represented hy the MRS-~mllSIhIPACK [3,5]. which has
- bc~idcsobcrs - hi-direc~ional~nferfncesto Pr(7/fIN(;INl:EK (CAD) ancl ANSYS
(FEM?.
 PANTOGRAPK/CATENARY DYNAMICS, CONTROL

Fig. 5. Tools to set o p a Hybrid MBS-Model for SIMPACK.

Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

The first draft of the model was designed with Pro/E and ProSIM, the interface
to SIMPACK and a tool for analysing the mechanical properties in the Pro/E
environment. The model contained rigid bodies and the full 3D kinematic accord-
ing to the technical drawings. First studies with ProSIM were performed to check
the body and joint definitions. The result yield SIMPACK datafiles which were
completed, e.g., by the flexible parts of the pantograph which have significant
influence an the dynamic behaviour using the SIMPACK pre-processor.
In the extensive model, these flexible bodies and their elastic behaviour had to
be considered out of the design and material, not reduced to special force
elements, which is sometimes done for simplified mass models. The MBS tool
SIMPACK can deal with elastic bodies represented by eigenfrequencies, eigen-
and staticmodes. T o get these data, precalculations, with the FEM-program
ANSYS were performed. The interface FEMBS ([39,40]) converts these FEM
results in a SIMPACK readable dataformat. For simple constructions, e.g., the
contact strips, it is sufficient to use the pre-processor of ANSYS for setting up the
FEM model. More complex FEM structures of bodies, e.g., the upper frame, can
easily be designed and meshed in the environment of the CAD-code (ProMESH).
According to these procedures the elastic bodies were considered in the model.
For identifying the characteristics of important force elements, e.g., progressive
springs between the pan head, a couple of experiments were carried out. The
non-linear characteristic of the contact strip springs are represented by a third
order polynomial (Fig. 6).
The other irr~portantforce element is the pneumatic component with a very
complex behaviour based on a bellow and a control unit. This unit has to hold a
constant air pressure. For low frequencies, it works well, whereas for higher
periodic excitation, it can be treated as a closed valve. In the mechanic model, an
assemble of linear force elements substitute this behaviour.
The spring c , represents the static characteristic with full working pressure
control unit. If the frequency increases, the damper d becomes more tight and so
the spring c, gains more influence on the cumulative behaviour. The keyvalues
were fixed according to dynamic and static measurements, which were carried out
at a test stand a1 the DB AG.
 174 G. POETSCH ET AL.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 6. Characteristic of the contact strip spring: Left: Experimental results. Right: Approximation.

The result is a 3D-Pantograph model which is based entirely on the manufac-

ture's assessment. The main advantages are, that modifications of the construction
directly influence the model and can be analysed according to their dynamical
effect and that one model is valid for different operation points and various
boundary conditions.
2.3. Modelling of the Pantograph-Catenary-Interaction
Since a wide variety of different methods is used for the simulation of both
catenary and pantograph, differences are also to be found in the treatment of the
dynamic coupling. The only point in common to all known simulation programs is
that the moving contact between pantograph (collector strips) and contact wire is
modelled as a point contact.
The coupling of both systems is described by coupling equations, which form a
connection between state variables of the two systems. These equations usually
lead to differential-algebraic equations (DAEs) which characterise the overall
system.
2.3.1. Coupling Equations
Apart from some simulations which make use of simplified overhead models (e.g.,
[411), the coupling has always been performed by equating the vertical force

ax - throstle -+
pneumatic bellow

pressure control

Fig. 7. Modeling of the pneumatic bellow of thc DSA 350 S.

 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 175

components acting on contact wire and collector strips. The values of the contact
force are then determined by a second condition. Several methods are available
here:
One method, proposed by [23], consists in equating also the displacements of
both system.s. This approach makes no assumption on the contact wire model
and can be used for strings as well as for beam models. In the original work
[23], linear interpolation was used to determine the deflection of the contact
wire. This results in highly oscillatory results for the contact force. However,
the use of high-order interpolation can reduce this effect significantly, conserv-
ing the "general-purposew-character of the method.
Another popular approach is the use of left-hand and right-hand spatial
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

derivatives of the contact wire displacement at the contact point. Beside the use
of static contact conditions which lead to wrong results by force, other
conditions ;are also problematic [32]. The reasons lie in the limited spatial
resolution of any numerical method, which makes it impossible to represent the
short-wavellength components which are necessary to model discontinuities of
spatial derivatives. As a result, highly oscillatory components in the contact
force occur which are not present in the real system.
Due to these problems, other approaches like penalty methods or the introduc-
tion of a "stiff spring" between pantograph and contact wire have occasionally
been used [42]. The coupling conditions - same forces acting on pantograph and
catenary, which have the same displacements at the contact point - are thereby
loosened, and the properties of the overall system can be misrepresented to an
unacceptable degree.
Only if the catenary equations are solved by "global" methods like model
analysis ([12,1.5,17,18]) or use of a RITZ approach [21], the coupling is unprob-
lematic. The contact point displacements can then be obtained by superposition of
modal function~s,and there is no need neither for interpolating nor for distributing
the contact force to adjacent discretization points, as with FDM or FEM.
The princip.al drawbacks of these methods, however, have already been stated
in Section 2.1. Problems arising in FDM/FEM-simulations are not that severe,
because they may be overcome by combination with semianalytical methods (see
Section 2.3.2).
Using "pure" FDM/FEM-methods will give rise to high-frequency compo-
nents of the contact force, which must be eliminated with low pass filters in order
to obtain realistic results. This deficiency is compensated for by the absence of
principal limitations imposed on the modelling.
2.3.2. Post-processing of Contact Force Values
An example of post-processing calculated contact force values is given in Fig. 8.
On the left, the contact force obtained for a model problem (travelling mass in
contact with a string of finite length) by application of a semi-analytical method is
displayed. The time history of the contact force shows the typical amplification of
contact force peak values as the mass approaches the support. The right figure
 176 G. POETSCH E T AL.

Fig 8. Contact Force, Semi-Analytical and Numerical Solutions.

FIN
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

300 - - semi-analytical solution, filtered

---- numericalsolulion, fillered
200 -

100 --

0
*
0 10 20 30 40 50 60 xlm

Fig. 9. Contact forces after postprocessing with a low pass filter.

shows the results obtained for the same problem with a .high-resolution FDM
calculation. Selectively, strong differences occur; nevertheless, both solutions
show similarities. Figure 9 shows both results after application of a low pass filter.
With respect to the filtered reference solution, the filtered numerical solution can
be considered as almost exact2.
The example shows that forces as calculated by a method convergent only for
contact point displacements, but not for contact force, can still provide reliable
information within a certain frequency range.
The post-processing of the contact forces has also been used on results of
full-scale catenary/pantograph simulations [35].In comparison with measurement
data, good agreement has been achieved.
Future enhancements of the coupling method will mainly affect computation
times, which is an interesting aspect for the assessment of effects caused by
parameter variations, or for performing a simulation-based system optimisation.

2.3.3 Summary
The analysis of catenary models and numerical methods given in Section 2.1 has
revealed several problem characteristics which require special attention in per-
forming dynamic catenary simulations. Accurate numerical handling of the cate-

The differences occuring when the travelling mass approaches the right suppon result from the
coupling conditions in the semianalytical solution, which permitted negalive values of the contact
force.
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 177

nary dynamics is of utmost importance for the analysis of the catenary-pantograph

interaction. A seemingly small misrepresentation of the contact wire deflection
will lead to strong errors in the determination of the contact force, which makes it
difficult to asse:ss the system quality.
Since the development of the first simulation programs, significant process has
taken place mainly with respect to structural modelling flexibility. Several pro-
grams have been developed which allow simulation of almost arbitrary catenary
designs (e.g., [22,29,31,35]). However, accuracy requirements have only rarely
been considered.
The catenary-pantograph coupling can be achieved using simple equations
("same contact force - same displacements"). These equations can be solved
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

iteratively for the forces, which allows to provide a general interface between
pantograph and catenary.
With state-of-the-art methods, it is now possible to extract reliable information
about the system quality by postprocessing of the numerically calculated contact
force values. :The development of fast algorithms which allow direct system
quality assessnlent is of major concern for future simulation tools. A possible
approach is the combination of FDM with semianalytical methods which provide
more exact trea.tment of the moving pantograph-catenary contact.
The DB AG uses a FDM-based simulation tool for catenary simulation, which
has been developed at HNI/Paderborn. Until now, only a few pantograph models
can be simulated; since the design of pantograph models "by hand" is an
expensive task, a program interface to the MBS System SIMPACK is now being
developed. It its planned to use the tool for engineering purposes, reducing the
costs of the de\relopment cycle and the need for costly measurements.

3. DATA ACQUISITION, PARAMETER DETERMINATION, VERIFICATION,

VALIDATION

3.1. Acquisition of the Model Data

In principle, modelling of pantograph/catenary interaction is relatively straightfor-
ward, provided that the separate pantograph and overhead subsystems are ade-
quately understood.
Whatever type of mathematical models are used, the key to the solution is to
define the contact point displacement as a function of contact force. Simultaneous
equations can be derived giving independent expressions in terms of pantograph
degrees of freedom, and in terms of overhead system degrees of freedom [2].
Recent developments in pantograph and catenary dynamics have focused
attention on techniques for modelling the subsystems to ever higher degrees of
detail. However, whilst developing more complex models of either the pantograph
or overhead designs is a valuable exercise in improving our understanding of the
system, i t may not necessar~ly lead to more accurate simulation of
pantograph/calenary interaction under realistic conditions.
 178 G. POETSCH ET AL.

The dynamic performance of a pantograph/catenary system is dominated by a

relatively small number of key system components, which are not easy to control
in a practical situation [43]. It must be appreciated that the accuracy of the
simulation can be no better than the accuracy of knowledge of these key system
parameters. Some of the most fundamentally important parameters are:
Contact wire tension - maintaining a high contact wire tension is crucial to
achieving a good standard of current collection, yet most auto-tensioning
devices are far from perfect. On balance-weighted systems, tension variations
of 10% have been observed within a few hours.
Contact wire sag - a useful technique to optimise pantograph performance is to
use the contact wire sag profile to compensate for variations in overhead
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

system compliance throughout the span, such that the pantograph runs with a
near constant height trajectory. However, errors of greater than 2 mm at each
dropper position can have a significant effect on system performance. In
practice, it is difficult to achieve an accuracy of better than 10 mm in contact
wire height.
Aerodynamic forces - forces on pantograph components are a crucial factor
influencing system performance, particularly wire uplifts. Yet, they remain
difficult to assess and can vary considerably during normal operation. Key
factors influencing aerodynamic forces are the direction of running, position on
the train, locomotive/power car design, wire height, track topography (em-
bankments/cuttings/tunnels), wind strength and wind direction.
Overhead line damping - this is difficult to measure and to model, but it is a
very important factor influencing system performance when several pan-
tographs are operated in multiple. The level of overhead damping determines
whether trailing pantographs will be running onto a relatively smooth wire, or a
high1y disturbed wire.

3.2. Data for the Verification

A crucial problem to be overcome during testing is the measurement of contact
force. This is the single most important parameter in the field of pantograph/over-
head dynamics, and yet it is very difficult to measure. The most common method
of recording contact force is to measure the forces at the support points of the
contact strips, and to compensate for the inertia of the contact strips by measuring
their accelerations. However, due to the flexibility of the contact strips this
synthesised measurement has a linear frequency response up to only 5 - 10 Hz
depending on the mass and stiffness of the strip [44]. A further inaccuracy is due
to the aerodynamic forces on the contact strips themselves, for which allowance
must be made. The nature of these aerodynamic forces can be asskssed by separate
aerodynamic testing with a tethered pantograph, but this can only give an
indication of the likely behaviour during the dynamic tests, as the forces are
subject to unpredictable fluctuations due to localised wind variation and changes
in wire height. All the instrumentation nust be very carefully designed to
minimise interference with the aerodynamic characteristics of the pantograph.
 PANTOGRAPH/CATENARY DYNAMICS. CONTROL 179

More direct. measurement of the true contact force has been achieved with a
thin contact strip supported by load cells all contained within the standard contact
strip profile [44]. This has proved very useful for the measurement of higher
frequency contact force variations at features such as neutral sections, and has
been used to validate traditional measurements. However, instrumenting the whole
swept width o:f both contact strips of a pantograph with such a system would be
difficult and expensive, and the life of the measuring devices would be short due
to wear of the thin contacting strips. The method is also limited to nonenergised
overhead equipment.
T o keep errors in the validation exercise to a minimum, it is vital to maintain
close control of key system variables. This is best achieved by testing on an
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

isolated, well maintained overhead length and running with a specially prepared
and aerodynamically controlled pantograph. However, aerodynamic effects can
never be completely defined, and it is advisable to incorporate an element of
aerodynamic testing into the experimental program. In this way, a truer picture can
be obtained of uplift forces acting on the pantograph during the test runs.
Provided that deviations in critical system parameters are kept to a minimum,
good agreement can be obtained between predicted and measured contact force,
even with relatively simple mathematical models. The benefits of enhancing the
models by using a more detailed representation of the pantograph mechanism, or
including the bending stiffness of conductor wires, or including lateral degrees of
freedom, depend on whether predictions are sought in the near region of existing
operation and equipment or new regimes and innovative designs are to be
considered.

3.3. Validation and Application

Pantograph models can be validated by means of laboratory bench tests. The
simplest option is to consider only forces and displacements at the level of the
contact strips. Presentation of the results in terms of the apparent mass of the
pantograph plotted against frequency will permit a general assessment of the
accuracy of the mathematical model.
The response of the pantograph is often dependent upon operating height,
amplitude of excitation, and the precise position of the input force. Fortunately,
evidence suggests that the combined pantograph/catenary system response is
relatively insensitive to the precise details of the pantograph model.

3.3.1. Low-order Models of Pantographs

The relatively simple two or three mass models appear to work up to frequencies
of at least 20 Hz; higher frequency analysis may require a more complex models
incorporating independent motion of the contact strips.
Figure 10 (left) compares the results from laboratory tests carried out on the
high speed paritograph DSA 350 S with simulated test results; the graphs show the
variation with frequency of the apparent mass of the pantograph. The simulated
results were obtained using a relatively simple three mass model based upon the
 180 G. POETSCH ET AL.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 10. Theoretical vs. experimental response of pantograph (DSA 350 S). Left: Initial model. Right:
Modified model.

accepted technical specifications for the design of pantograph. The initial compari-
son suggested that there might be some significant error in the mathematical
representation of the pantograph. However, relatively minor modifications to the
parameters of the model enabled the agreement to be improved to that shown in
Fig. 10 (right).

3.3.2. 3 0 Models of Pantographs

It is often the case that the first drafts of models based entirely upon the
manufacturer's assessments of the dynamic properties of the pantograph compo-
nents gives a relatively poor match with the experimental response. Accurately
quantifying more detailed components of the model requires extensive testing of
the individual linkages of the pantograph mechanism. This disadvantage is com-
pensated because a fully validated 3D- model of pantographs can be used for
different operating heights or other modifications of the operation conditions
without a new update of the significant parameters. Figure I I shows the responds
of the model of the DSA 350 S presented in Section 2.2.
The obvious difference to a mass model is the range of frequency, where the
model is defined. Here, results can be calculated up to 60 Hz, because the
flexibility of the contact strips are considered. Besides this, the model is valid for
different operating heights.

3.3.3. Caterlary
To validate the overhead model, a full scale resonance test of an installed overhead
system can be a major exercise involving considerable expense. Where this
procedure has been carried out, testing has confirmed that standard catenary
modelling techniques can predict resonant frequencies and mode shapes to a high
degree of accuracy. Overhead damping has been observed to have some frequency
dependency, but is generally in the ortler of I % of critical. The level of damping is
not too significant when the overhead model is used to simulate the response of a
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 18 1

1 10
Frequency [Hz]

DSA350S/H=l.5mF 0=120N:AF=20N
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

10
Frequency (Hz]

Fig. 1 1 . Simulated and measured apparent mass of the DSA 350 S camed out with two different
operation heights and load parameters.

single pantograph, but becomes much more important when multiple pantograph
operation is considered.
A simpler option of assessing the accuracy of an overhead model may be to
measure the response of an overhead system to a step forcing input. This will give
an indication as to the dominant natural frequencies of the system and enable an
assessment of overhead damping to be made.

3.3.4. Interaction
Validation of full pantograph/catenary interaction cannot be achieved without an
extensive propam of running tests at high speeds on a real overhead system.
Scaled line sy.stems with short span lengths and increased wire mass have also
been used with some success for predicting behaviour at very high speeds. Figure
12 shows the level of agreement which can be obtained between theoretical
predictions and experimental measurements of contact force, on a carefully
controlled overhead length. (These tests were carried out on an experimental
pantograph rur~ningon a have scale trolley wire overhead equipment - hence, the
peaks in contact force which occur at every structure).

3.3.5. Typical .4pplicarions

Where a completely new overhead system is to be installed, simulations can give a
guide to maximum operating speeds on the system. However, there is a need to
 G.POETSCH ET AL.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 12. Comparison of theoretical and experimental contact force time histories.

allow a safety factor to take account of the differences between a design case
installation and a real system. This factor will depend upon the quality of
maintenance which can be guaranteed and the extent to which occasional loss of
contact can be tolerated; it is best assessed with reference to existing applications
on the network.
There is a tendency for overhead systems and pantograph designs to evolve
together. However, a pantograph with a proven record of satisfactory operation at
high speed on one particular overhead design may prove to be incompatible with a
different high speed overhead system. Simulations can be used to check panto-
graph/overhead compatibility.
Operators may wish to run new train formations on an existing overhead system
with increased numbers of pantographs per train, at different spacings. Simulations
can indicate any possible deterioration in performance.
Options for upgrading overhead systems can be studied and compared with
existing performance. This enables the operators to identify the most cost effective
solution for the required increase in performance [3].
Having validated the response program for a basic open route overhead system,
the overhead models can then be developed, with some degree of confidence, to
include specific overhead features - for example, overbridges, overlaps, neutral
sections, etc.
These types of features are usually the weakness on any overhead network.
Simulations can identify the most appropriate design modifications to achieve a
localised improvement in current collection performance. Figure 13 illustrates a
typical application for a simulation program - assessing the benefits of proposed
modifications to the catenary design at a particular overbridge, which the operators
have identified as having current collection problems in high temperature condi-
tions. (Note that the short periods of apparent negative contact force are simply an
effect of filtering the calculated output functrons).
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 183

Bridge design No. 1.43 deg - High Speed Pan - 90 mil&

Train Speed = 40.2 mls - Filter Frequency = 15.0 Hz -Om Damping = 1.0%

E 2m
s
5 193
P
8 loo
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Bridge design No. 2.43 deg - High Speed Pan - 90 mil&

Train Spec:d = 40.2 mls - Filter Frequency = 15.0 Hz - OIH Damping = 1.0%

Fig. 13. Sirnulalion of responded catenary arrangements for an ovcrbridge.

The validation process has required a very precise knowledge of the dynamic
parameters of pantographs and overhead equipment and it is sometimes believed
that this makes the prediction procedure unsuitable for practical use. However, it is
the opinion of the authors that quite the converse is true: For a properly modelled
system, prediction based on simulation is correct, so that if varying a particular
parameter causes significant changes of the behaviour, then varying that parameter
would similarly changes the performance of the real equipment. Prediction rneth-
ods can therefore be used to rank different proposals with reasonable confidence.
 184 G. POETSCH ET AL.

4. PANTOGRAPH TECHNOLOGY: FROM THE PRESENCE T O THE

FUTURE

4.1. Pantographs, 'State of the Art'

The current collectors which are used all over the world today are passive systems.
The first widely circulated types have been symmetric pantographs. This concept
is well up to now, sometimes in the original design, sometimes in a modified
version, the asymmetric configuration.
The advantage of this basic concept is, that a relatively light construction is
combined with a high mechanical stiffness. Only little space is required, when
they are folded, but even though they serve a wide range of catenary heights.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Furthermore, it can be adapted very easily to different requirements. By modifying

some significant parameters (e.g., the length of the arms), the kinematics are
changed, or the suspension elements can be arranged in different ways. Even
combinations of both concepts are possible as it is shown in Fig. 20.
The French railway company SNCF and the manufacturer Faiveley tried to
solve the dynamic problems by using a dual stage pantograph. Field tests
demonstrated, that the quality of current collection could be improved, but the
wear of the shoes and the contact wire increased because of the higher static
contact force component. Additional a telescope pantograph was proposed, but
never realised [45].
Another kind of dual stage collectors was designed in Russia for trains
connecting Moscow and Leningrad. The construction of 1972 was outlined for
speeds up to 200 km/h and contains two combined frames, Fig. 15. A lower
frame build like a parallelogram was connected with an upper frame, a symmetric
pantograph containing the panhead. Each system is connected with mechanical
controlled pneumatic actuators. Experiments at the test stand were convenient [46].
In the year 1971 the British Railway Board tried to reach the speed-range over
200 km/h by using a pantograph based on the AM-BR. The ambition of this
prototype was to test different combinations of dampers and springs [47]. For
example adjustable viscous dampers (Fig. 17) and suspensions with a range from
35 mm to 75 mm were tested.
Some railroad administrations have favoured a policy of deliberate augmenta-
tions of the contact forces by aerodynamic uplift forces in an attempt to reduce the
different aerodynamic characteristics in the two directions of running. For example
the asymmetric pantograph of Brecknell, Willis & Co. Ltd. was improved by

Fig. 14. Left: Symmetric construction. Right: Assymetric construction.

 PAhTM;RAPH /CATE"IARY DYNAMICS. COYTROL 185
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

FIR. 1.5 Dual S t q c hlopraph of Fa~iclc!.

-
-
Upper fmmc

Lower frame

adding an aerofnil which compcnsstcs the ncrmlynnni~cdifference so [hat the

de5ign uplift force i~achiever! in h t h d~rcct~nnf. For thts reason the aerofoil s h a p
was delikrately choscn to act only in a F I ~ F I P dtrccr~on[#I.
A s a conwquencc. it i 4 u s u a l l ~xceptctl policl; that minimisine the acrody-
n:tniic LIP- or downllfr forcrli w ~ r h ~ v c dand ~ilrecrtnnof running is optimal,
althou~hi t seems ver?; unlikely that I< ha\ k e n ach~cbcdin the pnq.
 G. POETSCH ET AL.

Fig. 18. Catenary with Integrated Dampers (Japan).

In Japan, the railway companies try to solve the dynamic problems with
damped catenaries, [49]. The droppers are substituted by airspring-damper ele-
ments (Fig. 18). This variation enables a constant height of the contact wire, so a
very compact designed pantograph can be used. It must only serve a range of 550
mm which enables a very light and therefore highly dynamic construction.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

4.2. Improving Existing Concepts with Active Control

As we have discussed in the previous section, aerodynamic behaviour is complex
and is influenced by many parameters such as pantograph geometry, direction of
travel, roof environment, wind conditions etc. Although the magnitude of these
effects is difficult to calculate theoretically, it is a relatively simple matter to
measure on the vehicle the overall uplift force. From here it is a small step to use
this measurement for feedback control.
As a consequence also a number of other effects can .be improved significantly
such as span passing frequency effects and performance on gradients and trolley
wire. In terms of apparent mass: the effective frame mass of the pantograph can be
reduced by use of a serb . Moreover, the contact force increments will be reduced,

overhead wire. \
causing less concern ove local wire wear and imparting less disturbance to the

With high speed vehicles and railway trains now proposed, satisfactory power
collection by conventional pantograph/catenary systems arrived at its limit [SO].
Successful current collection at velocities up to 400 km/h requires either
- highly developed catenary systems using conventional passive pantographs or
- standard catenary systems using sophisticated active pantographs.
The more economic way is to solve the problems improving the present
concepts with active components.
Validated models of the passive system can be used as the basis for such
modifications. For example, a symmetric pantograph was improved by adding a
linear motor between the'mainframe and the contact shoe or placing a torque
motor between a mainframe rotating arm and the vehicle roof [S I], shown in Fig.
19.
In [52] it is proposed to modify an asymmetric pantograph by adding a panhead
based on the mainframe concept.
The active pantograph, shown in Fig. 20, can be described with the mathemati-
cal model, shown in Fig. 21. It includes a mass m , to represent the shoe, a mass
m, to represent the main frame and a force AF(t) to represent the dynamic contact
force between shoe and catenary.
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 187

-
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Load cell Accclrromclcr

-..-A&
I '.T

Fig. 19. Improving symmetric pantograph concepts.

Finally, two controlled hydraulic actuators are included to generate the sec-
ondary suspension force AF,(t) and the primary suspension force AF2(t).
According to the structure shown in Fig. 20, the control system depends on
controller R1 (contact force controller) and controller R2 (frame controller).
Controller R1 has a high bandwidth but the amplitude of the according command
signal is small.. Controller R2 has a small bandwidth but the amplitude of the

F(t) = Fo + AF(t)contact force

~l(t)
effective shoe mass

AFl(t): secondary suspension force

220)
effective frame mass

A AF2(t): primary suspension force

Fig. 20. Active panto~raphwith contact force controler R1 and frame controller R2.
 188 G. POETSCH ET AL.

-
contact

actuator
A?, c t,
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 21. Simplified Model of the Pantograph.

according command signal is large. This is necessary to use also overhead lines
with various height of the contact wire as in tunnels and under bridges.
Contact force controller R1:
From the measurements, shown in Figure 20, in the time invariable controller
R1 the force command AF,(t) is computed as

From the measurements shown in Figure 20, in the time invariable controller
R2 the force command AF2(t) is computed as

In praxis the velocity S , is estimating by a reduced observer because the signal

cannot be measured by a sensor [52].

Simulation investigations have been performed based on the equation given for
the pantograph and catenary system described with modal analysis technique using
eigenvalues, eigenfunctions, catenary mass and modal damping ratio [53,30]. The
linear time varying system is integrated numerically over one period (four spans)
using the computer program FAST (Fahrleitung/Stromabnehmer). The initial
conditions of the catenary and pantograph are assumed to be zero and wind forces,
hog-sag of the wire and the vertical motion were neglected.
Figure 22 and Figure 23 illustrates plots of the simulation results with passive
and active pantographs. The dynamic contact force and the secondary suspension
force are given in relation to the corresponding static forces, so that level 0
indicates a force between shoe and contact wire equal to that in the static case, and
level - 1 means zero force and hence loss of contact. The shoe height and the
 PANTOGRAPH/CATENARY DYNAMICS. CONTROL
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 22. Passive pantograph shoe height and contact force at v = 100 m/s.

actuator amplirude of the secondary suspension are given in relation to the static
midspan deflection of the catenary.

4.2.1. Control Concepts of the DSA 350 S

The Deutsche IBahn AG and the DLR proposed an active suspension of the contact
strip, [9], as well as Adtranz did in [45]. This concept was tested using the
non-linear MBS-model of the DSA 350 S whichwas set up in SIMPACK (as it is
described in S.ection 2.2.2), containing sensors at the suspension of the contact
strips and actuators as pictured in Fig. 24. The force of the actuator is induced
serially to the spring of the strips, the actuator itself is considered as a small mass,
which was arranged in parallel to the actuator spring. The sensor signal is the force

Fig. 23. Active pantograph shoe height and contact force at v = 100 m/s.
 G. POETSCH ET AL.

contact strip

actuator mass

pan hed
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Fig. 24. Schematic composition of a pan head control concept.

of the strip-springs. The catenary is reproduced by a spring c(t) with a variable

stiffness as a function of time (see Section 2.1.1). A data record is read in from a
test drive into the force element of the spring c(t). It represents the real existing
spring stiffness of the section.
For the first draft of a controller, it is not necessary to consider the full
non-linear model. So the linear system matrices were generated for the operation
point and implemented in a control loop in MatrixX [54]. For the first tests of this
concept a PI-controller (P = 100; T = 1) and ideal actuators were used. The aim
was to control the contact force to F = 180N. The comparison between the
controlled and uncontrolled contact force is pictured in Fig. 25.
The simulation shows, that the contact force of the strips, supported by the
controller, is more low-frequent than the uncontrolled signal. The maximal span of
the uncontrolled contact force A ,,,,,,,,,,, = 118 N and the controlled span
A,,,,,,,,,, = 1 1 N show the reduction of the force dynamics. Also the variance of
uuncontrolled= 19 N compared with u,,,,,,~,~ = 2.2 N presents a clear enhancement
of the force signal.

Contact Force

Fig. 25. Spring force of the pan head: with and without control loop.
 PANTOGRAPH/CATENARY DYNAMICS.CONTROL 191

To test the robustness of the modification measures, further simulations where

performed using the full non-linear model. Therefore the controller layout was
copied in the MBS tool SIMPACK completing the 3 D-model. The tests with
several boundruy conditions showed that the concept works well under different
operating heiglnts or excitation levels.
Another proposal of the DLR is to control the level of the contact force by
active components arranged at the pneumatic bellow or at the knee of the
pantograph comstruction. These actuators can for example be made out of piezo-
electrical elements as it is described in the pending patents [55] and [56].

4.2.2. Conclusion
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

Simulation results and test runs show, that long time power collection with passive
pantographs on standard overhead line is possible up to 250 krn/h. For higher
train velocities satisfactory current collection requires active pantographs with a
high bandwidth contact force control or special improved catenary equipment as it
is used by the SNCF.
For overhead lines with varying height of contact wire (as in tunnels and under
bridges) a frame controller with low bandwidth is included. Because the velocity
in for exampl~:Fig. 20 cannot be measured by a sensor, a reduced observer, as
shown in [52], is used to estimate the signal. Simulation results with included
actuator dynamics are also discussed in [52].
I
4.3. Developing new Concepts with Mechatronic and Concurrent Engineering
Methods
At present, investigations are being performed within a joint venture group,
consisting of DB AG, Adtranz and DLR, with the scope of developing an
innovative "high-performance pantograph". The objective is to produce a panto-
graph prototype which, being equipped with active contact control, can run on the
existing catenary network at higher speeds (up to 25%) without the operational
limitations mcntioned in this survey. Stricter noise protection regulations are
forcing the railway operators to reduce noise emission also for pantographs.
Reduction of noise is becoming more and more important for a positive response
from the public towards the railway. Aeroacoustic demands have been largely
neglected during the development of pantographs up to now. In the "high-perfor-
mance pantograph" project, aeroacoustic and aerodynamic specifications are now
to be integrated during the design of the new pantograph and potentials for
reducing noise: are to be explored. Among other things, it is planned to integrate an
"on-line diagnostics system" for early detection of faults which analyses the
status of the catenary/pantograph system from sensor signals during running and,
where appropriate, initiates automatic reactions or provides information concern-
ing the present status of the line to the system support. The findings gained during
development in the respective specialist areas such as mechanics,
aeroacoustics,~aerodynamics or regulation/control are to be further investigated
with regard to their useful application for improving already existing pantographs.
 G.POETSCH ET AL.

Fig. 26. Theoretical concepts of new advanced pantographs.

New concepts of current collectors must consider a wide range of requirements:

Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

- low aerodynamic drag,

- low aeroacoustic emission,
- uninterrupted and reliable transmission of electric energy,
- low abrasion of the contact wire and contact strip,
- robust and long life construction,
- keep restrictions of laws and conditions.

In the past,,especially the last point hindered the development of new collector
concepts in Europe. But not only developments in Japan demonstrated that new
kinematic solutions are possible.
Fundamental examinations of theoretical concepts will show the best compro-
mise of the above mentioned requirements, maybe it will base on a telescope or
one arm concept as sketched in Fig. 26. One aspect is sure: new concepts will be
composed of less components designed with larger cross-sections. Especially the
small parts of the today used constructions produce much noise, so that they have
to be redesigned. Furthermore, the dependency between the wear of the carbon
strips and the aerodynamic uplift must be controlled in the future.

5. CONCLUSIONS

This s u ~ e yhas been dealing with the complex system dynamics of pantographs
and catenaries. It has turned out that this complex dynamic interaction has become
a stumbling block for higher speeds, less wear, higher operational safety as well as
reduced (aeroacoustical) noise. A complete understanding of the physical phenom-
ena, a careful mathematical modelling as well as identification of the crucial
parameters is mandatory for analysing the shortcoming of the present systems as
well as to design and investigate new concepts.
As being typical for Europe many different systems are in operation with
numerous experiences and unification is slow because of the immense costs
involved. Even more advanced computational facilities are needed to predict with
a high degree of fidelity the improvements which can be achieved by minor
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 193

modification of' the infrastructure but by courageous steps with new and certainly
active pantographs.
Consideratica of overhead and pantograph as parts of one dynamic system
provides a basis for the idea to improve pantographs for better use of existing
overhead equipment, or less complex equipment for higher speeds. The ultimate
extension of this argument from a current collection standpoint is the potential use
of a single wire or trolley wire system. With the now existing CAE-simulation
tools and the validated models it is also possible to suggest more economical
designs for sonie new electrification projects.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

6. REFERENCES

1. Eastam. T. R.: Em goldenes Zeitalter f i r Hochgeschwindigkeitszuge? Spektrum der Wissenschaft.

Special 4, Schlusseltechnologien 1995.
2. Gostling, R. 1. and Hobbs, A. E. W.: The Interaction of Pantograph and Overhead Equipment:
Practical Applications of a New Theoretical Technique. Proc. 1. Mech.E. Vol 197 No. 13,
January 1983.
3. Scott. G. A. and Cook, M.: Extending the Limits of Pantograph/Overhead Performance. Better
Journey Time - Better Business. I. Mech. E. Conference Transaction 1995 - 8, Mechanical
Engineering Ibblications, Bury St. Edmunds, UK. ISSN 1356-1448. ISBN 0 85298 997 0.
4. Nibler, H.: Dynamisches Verhalten von Fahrleitung und Stromabnehmer bei elektrischen Haupt-
bahnen. Elekuische Bahnen. 2 1 (10): 234-241. 1950.
5. Gilbert, G. and Davies. H. E. H.: Pantograph motion on a nearly uniform railway overhead line.
Institution of Electrical Engineers: Proceedings. 113 (3): 485-492. March 1966.
6. Bauer. K. H., Bucksch. R.. Lerner F.. Mahrt R. and Schneider F.: Dynamische Kriterien fiir die
Auslegung von Fahrleitungen. ZEV Glasers Annalen. 103 (10): 365-370. October 1979.
7. Manabe. K.: Catenary-Pantograph System for Speedup of Shinkansen Train. Japanese Railway
Enginwring, 30 (1 17): 10- 13, March 1992.
8. Heigl. H.: MeSeinrichtung zur Registrierung von Kontaktunterbrechungen zwischen Fahrdraht und
Stromabnehrner. Elektrische Bahnen. 36 (7): 171- 174. 1965.
9. Barlage, J.: Snmulation aktiver Stromabnehmer. Diplomarbeit, lnstitut fur Regelungstechnik.
Technische Universifat Braunschweig 1996.
10. Ebeling. H.: Stromabnahme bei hohen Geschwindigkeiten; Probleme der Fahrleitungen und
Stromabnehmer. Elektrische Bahnen, 40: 26-39, 60-66. 1969.
11. Fischer, W.: E:ine Methode zur Berechnung des Schwingungsverhaltens von Kettenwerk und
Stromabnehmer bei hohen Zuggeschwindigkeiten. PhD thesis. Technische Hochschule Darm-
stadt, January 1975.
12. Levy. S.. Bain, J. A. and Leclerc. E. J.: Railway Overhead Contact Systems. Catenary-Pantograph
Dynamics for Power Collection at High Speeds. Transactions of the ASME, pages 692-700,
November 1968.
13. Abbott. M. R.: A Numerical Method for Calculating the Dynamic Behaviour of a Simple Catenary
Overhead Col~tactSystem for Electric Railway Traction. Technical Report 67156, Royal Aircraft
Establishment. July 1967.
14. Andrews. H. I.: Calculating the Behaviour of an Overhead Catenary System for Railway
Electrification.. Institution of Mechanical Engineers: Proceedings. 179 (25): 809-828. 1964.
15. Brodkorb, A. and Semrau. M.: Simulationsmodell des Systems Oberleitungskettenwerk und
Stromabnehmer. Elektrische Bahnen. 91 (4): 105- 1 13. 1993.
16. Bianchi. C. andl Tacci, G.: Stromabnehmer fur Gleichstrornbahnen. Elektrische Bahnen. 89 (1 1):
429-435, 199.1.
17. Hobbs. A. E. 'W.. Illingworth R. and Peters A. J.: New Developments in Understanding the
Dynamics of Overhead Current Collection Equipment for Electric Railways. Closed Loop, pages
3-9. February 1977.
 194 G. POETSCH ET AL.

18. Link. M. and Nowak, B.: Studie zur Auslegung eines Stromabnehmers fur den Intercity Experi-
mental (ICE) durch Simulation des Schwingungsverhaltensvon Stromabnehmer und Fahrleitung.
AbschluBbericht FGL 3/1985, Universitst - GH Kassel. 1985.
19. Manabe, K. and Fujii, Y.: Resonanzen des Oberleitungssystems bei Verwendung von mehreren
Stromabnehmem sowie Gegenmdnahmen. Elektrische Bahnen, 88 (1 1): 403-409. 1990.
20. Oda. 0.. Morikawa, T. and Kusumi. S.: Reduction Method of Contact Losses of Overhead-Catenary
and Pantograph System. Quarterly Reports, 27 (4): I 15- 120, 1986.
21. Renger, A.: Berechnung des dynamischen Verhaltens von Oberleitungskettenwerk und Stromab-
nehmer. VDI Berichte. (820): 141- 154. 1990.
22. Resch. U.: Simulation des dynamischen Verhaltens von Oberleitungen und Stromabnehmem bei
hohen Geschwindigkeiten. Elektrische Bahnen. 89 (1 1): 445-446. 1991.
23. Scott, P. R. and Rothman. M.: Computer Evaluation of Overhead Equipment for Electric Railroad
Traction. IEEE Transactions on Industry Applications, IA-10 (5): 573-580, September/October
1974.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

24. Becker. C., Konig. A., Resch. U. and Zweig, B. W.: Hochgeschwindigkeitsfahrleitung; Ein Thema
fir die Forschung. Eisenbahntechnische Rundschau, 44 ( 1-2): 64-7 1. 1995.
25. Vinayagalingarn. T.: Overhead Current Collection with a Servo-assisted Pantograph. PhD thesis,
University of Birmingham, August 1974.
26. Petri, K.: Vergleichende Untersuchung von Berechnungsmodellen zur Simulation der Dynamik von
Fahrleitungs-Stromabnehmersystemen. PhD thesis, Universitat GH Paderbom, January 1997.
Heinz Nixdorf Institut. Verlagsschriftenreihe, Band 14.
27. Bucksch. R.: Beitrag zum Verst.%ndnis des Schwingungsverhaltens eines Fahrdraht-Kettenwerkes.
Wissenschaftliche Berichte AEG-Telehnken, 52 (5): 250-262, 1979.
28. Smith, C. E.: Motions of a Stretched String Carrying a Moving Mass Particle. Journal of Applied
Mechanics. pages 29-37. March 1964.
29. Konig. A. and Resch. U.: Numerische Simulation des Systems Stromabnehmer/Oberleitungsliet-
tenwerk. Presseinfonnation pri 1379, AEG. November 1993. 28. Technixhes Presse-Colloquium
vom 3. bis 4. November 1993 in Berlin.
30. Meisinger, R.: Modellbildung und Simulation einer Eisenbahn-Fahrleitung mittels FEM-Modalap-
proximation. In Simulationstechnik. 9. Symposium in Stuttgart, pages 475-480. F. Vieweg &
Sohn Verlagsgesellxhaft, October 1994.
31. Reinbold. M. and Deckart, U.: FAMOS - Ein Programm zur Simulation von Oberleitungen und
Stromabnehmem. ZEV-Glasers Annalen. 120 (6): 239-243. June 1996
32. Poetsch. G. and Wallaschek. J.: Simulating the Dynamic Behaviour of Electrical Lines for
High-Speed Trains on Parallel Computers. In International Symposium on Cable Dynamics
(Proceedings), pp 565-572. A.I.M., Liege, October 1995.
33. Link, M.: Zur Berechnung von Fahrleitungsschwingungen mit Hilfe frequenzabKangiger finiter
Elemente. Ingenieur-Archiv, 5 1: 45-60, 1981.
34. Morris. R. B.: The Application of an Analogue Computer to a Problem of Pantograph and
Overhead Line Dynamics. Institution of Mechanical Engineers: Proceedings, 179 (25): 782(808.
1964.
35. Poetsch. G.: Untersuchung numerischer Verfahren zur Simulation von Stromabnehmer-Ketten-
werk-Systemen. Universitat-GH Paderbom (to appear).
36. Bucksch. R.: Theorie der Wechselwirkungen von Fahrdrahtwellen mit angekoppelten mechanis-
chen Systemen (I). Wissenschaftliche Berichte AEG Telefunken. 54 (3): 129- 140. 198 1.
37. Resch. U.. Konig, A.: Simulation der Kontaktkraft zwischen Fahrdraht und Schleifleiste des
Systems Oberleitung/Stromabnehmer. VDI-Berichte Nr. 12 19, 1995.
38. Rulta. W.: Bewegungsgleichungen fur hybride mechanische Systeme mit freiheitsgradpropor-
tionalem Rechenaufwand. DLR intemer Bericht IB 5 15-86/ 12, 1986.
39. Wallrapp. 0.: Standardisation of Flexible Body Modeling in Multibody System Codes. Part I:
Definition of Standard Input Data. Mech. Stmct. & Mach.. 22 (3). 283-308 1994.
40. Wallrapp. 0 . and D. Sachau D.: Space Flight Dynamic Simulations Using Finite Element Analysis
Results in Multibody System Codes. 2nd Int. Conf. on Computational Structures Technology.
Athens, Greece, 1994, Civil-Comp-Press.
41. Yagi, T., Stensson, A. and Hardell. C.: Simulation and Visualisation of the Dynamic Behaviour of
an Overhead Power System with Contact Breaking. Vehicle System Dynamics, 25: 31-49. 1996.
42. Doomernik, J., Valkenburg, C. and de Vet. B.: Anhebung der Ceschwindigkeit auf dem Gleich-
stromnetz der Nederlandse Spoorwegen. Elektrische Bahnen. 9 1 (4): 1 14- 1 19. 1993.
 PANTOGRAPH/CATENARY DYNAMICS, CONTROL 195

43. Scott, G. A,: High Speed Current Collection - System Sensitivity. Paper presented at the Institution
of Mechanica:l Engineers, Railway Division, High Speed Current Collection Seminar S079, 24th
September 1992.
44. Betts, A. I., Homes, R. and Hall, J. H.: Defining and Measuring the Quality of Current Collection
on Overhead Electrified Railways. I.E.E. Conference 'Electric Railway for a New Century',
September 1987.
45. Brand, W.: Adaptive Regelung eines Stromabnehmers f i r Hochgeschwindigkeitszuge. Diplomar-
beit, lnstitut und Lehrstuhl B f i r Mechanik. Technische Universitat-Miinchen 1995.
46. Belyaev, I. A.: Automatisch gesteuertes Stromabnehmersystem f i r elektrische Triebfahrzeuge.
Schienen der Welt 5, Mai 1974.
47. Beadle, A. R., Betts, A. I. and Smith. W. R.: Pantograph Development for High Speeds. Royal
Engineers Journal (REJ), Nov. 1975, pp 72-81.
48. Coxon, D. J., 0ostling. R. J. and Withland, K. M.: Evolution of a Simple High-Performance
Pantograph. Railway Gazette International, Jan. 1980, pp 44-47.
Downloaded by [TIB & Universitaetsbibliothek] at 03:31 27 February 2014

49. Goroskov, J. I.: Oberleitungs-Ausriistung und Stromabnahme bei Geschwindigkeiten von 160 bis
300 h / h . Schienen der Welt, Dez. 1974 S805-828.
50. Dupuy, J.: 380 km/h. Schienen der Welt (1981), pp. 316-321.
51. Galeotti, G., Cialanti. M., Magrini, S. and Toni P.: Servo actuated railway pantograph for
high-speed running with constant contact fore. lMechE 1993, Vol. 207.
52. Meisinger, R.: Systemuntersuchung und Bereitstellung eines Versuchssystems fur die
Enegieubertragung. BMR-Forschungsbericht TV 7926 1981
53. Meisinger, R.: Simulation von Stromabnehmer und Fahrleitung bei hohen Fahrgeschwindigkeiten.
ZAMM 64 (1,9841, Zeitschrift f i r angewandte Mathematik und Mechanik, S. T69-T7O.
54. Kriiger, W. ancl Eichberger, A,: SlMAX Manual. INTEC GmbH Ingenieurgesellschaft fur neue
Technologien, Munchner Strasse 20, 82234 Wessling, 1997.
55. Breitbach, E. and Biiter, A.: Adaptiver Stromabnehmer: Hubantrieb, Deutsche Pantentanmeldung
Aktenzeichen 195 40 914.0; 8.1 1.95.
56. Breitbach. E. arid Buter. A,: Adaptiver Stromabnehmer: Torsionselement im Anschlussbereich der
Oberschere, Ccutsche Pantentanmeldung Aktenzeichen 195 40 91 4.2; 8.1 1.95.