nd th

Joint 22 International Conference on Electrical Contacts & 50 IEEE Holm Conference on Electrical Contacts, Seattle - USA, 2004

The Process of Arc-Splitting Between Metal Plates

in Low Voltage Arc Chutes

Manfred Lindmayer, Erik Marzahn, Alexandra Mutzke,

Thomas Rüther, Matthias Springstubbe+
Institut für Hochspannungstechnik und
Elektrische Energieanlagen
Technische Universität Braunschweig
Schleinitzstr. 23a, D-38106 Braunschweig, Germany
e-mail: m.lindmayer@tu-bs.de,
Homepage: http://www.htee.tu-bs.de

Abstract — The behavior of arcs in low voltage switching devices plex processes in an arc chute, mainly experimental investiga-
is affected by several interactions. For studying simple arrange- tions have been carried out so far [2-6]. But increasing per-
ments of arc runners without metallic splitter plates (“deion formance of computers enables one to simulate more details of
plates”) a 3D simulation system has already been developed. It the switching arc. For this reason simulations have gained in
takes into account the plasma fluid dynamics, the current flow importance recently [7-16]. Simulations also help to obtain in-
within the electrodes and the plasma, and the magnetic field formation, which is usually not accessible by experimental
generated by both currents. In order to model the arc splitting investigations, and thus can reduce the time for development of
process simulations with an insulating barrier were done in a new switching devices.
first step.

For the simulation of the splitting process between metallic de- II. SIMULATION METHOD, PRINCIPLE
ion-plates the simulation model has to be extended. In experi- An arc consists of the arc roots at anode and cathode and of
ments including high-speed movies it is shown that this process is
a continuous transition of current flow from the still undivided
the arc column between them. The behavior of the arc column
arc to the new arc roots formed on the metal plates. is determined by coupled electromagnetic and gas dynamic
interactions as shown in Fig. 1.
A new simulation model representing the roots by a thin layer of
current-dependent resistivity is discussed and simulation results
are compared with switching experiments.

I. INTRODUCTION
Opening the contact pieces in low voltage switchgears
causes an arc, which has to be extinguished quickly. Several
methods of arc treatment are applied, especially for current
limiting interruption [1]:
• arc motion and elongation
• squeezing between insulating size walls
• cooling by gassing wall materials
• splitting into series arcs by metal plates
Of these possibilities, which are also used in combination, Figure 1. Interaction of processes in arc column
the most effective and hence most widely used principle is to
split the arc up into series of arcs in a stack of splitter plates, The current density of the arc leads to ohmic heating and
generally made of steel, because its ferromagnetic property magnetic forces, which cause gas flow and energy transport
helps to attract and keep the arc in the stack. Thereby the arc within and out of the plasma. This leads to a temperature and
voltage is increased by multiple anode and cathode falls. pressure distribution within the chamber, which affects the
electrical conductivity and all other properties of the plasma,
For the development of switching devices detailed knowl- such as thermal conductivity, viscosity, density and specific
edge about the arc behavior is necessary. Because of the com- heat. The local distribution of electrical conductivity deter-

+
The authors are indebted to Deutsche Forschungsgemeinschaft (German
Research Foundation) for support of this research project.
mines the electric potential and the current distribution, which does not affect current measurement. The arc is ignited be-
closes the circle. tween fixed arc runners by an exploding wire at the place indi-
cated.
All field processes in Fig. 1 can be described by second or-
der partial differential equations, which follow the form of the
so called “general transport equation”
∂ ( ρΦ ) G
+ div ( ρ v Φ ) − div ( Γ grad Φ ) = S . (1)
∂t Φ Φ

The transport takes place in a fluid of density ρ flowing at a

speed v. The dependent variable Φ represents the physical
quantity whose transport is described. This equation consists
on the left-hand side of a time-dependent term (storage term), a
convection term that describes the transport by flow, and a
diffusion term specifying the transport against the gradient of
Φ. ΓΦ is the diffusion coefficient of the quantity Φ. The solu-
tion of (1) yields the local and temporal evolution of the physi- Figure 2. Test switch for the investigation of arc splitting
cal quantity.
Typical oscillograms of the total arc current, the partial cur-
The classical fluid mechanics of compressible fluids are de- rent across the splitter plate, and the arc voltage for two se-
scribed by a system of three coupled transport equations, the lected experimental conditions are shown in Fig 3. The arc
mass balance, the momentum balance and the power balance appearance described below was taken from high-speed movies
per unit volume (“Navier-Stokes equations”). They are referred 16 000 or 32 000 frames/second. The movies were made with a
to as “gas dynamics” in Fig. 1 and characterize the dynamic rotating prism camera, model Hitachi HD16. No filters were
behavior of the arc plasma. The electrical and magnetic proc- used.
esses (Fig. 1) can also be represented by transport equations
(1), where the physical quantity Φ is either the electrical po-
tential U or the magnetic vector potential A. These quantities
are solved in the electrodes as well as in the plasma. The mag-
netic field is of special interest because the arc is moved by the
resulting magnetic Lorentz forces J x B. More details have
already been presented [7-13].
The current in the cathode and anode spots, flowing from
the electrodes into the plasma, needs a special treatment. As
described earlier [7-13] up to now the total current is parti-
tioned to the boundary elements of the regarded electrode and
plasma volume, respectively, with weighting factors depending
on the local electrode temperature. As shown later, the arc root
formation on splitter plates needs an extra treatment.

III. ARC SPLITTING PROCESS

If the arc chamber is provided with deion plates, the proc-
ess of arc splitting has also to be taken into account. The basic
process of arc splitting has already been discussed by Burkhard
in 1962 [17]. In a recent project, experiments were carried out
with newer measuring technique and higher currents up to
several kA [18]. The main aim is to gain knowledge and data in
order to model the process of arc splitting by metal plates.
A. Measurements of Arc Splitting
The experimental setup to investigate the process of arc
splitting is shown in Fig. 2. It consists of two fixed copper arc
runners 1.5 x 4 mm in a 12 mm wide chamber made of
“Dotherm DT700”, an insulating material based on cement,
with vents in the upper side, and a single, specially prepared Figure 3. Oscillograms of total current, current through splitter plate, and
arc voltage
splitter plate of 3 mm total thickness. It is partitioned into two
1 mm thick halves of unalloyed steel, separated by an In Fig 3a, arc-ignition by wire explosion is characterized by
insulating pressboard layer 1 mm thick. These halves are the 50 V peak (t=3.2 ms), after which the voltage stays around
connected by a distributed 0.1 mΩ shunt, evenly arranged be- 40 to 45 V (t=3.2 – 3.7 ms). During this time the arc sticks as a
tween the halves, for the measurement of the current across the broad band in the lower V-shape region of the electrodes. At
splitter plate. The shunt resistance is low enough, so its voltage t=3.7 ms the arc starts to elongate into the upper area, and to
drop does not impair the splitting process. Due to its arrange- bend around the splitter plate, accompanied by a nearly steady
ment the inductance of the shunt is also low enough, so that it
voltage rise. Up to t=4.2 ms the total current flows completely Fig. 2. Instead of one splitter plate there are two plates, each of
below the splitter plate, towards which the arc is moving. A 2 mm thickness. The measurement of current through the plates
steady increase of the fraction through the plate begins at remains out of consideration.
t=4.2 ms, until at t=4.5 ms it has reached one third of the cur-
rent of 1.3 kA. In the meantime the voltage increases up to
105 V peak. At t=4.5 ms there is a rapid back-commutation of
the arc current within about 10 µs into the lower V region ac-
companied by a voltage breakdown to half of the peak value
before. Shortly afterwards the arc moves upward again and the
fraction of current through the splitter plate begins to rise up to
forty percent of the total current (1.6 kA at t=4.95 ms) with
voltage increase to 100 V. At t=5.6 ms, after two more cycles
of back-commutation, the current through the plate reaches
nearly half of the total current of 1.85 kA. The final re-splitting
process begins with a steady rise of plate current and arc volt-
age (t=6 ms). The rising part of plate-current ends the parti-
tioning-process by reaching the total current on its decreasing
slope (t=9.25 ms). It remains there until current zero at
t≈ 11.4 ms.
The film (not shown here) clearly reflects the stages re-
ported here. The experiment shows that the partial current flow
through the plate is based on the fact of parallel existence of
arc current through and below the splitter plate. This is quite
normal for some time. Before the start of current through the
plate, the plasma is bent around the plate, to generate the volt-
age drop necessary for spot formation. It can be seen in the film
that the missing current portion after spot formation is due to a
plasma cloud which is still visible in the whole chamber out-
side the plate, and which disappears only towards the end of
the half cycle. During the parallel currents both parts are visi-
ble, and after t ≈ 9.25 ms the plasma channel below the plate
has disappeared. (There are some minor errors in the plate
current signals of Fig. 3a due to the differential measurement of
the shunt voltage on high common potential. In Fig. 3b this
was avoided by optical fiber transmission.).
Fig. 4 is a choice of high-speed movie frames belonging to
the oscillogram of Fig. 3b. Under these conditions of relatively
low current (1 kA RMS) there are no back-commutations after
the first splitting. Also a voltage hump at the beginning of the
splitting process can often be seen. It is less pronounced or not
found at all with higher currents.
The splitter plate, though thicker, is only indicated by a thin
line. The wire explosion, associated with a first voltage peak,
ends shortly before t =1 ms. Subsequently the arc moves up-
ward and elongates. At t =1.35 ms it begins to bend around the Figure 4. High-speed movie from experiment of. Fig. 3b, 1 kA RMS
metal plate, while the whole current still flows outside of it.
Only after t=1.5 ms does the current through the plate begin to
rise. Between t=1.59 ms and t=1.72 ms, during the fast initial
current rise, arc roots are being formed newly on both sides of
the plate, and they become brighter. At this time the voltage
has reached its maximum. The roots have not been formed at
the lower edge, as one might have expected, but in the center of
the plate. The reason is that the current loop formed by the
plasma, which generates the voltage drop necessary for the
formation of arc roots, must have a certain length. Both current
paths – through the splitter plate and around its lower edge -
are clearly visible. After t=1.77 ms the latter is only faintly
visible, but the oscillogram shows that until t=7 ms a minor
current portion still flows through residual plasma outside the
plate.
An oscillogram of a measurement with two splitter plates is
shown in Fig 5. The experimental setup is almost analog to Figure 5. Oscillogram of total current and arc voltage
 Arc-ignition (t=0.6 ms) and elongation into the upper V- tioned through the plates and undivided outside the plates - is
shaped area (t=0.6 – 1.6 ms) is similar to the oscillograms quite normal. Moreover, repeated fluctuations may occur be-
shown in Fig. 3. At t=2.3 ms a back-commutation is indicated tween both paths. In this experiment the processes of upward
by a voltage breakdown to 60 V, followed by repeated proc- motion and back-commutations can be attributed partly to the
esses of partial forward- and back-commutations with a high elongation on V-shaped arc runners, partly to the splitting
repetition frequency (t=2.3 – 5.9 ms). During this time, the arc process. When two or more deion plates in series are involved,
often appears in a distinct W-shape, caused by squeezing the their role becomes more important. Everyone dealing with low-
arc between the splitter plates. Between t=6.0 ms and t=6.7 ms voltage switchgear tests is familiar with such processes, which
the voltage reaches values of 150 V down to 135 V. This pla- are undesirable, but often unavoidable.
teau is followed by another sequence of forward- and back-
commutations until current zero at t=8.5ms. The reasons for back-commutations have been discussed
earlier [19]. The elongation along arc runners, and the splitting
process on metal plates both cause a voltage rise. This forces
more current to flow in the reverse, still hot and conducting
area behind (=below) the arc. This region is re-heated, and still
more current flows there, until finally the situation tilts, and the
current quickly commutates into this area. It has been shown in
simulations that, while the heating-up process may take up to a
millisecond without visible current change, the final commuta-
tion of the main current may occur within typically 5 to 10 µs.

B. Model of Arc Splitting Process

The process of arc splitting, as a result of the experiments
and from the considerations in [17], is schematically shown in
Fig. 7.

Figure 7. Principle of arc splitting by metal plates

Figure 6. High-speed movie from experiment of. Fig. 5, 2 kA RMS
During the subdivision of the arc on such a plate, an addi-
Fig. 6 displays a section of high-speed movie frames be- tional cathode and anode fall, respectively, has to be created on
longing to the oscillogram of Fig. 5. The splitter plates are either side of the plate. Their voltage requirement lies at 15 to
indicated as thin lines again. The chosen time-frame clarifies 20 V for both together. Before the formation of these new
the processes between t=5.882 ms and t=6.707ms. At spots, the voltage drop in the yet undivided plasma between
t=5,882 ms the arc starts moving upward and bends around the both sides of the plate must at least reach this value. This is
splitter plates, accompanied by a fast increase of voltage. The achieved by bending the plasma around and squeezing it
distinct W-shape at t=5.983 ms is broken during the next two against the front edge of the plate, thus elongating it, reducing
frames by forming arc-roots at the left splitter plate. They are its local cross-section and consequently increasing its voltage
formed at the lower third of the plate (the reason has already drop (a, b). For a certain time, current flows both in the new
been discussed before). The voltage has reached its local path across the new anode and cathode falls through the metal,
maximum of 150 V. As the partial arc at the left third begins to and in the old path in front of the plate (b, c).
move upward, the rest of the arc-column sticks at the lower
edge of the right splitter plate. During t=6.117 ms and Because of its complexity, the arc splitting process has not
t=6.484 ms there are only insignificant changes in arc-appear- been included in simulations so far. In [15] simple metal plates
ance. Especially at the right plate, the part of current flowing with their low resistance have been used. The disadvantage of
outside appears to be predominant, while at the left plate the this model is that the current flows completely through the
major part of current flows through it. The voltage decreases metal plates as soon as the arc reaches them because the path
down to 135 V. For a short period, at t=6.584 ms, there is a outside is shorted. The delay in splitting-up and back-commu-
noticeable current flowing outside the left plate. This situation tations cannot be taken into account. The application of inac-
is followed by a formation of three arcs in series. However, the tive insulating barriers instead of metal plates [16] also cannot
splitting at the right plate is not completed, but a minor part of describe the role of deion sheets completely.
current is still flowing outside. At t=6.707 ms the temporary
A good simulation tool should be able to model the process
arc-splitting ends with a re-ignition in the V-shaped part of the
of arc splitting (i.e. formation of new arc roots), the parallel
electrodes.
existence of current paths through and outside the splitter
The examples given here clearly show that the process of plates, and the back-commutation out of the splitter plates as
arc splitting between metal plates is by no means steady and well. The arc root model used so far by the authors and de-
straightforward. The existence of parallel current paths – parti- scribed shortly above, is not suitable for this purpose, because
the total current ought to be known and sub-divided according The electrical part (potential, current flow) of the simula-
to weighting functions. For the electrodes (“arc runners”), the tions of the arc splitting process were carried out by the Finite
total current equals the current from the external circuit. How- Element package ANSYS, coupled with the gas dynamics in
ever, the current through the splitter plates is not known, but it CFX which is able to solve systems of coupled transport equa-
is a consequence of all processes in the arc roots and in the tions [23]. Plasma properties were taken from [24].
plasma around the plates. Consequently the model must be
self-consistent and yield the portions of current flow outside IV. SIMULATION RESULTS
and through each splitter plate without any special assump-
tions. Because of the microscopic scale of the electrode fall As a preliminary step to model the arc behavior around de-
regions at anode and cathode it makes no sense to describe the ion plates, an insulating barrier of different length (20 and
physical processes in detail. Instead, a macroscopic and phe- 15 mm) between two parallel copper arc runners (Fig. 9) was
nomenological description is chosen, which is not in contra- used instead of a metal plate [25]. For the insulating barrier the
diction to the physical behavior. Though the physical processes thermal data of copper are used as well, but the electrical con-
are quite different in detail, anode and cathode are treated ductivity is set to zero. The sidewalls are modeled by stationary
equally, though their data could also be differentiated. This heat conduction boundary conditions with a temperature limi-
simplified treatment is justified, because the macroscopic di- tation to 2000 K [9].
mensions of anode and cathode roots were found similar for
kA-arcs. Rails 4.5x1.5, 58 long
Rail distance inside 8
The voltage drop across both the cathode and anode spots Side walls 4.5 apart, adjacent to rails
are known to be rather independent of the current, roughly Barrier 1.6 thick, 20 (15) long
U0 =10 V [20-22]. The arc spots are located in a narrow space Venting holes in upper wall
sheath in front of the metal plates. They can be modeled by a
25% of cross-section
shell of elements with a nonlinear local electrical conductivity,
which is chosen according to the current density – voltage
characteristics given in principle in Fig. 8.

All dimensions in mm
Figure 9. Rail electrodes with insulating barrier

Fig. 10 depicts the simulation results for a 20 mm long bar-

rier with a current of 5 kA RMS and a current at ignition of
450 A.

Figure 8. Principle of voltage – current characteristics of a shell for

modeling the formation of an arc spot

For currents above a certain threshold the voltage ap-

proaches asymptotically U0=10 V. To enable a continuous
current transition between the arc through and the plasma in Figure 10. Simulated arc motion with 20 mm barrier, I=5 kA RMS,
front of the spot, the voltage approaches zero for low current current at ignition 450 A
density. This means that when the voltage difference between
splitter plate and adjacent plasma is small, only a small current The arc is ignited asymmetrically in the lower area and then
flows through the plate – provided the current can continue on moves up. At t=0.2 ms it starts to bend around the barrier. In
the other side. When it approaches U0, the value necessary for the following the arc roots move further up and form a current
spot formation, a nearly unlimited current can flow through the path in a distinct U-shape. Since the arc is elongated and
plate, but also a residual current outside due to this voltage squeezed in front of the barrier the arc voltage increases.
difference and the electrical plasma conductivity in the outside The results of a simulation with an only 15 mm long barrier
area. An additional voltage hump in the characteristics (dashed are given in Fig. 11. Until t=0.675 ms it resembles the simula-
line) could also be taken into account, in order to model the tion with the long barrier. However, at t=0.7 ms plasma jets
hump observed in experiments under certain conditions (sec- blow out of the venting holes on both sides of the barrier and
tion III A). This peak value could be assumed temperature- touch each other. They close a shorter current path above the
dependent in order to model the formation of arc spots out of barrier which exists parallel to the one under the barrier until
the contact between hot plasma and cold electrodes as well. t=0.725 ms. At t=0.75 ms only the upper path burning through
There the voltage necessary lies much higher, e.g. 200 to the holes is visible.
300 V.
 with the experiments cannot be expected yet, but the general
qualitative behavior of the arc splitting process can be mod-
eled.
Fig. 13 shows a sequence of local current density distribu-
tions and Fig. 14 the respective current through the splitter
plate and arc voltage vs. time. The arc is ignited at the lower
edge of the chamber and starts to move up. The air is warmed
up by ohmic heating, hence the resistivity and arc voltage,
respectively, decrease strongly at the beginning. After
t=0.24 ms the arc reaches the splitter plate and begins to bend
around it. Because of arc elongation and squeezing under the
plate the voltage drop increases, until at t=0.48 ms the first arc
roots on the plate are visible. According to the experiments the
arc roots do not form at the lower edge of the metal plate, but at
about one third of the height of the plate. The current through
the splitter plate starts to rise and the voltage drop decreases for
a short time. A similar behavior could be observed in the ex-
periments. Until t=0.66 ms two parallel current paths, one
through the plasma and one through the splitter plate, exist, but
the current through the plate and the voltage continuously in-
crease. At t=0.72 ms the current path through the plasma has
vanished, nearly the whole current (I=95 A) flows through the
plate. When the current approaches the upper edge of the plate
(t=0.81 ms), a new current path forms in the exhaust volume.

Figure 11. Simulated arc motion with 15 mm barrier, I=5 kA RMS,

current at ignition 450 A

First experience with the new model of arc root formation

on splitter plates was gained in simulating in the small arc
chute with a single metal plate shown in Fig. 12.

Rails 4.5x1.5, 16 long

Rail distance inside 8
Side walls directly adjacent to rails
Copper Plate 2 thick, 7 long
Upper side open (exhaust volume)

All dimensions in mm

Figure 12. Arrangement for the simulation of arc splitting

The plasma parameters were made dependent on the local

temperature and pressure according to published data for air
[24]. For the rails and the splitter plate the data for copper were
taken. The arc root area of the plate was represented by a shell
of 0.1 mm thickness completely surrounding the metal plate.
The electrical resistivity in this area was calculated as a func-
tion of the local current density according to the J-U-curve in
Fig. 8 (solid line). The current density and the resistivity distri-
bution have to be iterated within each time step. The arc cur- Figure 13. Simulated process of arc splitting on a metal plate between
rent of 100 A DC was chosen smaller than in the experiments two straight arc runners (Fig. 12), I=100 A DC
and lies in a region where complete splitting is easier. The
magnetic self-field of the arc was neglected. The arc is driven
by a constant blowout field which was assumed as 0.01 T.
Because of these simplifications a close quantitative correlation
 [8] F. Karetta, M. Lindmayer: Simulation of arc motion between divergent
arc runners. 19th Int. Conf. on Electric Contact Phenomena, Nürnberg,
Germany, 1998, pp. 361-367.
[9] F. Karetta: Dreidimensionale Simulation wandernder Schaltlichtbögen.
Thesis TU Braunschweig, Fortschritt-Bericht VDI Reihe 21, No. 250,
VDI-Verlag, Düsseldorf, 1998.
[10] M. Lindmayer, M. Springstubbe: 3D-Simulation of arc motion including
the influence of ferromagnetic material. 47th IEEE Holm Conference on
Electrical Contacts, Montreal 2001.
[11] M. Lindmayer, M. Springstubbe, E. Marzahn: Neues über die Komplett-
simulation des Lichtbogenverhaltens in Schaltgeräten. 16. VDE Fach-
tagung „Kontaktverhalten und Schalten“, Karlsruhe 2001. VDE Fachbe-
richt 57, VDE-Verlag Berlin 2001.
[12] M. Lindmayer: Complete Simulation of Moving Arcs in Low-Voltage
Switchgear. XIV Int. Conf. on Gas Discharges and their Applications,
Liverpool 2002.
Figure 14. Current through splitter plate and arc voltage vs. time
[13] M. Lindmayer, E. Marzahn, A. Mutzke, M. Springstubbe: Low-Voltage
Switching Arcs – Experiments and Modeling. XVth Symposium on
V. SUMMARY AND CONCLUSION Physics of Switching Arc, Brno, 2003.
[14] G. Andre, K. Berger, B. Gessl, W. Rieder, W. Schneider: Lichtbogenbe-
The system developed for the 3D simulation of the com- wegung bei geknickter Laufschiene. 17. VDE Fachtagung
plete arc behavior in low-voltage arc chutes has been summa- „Kontaktverhalten und Schalten“, Karlsruhe 2003. VDE Fachbericht 59,
rized. As a first step to model the process of arc splitting simu- VDE-Verlag Berlin 2003.
lation examples with insulating barriers are given. Special [15] H. Stammberger, Th. Daube, C. Dehning, M. Anheuser: Arc Simulations
experiments to study the process of arc splitting have been in realistic low-voltage arcing chambers. 21st Int. Conf. on Electrical
made. Based on their results, a new model of the arc root for- Contacts, Zürich 2002
mation on splitter plates has been suggested. It replaces the arc [16] B. Swierczynski, J.J. Gonzales, P. Freton, A. Gleizes: Contribution 3D à
root area by a thin layer of elements with a nonlinear electrical l’étude d’un disjoncteur basse tension. 6émes Journees d’Études sur les
conductivity depending on the local current density. First sim- Fluctuations des Arcs, LAEPT, Universté Blaise Pascal, Clermont-
plified simulations and comparison with measurements show Ferrand 2003.
that the process of arc splitting by metal plates can be modeled [17] G. Burkhard: Über das Lichtbogenverhalten in Löschblechkammern und
in principle by the suggested method. deren Bemessung. Thesis TH Ilmenau 1962.
[18] E. Marzahn: Thesis TU Braunschweig, to be published.
Further work is necessary to incorporate this phenomenol-
ogical arc root model into the complete simulation process, and [19] M. Lindmayer: Simulation of stationary current-voltage characteristics
and of back-commutation in rectangular arc channels. 17th Int. Conf. on
to model complete arc chutes with stacks of metal plates. This Electrical Contacts, Nagoya 1994.
arc root model, modified by a temperature and field depend-
[20] A. Erk, M. Schmelzle: Grundlagen der Schaltgerätetechnik. Springer-
ence, might also prove to be an acceptable model for the arc Verlag, Berlin 1974.
roots on the contacts and rail electrodes. So any differentiation
[21] G. Hertz, R. Rompe: Einführung in die Plasmaphysik und ihre
in the treatment of different kinds of metallic parts in the arc technische Anwendung. Akademie-Verlag, Berlin 1965.
chute could be omitted. This would also enable to model the
[22] M. Lindmayer: Schaltgeräte. Springer-Verlag, Berlin 1987.
process of commutation from one electrode to another, which
is often utilized in low-voltage switching devices. [23] CFX User Manual. CFX International, 8.19 Harwell, Didcot,
Oxfordshire OX11 ORA, UK 1997.
[24] J. Yos: Revised transport properties for high temperature air and its
REFERENCES components. Avco Space Systems Division, Technical Release, 1967.
[1] M. Lindmayer: Medium to high current switching. In: P. Slade (editor): [25] M. Springstubbe: Thesis TU Braunschweig, to be published.
Electrical Contacts, Marcel Dekker, New York 1999.
[2] N. Behrens: Lichtbogenwanderung in Leitungsschutzschaltern. Thesis
TU Braunschweig, 1980.
[3] H. Weichert: Lichtbogenwanderungs- und Kommutierungsvorgänge in
strombegrenzenden Löschkammeranordnungen. Thesis TU Braun-
schweig, 1988.
[4] G. Clemens: Untersuchungen an strombegrenzenden Löschkammern.
Thesis TU Braunschweig, 1989.
[5] P. M. Weaver and J. W. McBride: Arc motion in current limiting circuit
breakers. 16th International Conference on Electrical Contacts, pp.
285-288. Loughborough, UK, 1992
[6] E. Gauster and W. Rieder: Arc lengthening between divergent runners:
Influence of geometry and materials of runners and walls. 17th
International Conference on Switching Arc Phenomena, pp. 285-288.
Lodz, Poland, 1993.
[7] F. Karetta, M. Lindmayer: Simulation of the gasdynamic and
electromagnetic processes in low voltage switching arcs. IEEE Trans.
CPMT-21 Part A (1998), pp. 96-103.