Electric Arc Modeling in Circuit Breakers for Electromagnetic

Transients Analysis
Kevin Pedro
[email protected]

Instituto Superior Técnico, Lisbon, Portugal

April 2017

Abstract
Circuit breakers play a very important role in Eletrical Power Systems. They are responsible for
switching operations, connecting and disconnecting network elements, as well as interrupting currents
harmful to the system. However, the circuit breaker doesn’t respond instantaneously and exhibits also
a non-linear behavior, resulting from the appearance of an electric arc in the dielectric. For this reason
the modeling of the electric arc takes great importance in its representation. To approach this subject,
the Black Box models are implemented for two distinct faults: terminal fault and short-line fault,
in order to study their arc parameters, as well as their ability to represent the characteristics of the
electric arc.
Keywords: Circuit breaker, Eletric Arc, Black Box Models, Terminal Fault, Short-line Fault,
Interruption, Re-ignition

1. Introduction 2. The Switching Arc

The switching arc present in the circuit breaker,
Circuit breakers are the equipment responsible for during a switching operation, assumes a funda-
switching operations in the power system and very mental role in the process. The electric arc is a
important in the interruption of harmful currents to plasma channel that is established between the cir-
the system. Normally it is considered, for simplic- cuit breaker contacts after a discharge in the extin-
ity of representation, that the change of states of guishing gas. When a current flows through a cir-
the circuit-breaker takes place instantaneously. In cuit breaker and the contacts begin to move away,
reality, the circuit-breaker does not respond instan- the magnetic energy stored in the inductances of
taneously and also presents a non-linear behavior, the power system prevents the current from being
during the switching operations, resulting from the interrupted, forcing it to continue [16]. Immediately
appearance of an electric arc on it’s dieletric. For before contacts separation, the contact area is very
that reason, is fundamental to understand the influ- low and the high current density melts the circuit
ence of the electric arc in the interruption process. breaker contacts. The contacts material that has
melted ”explodes” and causes a discharge in extin-
The literature of the specialty presents several guishing medium [16]. As the kinetic energy of the
models to represent the behavior of the electric arc molecules increases, matter changes from the solid
in the circuit-breaker. The physical models are the to the liquid state and from the latter to the gaseous
most complex ones, and are based on the fluid dy- state. A further increase in temperature, equivalent
namics equations to translate the behavior of the to the increase of energy state, cause molecules dis-
arc as a plasma. The simplest ones, Black Box sociation into separate atoms. If the level of energy
models (also known by P-τ models) translate the continues to increase, orbital electrons of the atoms
nonlinear variation of arc conductance as a function dissociate into free electrons, leaving positive ions.
of the arc voltage and current. While the first type This state of matter is called plasma. Because of
of models is used by manufacturers in the develop- the free electrons and the heavier positive ions in the
ment of equipment, the second is used in studies of high-temperature plasma channel, the plasma chan-
electromagnetic transients since they translate the nel is highly conducting and the current continues
interaction between the electric arc and the circuit to flow after contact separation. The plasma con-
under analysis. ductivity is function of temperature and increases
rapidly with its magnitude [16].

1
 ous interaction of various phenomena (electrical and
magnetic).
In the electric arc extinction process two physical
regimes (requirements) are involved [3]:
• Thermal regime: The arc channel has to be
cooled down to a sufficiently low temperature
in which it ceases to be electrically conducting.
• Dielectric regime: After the arc extinction,
the insulating medium must withstand the
rapidly increasing voltage that appears across
circuit breaker terminals. This voltage is called
the recovery voltage and its transient compo-
nent, transient recovery voltage (TRV) which
is motivated by the continuity of electric and
magnetic energy.
If any of the requirements is not achieved, the
current will continue to flow until, after half a cycle,
it reaches zero again. The verification of the neces-
Figure 1: Electrical arcing in a gaseous medium [15] sary conditions for current extinction is repeated in
each passage for zero.
The thermal ionization, as a result of the high In figure 2 a) is presented the equivalent circuit
temperatures in the electric arc, is caused by en- for the terminal fault. In figure 2 b) is represented
ergy transfer between electrons and photons that the short-circuit current and the recovery voltage,
have high kinetic energy, and positive ions and neu- for the fault in question, where is shown the thermal
tral atoms with reduced kinetic energy (see Figure and dielectric regime [3].
2.1). At the same time, there is also a recombina- The instant of time t1 represents the moment of
tion process when electrons and positively charged contact separation, t2 the moment of arc extinction
ions recombine to a neutral atom. When the ion- and S the rate of rise of recovery voltage.
ization rate is equal to the recombination rate, the
thermal equilibrium is reached. Normally, it is as- 2.1.1 Thermal Regime
sumed that the electric arc is in thermal equilib- After the opening of the circuit breaker, an elec-
rium during the entire transient phenomenon, since tric arc arises sustained by the dissipated energy
the time constants of the ionization and recombi- under joule effect which subsists while the temper-
nation processes are negligible when compared with ature of the medium is very high [15]. The electric
arc variations [16]. arc interruption is carried out when current mag-
The electric arc consists of three regions: nitude is near zero in case that the ionized channel
rapidly recovers its dielectric characteristics. As the
• Column region current approaches zero, arc conductance decreases
with it. When the current theoretically reaches the
• Cathode region
zero magnitude, the arc conductance decreases as
• Anode region a function of the deionization time constant of the
medium. This time constant is inherent to the iner-
In the arc column, the current flow is maintained tia of the medium in recovering its dielectric char-
by electrons, and there is a balance between the acteristics. In this instant of time, the conductance
electron charges and the positive ion charges. The assumes a small value diferent than zero, which, de-
peak temperature in the arc column can range from pending on the applied arc voltage, can produce a
7000 to 25 000 K, depending on the arcing medium current of some amperes called post-arc current (see
and the configuration of the arcing chamber [16]. figure 3). The current interruption depends on an
The voltage established between the circuit breaker energy balance. If the arc input power exceeds the
terminals due to the arc and contacts resistance is circuit breaker cooling power the medium recovers a
called arc voltage [15]. conductive state, caused by thermal failure, and the
current continues to flow. Otherwise, a successful
2.1. Arc Extinction Regimes thermal interruption is obtained [15].
The current interruption process in a circuit breaker Figure 3 illustrates the inertia in the electrical
assumes a complexity arising from the simultane- conductivity of the arc. The thermal interruption

2
 after current zero [3, 10].

2.1.2 Dielectric Regime

The dielectric regime begins with the extinction of

the post-arc current and the manifestation of the
transient recovery voltage (TRV) that appears on
the circuit breaker terminals. At this time, the
extinguishing/isolating medium is no longer elec-
trically conducting, but it still has a much higher
temperature than the ambient. This reduces the
voltage withstand capacity of the contact gap. The
dielectric interruption success will depend on the
rate of recovery of the dielectric strength and the
rate of rise of the TRV (see figure 4).

(a) Simplified equivalent circuit for terminal fault

Figure 4: Dielectric regime [3]

As we can see at figure 4, whether the recov-

ery rate of dielectric strength (contact disruption
(b) Curves of short-circuit current isc and recovery voltage
ur voltage - dashed curve) is higher than TRV rising
rate (TRV voltage - continuous curve) a success-
Figure 2: Arc extinction regimes - terminal fault [3] ful dielectric interruption is obtained. Otherwise,
an unsuccessful interruption results and the current
will start over by dielectric failure [3]. The resump-
tion of current between the contacts of a mechanical
switching device during a breaking operation with
an interval of zero current of less than a quarter
cycle of power frequency is entitled re-ignition [1].
Otherwise it is called restrike [2].

3. Electric Arc Models

Existing arc models can be classified into two cate-
gories [16, 6]:

• Physical models

• Parameter Models
Figure 3: Current shape at interruption (the time
scale is in the microsecond range) [3]
In parameter models another category called
Black Box Models is defined. The black box models
regime corresponds to the period of time starting are parameter models that use functions and sim-
some microseconds before current zero, until extin- pler tables to determine the parameters of the mod-
guishing of the post arc current, a few microseconds els. [16, 6].

3
3.1. Physical Models  
Engineers responsible for designing circuit breakers 1 dg 1 ui
= −1 (2)
work largely with physical arc models when they g dt τ P
plan to create a new prototype. Physical mod-
where
els are the most complex ones, since electric arc
description are based on fluid dynamics equations g = the arc conductance [S]
and laws of thermodynamics, in combination with
Maxwell’s equations [16, 6]. These models are based u = the arc voltage [V ]
on the conservation equations of mass, momentum i = the arc current [A]
and energy. Obtaining the solution through these τ = the arc time constant [s]
equations requires simplifications in order to reduce P = the cooling power [W ]
complexity.
3.2. Black Box Models Due to the assumed considerations, this model is
Black box models (also called P - τ models) describe suitable for the thermal regime (low current inten-
the electric arc through differential equations. They sities) [6].
are considered mathematical models based on phys- The Habedank’s arc model is the junction of the
ical considerations that establishes the relationship two previous models, and is defined by the following
between arc conductance and arc quantities like differential equations [9]:
voltage and current [6, 7].
Although these models are not suitable for de-
1 u2
 
1 dg
signing circuit breakers, they are very important = −1 (3)
to simulate the arc-circuit interaction, where they gc dt τc Uc2
 
present a very reasonable level of precision [6]. For 1 dg 1 ui
= −1 (4)
this simulation purpose is fundamental the behav- gm dt τm P0
ior of the electric arc quantities, rather than the 1 1 1
inherent physical processes characterization. The = + (5)
g gc gm
formulation of these models comes from physical
simplifications, and it is therefore important to re- where
alize their applicability limit [6, 7].
Cassie and Mayr’s arc models are known as the gc = the arc conductance described by Cassie’s model [S]
classic black box models. The other models degen- gm = the arc conductance described by Mayr’s model [S]
erate from the classic models. Cassie assumed that g = the total arc conductance [S]
the arc temperature, current density, and electric
field are constant. Thermal convection is the main The arc conductance g results from the sum of
energy removal phenomenon and the arc cross- the inverse conductance of both models, gc and gm .
section varies with current and time [6, 4, 5, 12]. Habedank’s arc model resulted from the aggrega-
Taking into account these premises, result the tion of two complementary models, allowing a more
Cassie’s arc model equation [5]: rigorous representation of the various currents in-
tensities .
1 u2
 
1 dg
= − 1 (1) Schavemaker’s arc model is a modified Mayr’s arc
g dt τ Uc2
model with a time constant τ and a cooling power
where as a function of the electrical power input.
g = the arc conductance [S] The Schavemaker’s arc model equation is known
by [13]:
u = the arc voltage [V ]
τ = the arc time constant [s]  
Uc = the constant arc voltage [V ] 1 dg 1 ui
= −1 (6)
g dt τ max (Ua |i| , P0 + P1 ui)
This model is particularly suitable for studies in-
volving high intensity currents due to the premises where P1 and P0 are both cooling constants.
assumed [6]. Schwarz’s arc model is also a modified Mayr’s
Unlike Cassie, Mayr’s model assumed that the arc arc model which the time constant and the cooling
cross section is constant and loses energy exclusively power are dependent on the arc conductance. The
by thermal conduction. The arc conductance is a Schwarz’s arc model equation is known by [6, 4, 11]:
function of the internal energy of the arc [6, 4, 12].
 
Given these premises, result the Mayr’s arc model 1 dg 1 ui
equation [6, 4, 12]: = a −1 (7)
g dt τg P gb

4
 where a is the parameter that influences the con- In order to developed the Matlab program were
dutance dependency of τ and b the parameter that considered equivalent circuits that caracterize the
influences the condutance dependency of P . terminal fault (see figure 5) and short-line fault (see
Finally, KEMA’s arc model consists of three mod- figure 6).
ified Mayr’s arc models described by equations (8)
to (13) [14].

dg1 1 λ1 2 1
= g1 ua1 − g1 (8)
dt τ1 Π1 τ1
dg2 1 λ2 2 1
= g u − g2 (9)
dt τ2 Π2 2 a2 τ2
dg3 1 λ3 2 1
= g3 ua3 − g3 (10)
dt τ3 Π3 τ3
1 1 1 1 Figure 5: Equivalent circuit - Terminal fault
= + + (11)
g(t) g1 (t) g2 (t) g3 (t)
u = ua1 + ua2 + ua3 ; (12)

where

τ1 τ2 Π2
τ2 = τ3 = Π3 = (13)
k1 k2 k3

Figure 6: Equivalent circuit - Short-line fault

g = the total arc conductance [S]
un = the voltage across the n-th arc [V ] The circuit breaker is represented in the circuits
of the figures 5 and 6 by the arc models presented
u = the total voltage across the arc [V ]
above.
τn = the time constant of the n-th arc [s] The implementation of the referred circuits and
Πn = the (cooling) constant of the arc models in a Matlab program requires the use of
numerical methods.
the n-th arc A(λ−1) V (3−λ)
The elements of the circuits(R, L and C) were
kn = the free parameters replaced by Norton equivalents.
λn = Cassie-Mayr control of the n-th arc The ordinary differential equations (ODE’s),
which govern the electric arc behavior, were im-
If λ = 1, results in a Cassie’s arc model equa- plemented in three numerical methods, the Euler
tion. If λ = 2, results in a Mayr’s arc model equa- method, the Trapezoidal method, and the 2nd or-
tion. The kn parameters depend on the breaker der Runge-Kutta method.
type while τ1 , Π1 , Π2 depends on the current state In order to validate the correct implementation of
of the circuit breaker [8, 11]. the models, it is possible to compare them with the
Although it is a model deduced from physi- Arc Model Blockset developed by the University of
cal considerations, it is considered a mathematical Delft in the year 2001 [11]. It is an extension of
model, where the physical description is no longer Matlab Simulink/Power System Blockset that can
perceptible. In this way, its application is mostly be used to make arc-circuit interaction studies dur-
practical and requires experimental tests. ing the breaker interrupting process. It allows to
simulate through Matlab Simulink certain electric
4. Numerical Implementation of Arc Models
arc models that are a priori defined (see figure 7).
In this work, we proceed with the implementation, The purpose is to compare the simulations per-
in Matlab, of the different arc models previously formed in Simulink with those of the implemented
presented for the purpose of studying their char- Matlab program.
acteristics. With this objective it was decided to In order to choose the numerical method to be
implement the arc models for two distinct cases of used in the results, was simulated the comparison
fault: of the three integration methods with the Arc Block
Set1 solution (see figure 8).
1. Terminal fault
1 Simulated with a trapezoidal method which has built-in

2. Short-line fault regressive differentiation formula (2-state method).

5
 4
x 10

2.5 Arc Blockset

Script
2

1.5

0.5

uarc [V]
0

−0.5

−1

−1.5

−2

−2.5

0 0.005 0.01 0.015

Time [s]

(a) Arc voltage

Figure 9: Arc voltage - Cassie’s arc model

Figure 7: Arc Model Blockset [11]

4
x 10 As can be seen from figure 9, the results of the
Euler
1
Trapezoidal Arc Blockset and Matlab script are identical, and
Runge Kutta
Arc Blockset therefore, Cassie’s arc model was correctly imple-
0.5
mented for the terminal fault.
0
5. Results
uarc [V]

In this chapter we proceed with the analysis of the

−0.5
Schavemaker’s arc model simulations for the termi-
−1
nal fault as well as the Mayr’s arc model simulations
for short-line fault. Simulations were performed for
−1.5 variations of the different models parameters, being
5 5.05 5.1
Time [s]
5.15 5.2 5.25
−3
verified the consequences in the electric arc behav-
x 10
ior.
(a) Mayr’s arc voltage
5.1. Terminal Fault
Figure 10, is obtained for τ = 100 µs and τ = 1 µs,
9500 Euler
Trapezoidal
allow us to conclude that the decrease in arc time
9000
Runge Kutta
Arc Blockset
constant, associated with a decrease in the deion-
ization time of the medium, allows to change the
8500
re-ignition paradigm to interruption.
The post-arc current occurs because the medium
uarc [V]

8000

7500
does not acquire its dielectric properties instanta-
neously (see figure 11 b)). The transient recov-
7000
ery voltage (TRV) is the voltage across the cir-
6500 cuit breaker terminals and results from the energy
stored in the coils and capacitors at the time of cur-
5.0185 5.019 5.0195 5.02 5.0205 5.021 5.0215 5.022 5.0225
Time [s] x 10
rent interruption (see figure 10 b)).
−3

Figure 12 a) results from a simulation for an in-

(b) Detail of arc voltage for t  [5.018, 5.0225] ms
crement of 3 kV in the arc voltage constant pa-
rameter. Arc voltage tends to the predefined level
Figure 8: Comparison of numerical methods for an (U = 5 kV ) and there is an accentuation of the
a
interruption of Mayr’s arc model voltage rise at the second zero of the current.
In relation to the cooling power P0 , two simula-
tions were performed in the figure 12 b). The in-
It can be concluded from the graph of figure 8 b) crease in the cooling power favors a greater medium
that the trapezoidal method has the closest result deionization in the vicinity of the zero of current. If
to the Arc Blockset method and, therefore, was the this effect does not occur in order to interrupt the
method chosen to perform the simulations. current, a high voltage arises, resulting from the
Figure 9 illustrates one of the simulations car- increase of the medium dielectric strength (greater
ried out with the purpose of validating Cassie’s arc resistance on the circuit breaker terminals).
model implementation for the terminal fault. Finally, the dimensionless parameter P1 proves

6
 5 4
x 10 x 10
2.5 1 2500

2 0.8 2000

1.5 0.6 1500

1 0.4 1000

0.5 0.2 500

iarc [A]
uarc [V]

uarc [V]
0 0 0

−0.5 −0.2 −500

−1 −0.4 −1000

−1.5 −0.6 −1500

τ = 100 µs
τ = 1 µs
−2 −0.8 −2000 τ = 1 µs
τ = 100 µs

−2.5 −1 −2500
0 0.005 0.01 0.015 0.02 0.025 0 0.005 0.01 0.015 0.02 0.025
Time [s] Time [s]

(a) Arc voltage (a) Arc current

4000 200 0.08

τ = 1 µs
2000 τ = 100 µs 100
0.06
0 0
0.04
−2000 −100

−4000 −200 0.02

iarc [A]
uarc [V]

uarc [V]

−6000 −300 0

−8000 −400
−0.02
−10000 −500
−0.04
−12000 −600
τ = 100 µs
−0.06 τ = 1 µs
−14000 −700

−16000 −800 5.032 5.034 5.036 5.038 5.04 5.042

5 5.05 5.1 5.15 5.2 5.25 Time [s] −3
Time [s] −3 x 10
x 10

(b) Post-arc current

(b) Transient recovery voltage - details of a) for
t  [4.95, 5.3] ms
Figure 11: Arc voltage and current considering
Figure 10: Arc voltage and current considering P0 = 1 kW , P1 = 0.8 and Ua = 2 kV
P0 = 1 kW , P1 = 0.8 and Ua = 2 kV
decreases more rapidly, at zero current, with the
to be very sensitive, which is why an increment of decrease of τ (higher deionization of the medium).
only 0.1 was made (see figure 13). Through figure Figure 15 provides the arc voltage considering a
13, it is possible to observe an accentuation of the variation of the cooling power, P = 30.9 kW and
voltage in the current passage by zero, as verified P = 100 kW , with τ = 0.3 µs. As we can see,
for the previous parameter. from figure 15 a) and b), current is interrupted for
both cases. The consequence of this variation is an
5.2. Short-line Fault increment in the arc voltage amplitudes.
Figure 14 illustrates the voltage, the current and the
conductance of the Mayr’s arc model for a variation 5.3. Model Comparison
of the arc time constant. It is verified, from figure In previous paragraphs, the parameters of the
14 a), that there is a reignition for τ = 30 µs and Schavemaker’s and Mayr’s arc models were studied,
an interruption for τ = 0.30 µs. as well as their influence on the description of the
In the first case, arc voltage assumes a relatively electric arc. However, it is important to evaluate
low value. Current continues to flow, as can be seen if for certain parameters/conditions of the circuit
from figure 14 c), and the conductance increases there are current interruption, and if this result is
and decreases as the current goes through zero, see transversal to all implemented models.
figure 14 d). Thus, it is intended to compare the various arc
In the second case, the current is interrupted at models in function of the possible current interrup-
the first current zero (see figure 14 c)). At the cir- tion, for the same simulation conditions, without
cuit breaker terminals the TVR appears (see fig- taking into account the evolution of the arc mag-
ure 14 b)), which presents a triangular shape from nitudes. For this purpose, simulations were per-
the voltage waves propagation in the line when the formed for both faults with the parameters indi-
current is interrupted. In figure 14 d), the con- cated in table 1. As shown by table 1, and although
ductance decreases, at the time of opening, from some of the models have specific parameters, an ef-
its initial value (104 S) to zero very rapidly. The fort was made to standardize them, thus allowing
reason for changing the reignition situation to in- their comparison. On the other hand, as the current
terruption is due to the fact that the conductance interruption is a function of the chosen parameters,

7
 5
x 10
14000 2000 4
Ua= 2 kV τ = 30 µs
12000 Ua= 5 kV τ = 0.3 µs

10000
1000 2
8000

6000

uarc [V]

uarc [V]
uarc [V]

4000 0 0

2000

0
−1000 −2
−2000

−4000
−2000 −4
−6000 0 0.005 0.01 0.015 0.02 0.025 0.03
0 0.005 0.01 0.015 0.02 0.025 Time [s]
Time [s]

(a) Arc voltage considering P1 = 0.8 and (a) Arc voltage considering τ = 30 µs e τ =
P0 = 1 kW 0.3 µs

4
x 10
14000 τ = 30 µs
P0=1 kW τ = 0.3 µs
12000 P0=30 kW
0 0
10000

8000

uarc [V]

uarc [V]
6000
uarc [V]

4000

2000

−2000

0.0101 0.0101 0.0101 0.0102 0.0103 0.0103 0.0104

−4000 Time [s]
0 0.005 0.01 0.015 0.02 0.025
Time [s]

(b) Transient recovery voltage - detail of a)

(b) Arc voltage considering Ua = 2 kV and for t  [0.0099, 0.0104] s
P1 = 0.8

2500
Figure 12: Arc voltage considering τ = 100 µs 2000

1500

1000
12000
P1= 0.8 500
iarc [A]

10000 P1= 0.9 0

−500
8000
−1000
6000
−1500
τ = 0.3 µs
uarc [V]

4000 −2000 τ = 30 µs

−2500
0 0.005 0.01 0.015 0.02 0.025 0.03
2000
Time [s]

0
(c) Arc current considering τ = 30 µs e τ =
−2000 0.3 µs
−4000
0 0.005 0.01 0.015 0.02 0.025
Time [s] 2000
τ = 30 µs
1800 τ = 0.3 µs

1600
Figure 13: Arc voltage considering τ = 100 µs, 1400

P0 = 1 kW , Ua = 2 kV 1200
garc [S]

1000

800

the result obtained will be different according to the 600

chosen parameter selection. However, this fact does 400

200
not invalidate the analysis in question since the pur-
0 0.005 0.01 0.015 0.02 0.025 0.03
pose is to conclude that all arc models, under the Time [s]

same simulation conditions, present the same result (d) Arc conductance considering τ = 30 µs
independently of it. The interesting information of e τ = 0.3 µs
the performed simulations for the terminal fault is
synthesized in table 2. Figure 14: Arc magnitudes considering P =
It is possible to conclude, from table 2 , that con- 30, 9 kW
trary to what would be expected for the same sim-
ulation conditions, not all models have the same
result. reveal a re-ignition, whereas the remaining arc
Cassie’s arc model and Schwarz’s arc model models, Mayr’s, Habedank’s, Schavemaker’s and

8
 5
x 10
2.5

2 Table 2: Summary of interruptions/re-ignitions for

1.5
the terminal fault
1
uarc [V]
0.5
Arc models Interruption / Re-ignition
0

−0.5
Cassie Re-ignition
−1
Mayr Interruption (2nd zero)
−1.5

Interruption (2nd zero)

P=30.9 kW
−2
P=100 kW Habedank
−2.5
0 0.005 0.01 0.015 0.02 0.025 0.03
Time [s] Schavemaker Interruption
(a) Arc voltage considering P = 30.9 kW e Schwarz Re-ignition
P = 100 kW

4
KEMA Interruption
x 10
2
P=30.9 kW
P=100 kW
1

0
lations carried out, it was possible to conclude that
each arc model has its own intrinsic characteristics,
uarc [V]

−1
since for the same simulation situation, the arc mag-
−2 nitudes are represented differently. This fact is as-
sociated to the different differential equations that
−3
constitute the several arc models implemented.
−4
0.01 0.0101 However, the parameters present in most mod-
0.0102 0.0103 0.0104 0.0105
Time [s]
els, such as the arc voltage constant Ua , the arc
(b) Transient recovery voltage - detail of a) time constant τ and the cooling power constant
for t  [0.01, 0.0105] s
P , showed an influence on the arc characterization
transverse to all models.
Figure 15: Arc magnitudes considering τ = 0.3 µs
The constant arc voltage Ua defines the arc volt-
age level at a reignition and the higher its value, the
Table 1: Parameters considered in the simulations higher the set level. The arc time constant τ influ-
for both faults ences the response time of voltage after the opening
Parameters of the circuit-breaker, and allows to change the suc-
Arc Models cess or failure of the interruption. Thus, the lower
τ [µs] U [kV ] P [kW ] a b its value is, the faster arc conductance decreases
Cassie 10 10 - - - (more effective medium deionization), and greater
the probability of current interruption. Cooling
Mayr 10 - 30.9 - -
power constant P has an influence on the maxi-
Habedank 10 10 30.9 - - mum amplitude of arc voltage and contributes in
Schavemaker 10 10 - - - the same way to the current interruption. The en-
ergy removed from the arc is proportional to this
Schwarz 10 - 30.9 1 1 constant, consequently, the higher its value is, more
KEMA 10 - - - - rapidly the medium recovers its dielectric character-
istics. Therefore, the type of parameters influence
is the same for all models and regardless of the fault
KEMA’s reveal current interruption. type. It was also possible to conclude that they have
In spite of this, Mayr’s and Habedank’s arc mod- a range of applicability depending on the fault type
els show that there was a restrike during the in- and model in question.
terruption process, since the current is only inter- When compared to each other, for the same pa-
rupted in the second passage by zero. rameters and conditions, arc models dont exhibit
all the same result in terms of current interruption.
6. Conclusions The fact that the models are made up of different
The objective of this work was to study the Black differential equations is one of the reasons for the
Box models in the electric arc modeling on a cir- results uniformity lack. On the other hand, since in
cuit breaker, when the short circuit current is in- the limit all arc models would have to present the
terrupted. same result, it is possible to conclude that the pa-
With this objective, the different Black Box mod- rameters are not adjusted to the model and fault in
els were implemented for two different faults, a ter- question. In order to complement this analysis, it
minal fault and a short-line fault. From the simu- would have been opportune to extract the parame-

9
ters of an experimental test, something that could [10] J. A. Martinez-Velasco. Power system tran-
not be achieved in this work. sients: parameter determination. CRC press,
In general, they are versatile and easy to imple- 2009.
ment models, with the ability to describe reigni-
tions and interruptions with and without restrikes. [11] P. Schavemaker. Arc model blockset for
However, they are limited to the interruption ther- use with matlab simulink and power system
mal regime and the physical processes perception is blockset–users guide. Delft University of Tech-
vanished by the mathematical formulation. nology, 2001.
In the impossibility of doing it on this work, it [12] P. H. Schavemaker. Digital testing of high-
would be interesting to apply the arc models, in voltage SF6 circuit breakers. PhD thesis, TU
future works, to experimental results in order to Delft, Delft University of Technology, 2002.
evaluate their prediction and representation of the
current interruption process. Another opportune [13] P. H. Schavemaker and L. Van der Slui. An im-
topic would be to study the ability of Black Box proved mayr-type arc model based on current-
models to represent current chopping. zero measurements [circuit breakers]. IEEE
Transactions on Power delivery, 15(2):580–
References 584, 2000.
[1] IEC 60050 - International Electrotechni-
cal Vocabulary - Details for IEV num- [14] R. Smeets and V. Kertesz. Evaluation of high-
ber 441-17-45: ”re-ignition (of an AC voltage circuit breaker performance with a vali-
mechanical switching device)”. http: dated arc model. IEE Proceedings-Generation,
//www.electropedia.org/iev/iev.nsf/ Transmission and Distribution, 147(2):121–
display?openform&ievref=441-17-45 . 125, 2000.
Accessed: 2016-10-5.
[15] S. Theoleyre. Mv breaking techniques. Cahiers
[2] IEC 60050 - International Electrotechni- Techniques (Schnedier Electric Technical Doc-
cal Vocabulary - Details for IEV num- uments), Cahier, 193, 1999.
ber 441-17-46: ”restrike (of an AC
mechanical switching device)”. http: [16] L. Van Der Sluis. Transients in power systems.
//www.electropedia.org/iev/iev.nsf/ John Wiley & Sons Ltd, 2001. ISBN:0-471-
display?openform&ievref=441-17-46 . 48639-6.
Accessed: 2016-10-5.
[3] A. Alfredsson. Live tank circuit breakers ap-
plication guide. ABB.
[4] T. E. Browne. A Study of A-C Arc Behavior
Near Current Zero by Means or Mathemati-
cal Models. Transactions of the American In-
stitute of Electrical Engineers, 67(1):141–153,
1948.
[5] A. Cassie. Theorie nouvelle des arcs de rupture
et de la rigidité des circuits. Cigre, Report,
102:588–608, 1939.
[6] W. Cigre. 13-01: Applications of black-box
modelling to circuit breakers. Electra, 149:40–
71, 1993.
[7] Cigré Working Group 13.01. State of the art
of circuit-breaker modelling. Cigré Brochure,
(135), 1998.
[8] N. Gustavsson. Evaluation and simulation of
black-box arc models for high-voltage circuit-
breakers. 2004.
[9] U. Habedank. On the mathematical-
description of arc behavior in the vicinity of
current zero. etz Archiv, 10(11):339–343, 1988.

10