CIRED 19th International Conference on Electricity Distribution Vienna, 21-24 May 2007

Paper 0001

UNIVERSAL ARC RESISTANCE MODEL "ZAGREB" FOR EMTP

Sandra HUTTER Ivo UGLEŠIĆ

HEP d.o.o., Elektra Zagreb – Croatia FER – Croatia
[email protected] [email protected]

ABSTRACT
We developed the arc resistance model for use with EMTP.
The electrical arc was simulated using a black box model
based on the Cassie and Mayr arc equation. The model was
tested on a benchmark circuit and the simulation results
match well with the referenced models.

INTRODUCTION
The need for an appropriate arc model arose repeatedly in
numerous EMTP studies on problems related to overvoltage
calculation. An arc model would be useful for various
applications such as breaking of small inductive currents,
calculation of VFTO in GIS or modeling of different types
of SF6 and pneumatic circuit breakers applied in studies of
switching overvoltages (e.g. interruption of short circuit Fig. 1. A.C. arc extinction [2].
currents).
If the input power, during the short time following the
DEVELOPMENT OF THE ARC MODEL instant of current zero (<1 ms) exceeds the arc power loss
the arc is going to reignitiate.
Black box models use a mathematical description of the
electrical behavior of the arc. These types of models do not Cassie and Mayr arc equation
give a full representation of the physical processes that goes Different arc models use different functions to describe the
on inside a circuit breaker [1]. relationship between the arc cooling power, the thermal
time variable and conductance.
Recorded voltage and current traces during the "thermal
period" are used to obtain circuit breakers parameters that The description of these different approaches is given and
are later on substituted in differential equations. We used explained in [1].
the most basic Cassie-Mayr equation, as well as its
modifications from the Schwarz-Avdonin model. Assuming The electrical arc was simulated using its conductance-
that the circuit breakers parameters are known, this model dependent parameters described by the Cassie and Mayr arc
can also be used with other equations derived from the equation:
Cassie-Mayr equation.
dg 1 i2
= ( − g) (1)
PHYSICAL PROCESS: dt τ (g ) P( g )

A short theoretical explanation of the physical process that where g represents the arc conductance, i - the arc current,
occurs during AC arc extinction in the case of short line P(g) - the arc cooling power and τ (g) - the arc thermal time
fault (SLF) is given below [2]. constant. Calculation of the arc conductance requires data
on the cooling power and the thermal time constant.
The main task of a circuit breaker is current interruption,
which requires that the interelectrode gap changes from a Parameters P(g) and τ(g) are conductance dependent. The
conductive plasma into an insulating gas. At the instant of cooling power P and the thermal time constant τ depend
current zero, the input power is zero, because the restriking on temperature, circuit breaker type and design, and they
voltage is zero. However, after that zero time point a small can be defined as a function of conductance f(g).
post arc current is going to flow. Recovery voltage between
circuit breaker contacts will increase rapidly. P = Po ⋅ g β (2)
α
τ =τo ⋅ g (3)

CIRED2007 Session 1 Paper No 0001 Page 1 / 4

CIRED 19th International Conference on Electricity Distribution Vienna, 21-24 May 2007

Paper 0001

Cassie and Mayr arc equation could be solved if a

sufficiently small time interval ( Δ t ) can be observed in
which P and τ are constant. The differential equation was
solved numerically with the Euler method for differential
equations of the first order. The solution is:

⎞⎡
−Δ t ⎤
⎛ i2
g (t + Δ t ) = g (t ) + ⎜ − g (t ) ⎟ ⎢1 − e τ ( g ) ⎥ (4)
⎜ P (g) ⎟⎢ ⎥
⎝ ⎠⎣ ⎦

Depending on the difference between input (power) and Fig. 2. Test circuit (60 Hz) and arc model parameters:
cooling thermal energy (power loss), the arc temperature P0 = 4 MW and β= 0.68; τ = 1.5 μs and α= 0.17 [3].
(and conductivity) is either going to increase or decrease.
We used a time step of 10 ns. Such small time step can be
Benchmark circuit problematical in some applications and can prolong a
We used the test circuit described in [3]. The benchmark computation time, however in EMTP a time step has to be
circuit is depicted in Fig. 2. fixed.

The first step in solving equation (4) is to find the solutions Calculated post zero arc currents and peak values of the arc
of the network equations (in EMTP) with the initial voltage were compared to three different models developed
conductance value go = g (to). The solution to these at Delft University of Technology and are shown in Fig. 3-
equations gives the value for the current i(to). The initial 5. below.
values P( g o ) and τ ( g o ) are also calculated for the
conductance initial value. Figures 3a., 4a., and 5a. show the graphs of the calculated
values with the model ZAGREB (Appendix A) and the
The following equations are used: figures 3b., 4b., and 5b. (in the second column) show the
graphs obtained with the three different models in the same
test circuit described in [3], which are: EMTP96, XTrans
P = Po ⋅ g β = 4 MW ⋅ g 0.68 (5)
and MATLAB Simulink/PSB/AMB based models.
α 0.17
τ = τ o ⋅ g = 1.5 μs ⋅ g (6)
Graphs of the post zero arc current and arc voltage give
Initial resistance value was in this case 0.0001 Ω. more information than the arc resistance graph and are more
With known values at the time step to the next time step is suitable for comparison. On Fig.3a. current zero is at the
solved g (t1 ) = g (t o + Δ t ) in MODEL section of the EMTP. time point 8.1749 ms from the beginning of simulation, and
a time interval of 4.1 μs is shown. On Fig 4.a. and 4.b. are
The output value from MODEL section is resistance and it
is used as input value for nonlinear resistance component shown, respectively, time intervals of 648 μs and 10 μs
R(TACS) Type 91. before the time instant of current zero.

0.00
[A]
-0.05

-0.10

-0.15

-0.20

-0.25

-0.30

-0.35
8.1479 8.1487 8.1495 8.1504 8.1512 [ms] 8.1520
Fig. 3a. Post zero arc currents:
- developed model ZAGREB Imax = -0.35 A Fig. 3b. Post zero arc currents:
- comparison with referent models

CIRED2007 Session 1 Paper No 0001 Page 2 / 4

 CIRED 19th International Conference on Electricity Distribution Vienna, 21-24 May 2007

Paper 0001

3500
[V]
3000

2500

2000

1500

1000

500

0
7.5 7.6 7.7 7.8 7.9 8.0 8.1 [ms] 8.2
(f f )
Fig. 4a. Arc voltage: developed model
Fig. 4b. Arc voltage: comparison with referent models
3500
[V]
3000

2500

2000

1500

1000

500

0
8.138 8.140 8.142 8.144 8.146 [ms] 8.148
Fig. 5a. Arc voltage – detail:
-developed model ZAGREB Umax = 3442 V Fig. 5b. Arc voltage – detail: comparison with referent
models
Air blast and SF6 circuit breakers
In order to compare the behavior of the air blast and SF6 Different quantities (post arc current peak zero, voltage
circuit breakers, parameters of the model "Zagreb" used in range after zero and thermal period duration) characterizing
the benchmark circuit were replaced with the following the thermal period of the switching process in the air blast
values from the literature [4]: and SF6 circuit breakers are compared in table 1. The given
P = Po ⋅ g β = 16 MW ⋅ g 0.5 (7) values are valid only for the given benchmark circuit.
τ = τ o ⋅ g α = 6 μs ⋅ g 0.2 (8) The air blast circuit breaker has in general a longer thermal
period (Fig. 6), for this benchmark circuit and chosen circuit
breaker parameters the thermal period and the peak value of
Air the arc voltage were respectively 2 and 2.7 fold higher.

Table 1. Obtained values of post-arc current peak zero,

voltage range after zero and thermal period duration for air
blast and SF6 circuit breakers
Air
Air blast SF6 Ratio
Thermal period
6 μs 3 μs 2
duration
Postarc current
1.885 A 0.35 A 5.4
peak zero
t i Air
Voltage range after
9314 kV 3442 V 2.7
zero
Fig. 6. Thermal period for short line fault interruption [1].

CIRED2007 Session 1 Paper No 0001 Page 3 / 4

CIRED 19th International Conference on Electricity Distribution Vienna, 21-24 May 2007

Paper 0001

Application [2] C.H Flurscheim, et Al, 1982, Power circuit breaker

This model was successfully applied in the switching theory and design, IEE, Peter Peregrinus Ltd, London,
overvoltage analysis conducted for a 400/110 kV UK, 191.
substation.
[3] P. H. Schavemaker, L.Van Der Sluis, 2002, "The Arc
Analysis showed that consecutive switching of two or more Model Blockset", Proceedings of the Second IASTED
SF6 or pneumatic circuit breakers will not cause high International Conference, POWER AND ENERGY
overvoltages, due to the fact that the electrical arc will be SYSTEMS (EuroPES), 644-648.
extinguished at the moment of the current zero passing.
[4] V. Phaniraj, A.G. Phadke, 1988, " Modeling of Circuit
Similar model with different parameters was also applied in Breakers in the Electromagnetic Transients Program",
calculation of VFTO in GIS during disconnector switching. IEEE Trans. On Power Systems, vol. 3, no.2, 799-805.

CONCLUSION APPENDIX A
Black box arc resistance models for the SF6 and air blast MODEL ZAGREB
circuit breakers were developed for use with EMTP. comment--------------------------------------
| Schwartz arc model |
Arc resistance is calculated for each time step in MODELS -----------------------------------endcomment
section, from equation (4), which is a numerical solution of INPUT U1, U2
the differential Cassie and Mayr arc equation. This solution OUTPUT RB
is valid if sufficiently small time intervals can be observed. VAR I, RB2, RB, G, G2, TAU
INIT
Circuit breakers parameters: cooling power and the thermal RB:=0.0001
time constants, were obtained from the literature. ENDINIT
EXEC
Depending on the circuit breakers parameters, different I:=(U1-U2)/RB
types of circuit breakers will have different thermal period IF (ABS(I)>1.E-12) THEN
duration. In general, a rate of rise of the electrical arc G:=(1./RB)
resistance is higher for the SF6 circuit breakers, which TAU:=(1.5E-6*(G**0.17))
means that the dielectrical strength between the circuit G2:=((I**2.)/(4000000.*(G**0.68))-G)*(1.-
breaker contacts recovers faster. -1./EXP(timestep/TAU))
RB2:=(1./ABS(G+G2))
REFERENCES RB:=RB2
ENDIF
[1] CIGRE WG 13.01, 1993, "Applications of Black Box ENDEXEC
Modelling to Circuit Breakers", Electra no.149, 41-71. ENDMODEL

CIRED2007 Session 1 Paper No 0001 Page 4 / 4