4’

“4,

IEEE Transactions on Power Systems, Vol. 3, No. 2, May 1988 799

M O D E L L I N G O F C I R C U I T ERECSKERS I N T H E

ELECTROMAGNETIC TRANSIENTS PROGRAM

V I PHANIRAJ A,G,PHADKE
Student Member, I E E E Fellow, I E E E

V i r g i n i a Polytechnic I n s t i t u t e and S t a t e U n i v e r s i t y

ABSTRACT - The recent publication of experimental and Since a large proportion of all power system transients are
theoretical results from verified arc models has made pos- initiated by circuit breaker operations, the need to repre-
sible the implementation and testing of a dynamic circuit sent breakers accurately in EMTP has long been recog-
breaker model in the Electromagnetic Transients Program nized [3]. Until now, the approximations used for
(EMTP). An estimator was developed to obtain model modelling circuit breakers included voltage or current
parameters from test data. Results from this are given, controlled switches, and predetermined time-dependent or
and it’s data requirements specified. To illustrate an ap- nonlinear resistances. These are adequate for cases such
plication of the model not previously possible with the ex- as Transient Recovery Voltage (TRV) calculation, but are
isting capabilities of EMTP, simulations of load current not suitable for many other applications. In particular,
interruption in a multi-terminal HVDC system were per- phenomena such as thermal and dielectric failure cannot
formed. Results from these are included, along with a be modelled. A more realistic model would perform these
discussion of the effects of system and model parameter and other functions, and find application in the following
variation on the interruption process. types of studies:
INTRODUCTION Circuit Breaker Testing : With the increase in short-
circuit capacity of power systems and the consequent rise
The Electromagnetic Transients Program (EMTP) is an in interrupting capabilities, direct testing of breakers is
extensively used tool for the analysis and simulation of becoming more difficult because of the high short-circuit
power system transients [1,2]. It contains a large number capacity test source requirement. Though synthetic test-
of independent modules, each modelling a different com- ing is commonly used, an accurate EMTP model could be
psnent such as a transformer, a synchronous machine or used to simulate a direct test, and supplement the syn-
a transmission line. Inductances and capacitances are re- thetic tests.
presented by an equivalent circuit consisting of resistances
in parallel with current sources. These equivalent circuits Interruption of Small Inductive Currents : This situation
are obtained by the application of the trapezoidal rule of , which arises often in transformer and reactor switching,
integration [I]. Through this transformation, the nodal may lead to current chopping and multiple restrikes which
admittance matrix [Y] becomes real and symmetric, and in turn cause dangerous overvoltages. The model could be
the differential equations for capacitances and used to identify such problems, and determine the nature
inductances become algebraic equations. Thus the gov- of corrective action, such as the ratings of a surge arrester
erning equation of the system is : to be connected.
[U.e(r) = i(r) - [I] (1) The need for a better model having been established, the
where e([)and i(r) are column vectors of the node voltages next task was to choose appropriate models to fulfill the
and injected currents, and [I] is the known vector of requirements. The main criteria used for selection were :
equivalent current sources representing the system history. 1. Availability of Model Parameters and Test Re-
Eq. (1) is to be solved repeatedly as the simulation
progresses, and therefore in order to enhance the program sults.
speed, the [Y] matrix is factored and stored in the L-U 2. Numerical Simplicity and Robustness.
form. This technique requires that the [Y] matrix remain
constant, making it difficult to simulate switches, 3. Range of Applications.
saturable devices and elements whose impedance is time-
varying, such as circuit breakers. The effect of switching Based on these factors, three models were chosen, and
operations however can be pre-computed and the resulting these are discussed in the next section.
modifications made directly to the L-U factors.
ARC MODELS
The three models chosen for implementation were, (1) the
modified Mayr or Avdonin model [4] ; (2) the Urbanek
model [4] ; and (3) the Kopplin model [5].
Avdonin Model :
The Avdonin model equation is :
This papa was sponsored by the IEEE Power EnSineering Society
for presentation at the IEEE Power Industry Computer Applica- 1-a rl-a-p
tion Conference, Montreal, Canada, May 18-21,1987. Manuscript d r - r --.i.v
was published in the 1987 PICA Conjerence Record. dt A A.B

0885-8950/88/0500-0799$01 .WO1988 IEEE

c

800

where v,i and r are the arc voltage, current and resistance stage proccdurc. At first, the nonlinear element is
respectively, and and A, B, a and p are the breaker model open-cii cuited and the rcsponsc of the nctwork calculated,
parameters. This model is a derivative of the Mayr model specifically the open-circuit voltage V, and the Thevenin
with the time constant 0 replaced by A.ra and the power impcdance Z,heY. If the nonlinearity can be characterized
constant P replaced by Brp. This model has been testcd at by the equation :
Hydro-Quebec, and publiqhcd results arc availablc [4] for
validation of the EMTP simulations. The model is capable Vn =AI,) (5)
of representing arc interruption and thermal failure, and then the actual value of I, can be evaluated from
has been used for modelling current chopping. It cannot
simulate diclectric breakdown or multiple restrikes, but v,, - Zthev.zn =AI,> (6)
has the advantages of being computationally simple and
robust. where for multiple nonlinearities,V, and I, are column
vectors and Z,hevan impedance matrix. Having calculated
Urbanek Model : I, , usually by an iterative procedure, the final network
solution is obtained by superimposing the response due to
This model can be described by the differential equation I, upon the open-circuit response.
The use of this method constrains the types of networks
that can be simulated, chiefly in that two nonlinear ele-
where g is the arc conductance, V the arc voltage for large ments cannot be electrically adjacent. This restriction can
currents, ( the breakdown voltage of the cold arc channel be overcome by the use of stub lines, with short travel
and the other variables arc as dcfined previously. This times, but this artifice complicates and distorts the system
model is the most complex of the three, and can represent modelling. It could be removed by solving all nonlinear
arc interruption and both thermal and dielectric failure. equations of the form of (6) simultaneously, but this
In addition to these, it can model current chopping and would be computationally difficult, and would require the
re-ignition. An estimation program to determine the interfacing of all modules that use the compensation
model parameters has been written [6], and test results are method (such as synchronous machines and circuit break-
available [4]. ers) in a common segment of the program. EMTP is not
structured to allow this at present, and the breaker model
Kopplin Model : interface developed here cannot be connected directly to
othcr nonlinear elements.
This equation, which was used in EMTP to model gener-
ator circuit breakers is described by : The iterative procedure used to solve Eq. (6) is of the
predictor-corrector form. Since near an a.c current zero
(4) the current decreases practically linearly, this slope can be
used to provide an initial guess for the current at the next
where r(g) = k,. (g +0.0005)o.25 and time-step. The slope is evaluated by a recursive moving-
P(g) = k ( g + 0.0005)0.6, k and k, being model param- average equation, and is continuously updated. The first
eters. TI& model is also d e h e d from the Mayr model, guess therefore is :
and can be used with larger time-steps than the Avdsnin I(t + At) = l ( t ) + At.-d l (7)
or Urbanek models. It simulates thermal breakdown, and dt
test results are available [5]. The next section describes the
incorporation of these arc equations into a complete model and the initial arc voltage estimate is
for a circuit breaker. V(t + At) = I(t + At).r(t) (8)
MODEL IMPLEMENTATION AND TESTING The rate of change of resistance (drldt), can be found from
the appropriate arc equation, and therefore an estimate
Implementation made of the arc resistance by the trapezoidal rule
The arc equations described above can be used only dur- r(t + At) = r(t) + 0.5(--,dr
dt
+ -drp ) A t (9)
ing a short interval commencing about 100 microseconds
before the current zero, while the overall interruption where (drldt), is the rate of change saved from the previ-
process may take a few periods at fundamental frequency. ous time-step. The estimate for the current is then cor-
Therefore a complete model had to be developed, of which rected by the equation
the arc model would be a crucial component.
The first major change in breaker status during opening
occurs when the contacts part, when the arc voltage starts
to build up and the arc resistance increases. It is known
from experimental evidence that the arc voltage rises until and a convergence'check made by comparing the two arc
a certain value, and then remains steady until just before voltage estimates
the current zero, at a value that depends on the inter-
rupting medium used. Therefore, in EMTP, the initial
stages of the opening are modelled by a constant arc volt- and
age until the inception of the arc instability period, with
the voltage determined by any of the equations described V2 = r(t + At) .I(t + At) (12)
previously. This procedure is fairly realistic, and provides
suitable initial values for the arc equations and minimizes If they agree to within a specified tolerance, the iteration
the numerical problems due to a sudden voltage change. is over , otherwise the process returns to Eq. (8) with an
updated estimate for the current from Eq. (IO). It was
The breaker is interfaced to EMTP as a nonlinear com- found that 3 or 4 iterations were generally sufficient, and
ponent by the compensation method [3], using a three- convergence problems were rare.
 801

Once the arc resistance and current have been deter-

mined, the routine is ready to return them to the main
program for processing. The last stage is to evaluate the
breaker status for the next time-step , which is necessary
only in the arcing regime. There are three possibilities : ( I )
The interruption succeeded ;(2) that it failed :or (3) there
was no change and the arc calculations should be contin-
ued. A decision is made by the following criteria, and the
appropriate flag set for future use. The criterion for de-
claring the interruption to be a success is that either the
arc resistance should exceed IO’OR or that the rate of
change of resistance should be greater than 1018R/second.
In real interruptions, when failure occurs, the arc resist-
ance becomes zero, but in a numerical simulation it usu-
ally overshoots and becomes negative. This would cause
computational difficulties in EMTP, and therefore the in-
terruption is deemed to have failed when the rate of
change of resistance starts to decrease. This hypothesis
was tested by conducting many simulations, and in no
case was the interruption successful after a decrease in the
rate of change of arc resistance following the current zero.

Testing
Two basic test circuits were used, Circuit 1 for the Circ . Rd Ld microF nanoF
‘od
Avdonin and Urbanek models and Circuit 2 for the ohms mH ‘d
Kopplin model. These circuits and the results from tests A 67,38 6,90 1,066 0,O
thereof are taken from [4] and [SI respectively, and are
shown in Figures 1 and 2. The component values for Cir- B 60.34 6.90 1,037 22.66
cuit 1 are given in Table 1 and the arc model parameters
in Table 2.
c 62#77 6.90 1,029 44.26

TABLE 1: C i r c u i t 1 Component Ualues

Ld

I fiUDONIN I URBANEK I
AIR OIL SF6
A 6,E-6 6.E-6 1-3E-6 P = 3,OE4
B 1.6E7 1.OE8 1.OE6 v = 8000.
a( -0,ZO ~ ~~
-0.15
~~~
-0,15 e = 2,E-6

F i g u r e 1: Teot C i r c u i t 1
IRI -0.50 T-0.60 I -0-26 I%= 46.E4

TABLE 2: ARC HODEL PARAMETERS

l-C 4.14 4.14 5.59 5.61 8.70 8.67 3.61 3.60

TABLE 3: Comparison o f Benchmark ( Reference 4 1 and EMTP R e o u l t r 1

802

325 p H

Figure 2 : Test C i r c u i t 2
60 I20
F i g u r e 6: Kopplin Model Arc Voltage
X : Micromecondo ; Y: U o l t s

0 30 60 90
F i g u r e 3: Avdonin Model Arc Uoltage
X: Time i n Microsecond@.
F i g u r e 8 : Kopplin Model Post-Arc Current
Y : Uoltage i n K i l o v o l t s .
X: Microsecond@ I Y: Ihrperss
The input is assumed to consist of N regular samples of
the arc voltage and current, at known instants of time.
These are transformed to values of arc resistance, and
used as input to a least-squares curve fitting routine which
produces the best polynomial r(t) fitting the points sup-
plied. r(t) is analytically differentiated and a set of values
for i(t), the derivative obtained. From the Avdonin
equation, given a parameter vector X = ( A , B, a, p)' the
derivative R(t,X) can also be computed. The estimation
problem then reduces to that of choosing X such that i ( f )
-m and R(t,X) are as close as possible. This is solved by an
iterative nonlinear least-squares estimator.
F i g u r e 4: Avdonin Model Post-Arc Current If A Y is the vector of differences between tft) and R(t,x),
X: Time i n Microseconda. the correction to the iterate X is given by
Y : Current i n -pores.
AX = [H'H]-l["j'.AY (13)
Kt Reference C151 EMTP where H is the,Jacobian matrix, consisting of the partial
microaec
derivatives of R(t,X) with respect to A, B, a and p, and
' 13 SUCCESS SUCCESS has 4 rows and N columns. The new estimate X' =
15 SUCCESS SUCCESS X + AX is used to update A Y . If the norm of AY is suf-
ficiently small, the iteration has converged ; otherwise A X
17 SUCCESS SUCCESS is recalculated and further iterations made.
I 1'8 I FA ILUR E I FAILUR E I The choice of data samples hinges on two factors : the
number of samples and the sampling instants. Based on
TABLE 4: KOPPLIN MODEL RESULTS many trials, the minimum number of samples was found
to be about 10, while using more than 40 points resulted
MODEL PARAMETER ESTIMATION in no significant improvement in the estimate. All results
in this section are based therefore upon N = 30 , at a
As mentioned previously, one of the chief problems faced spacing of half a microsecond. If the samples start much
during the implementation was the lack of standard data before the current zero, the rate of change of resistance is
for arc model parameters and field test results. Therefore very small and hence accuracy is lost in computing r(t)
in order to render the EMTP model more useful, it was and it's derivative. On the other hand, if the samples are
neccessary to develop an auxiliary routine to estimate the all from the post current-zero period, r(t) and r ( f ) are ex-
parameters from user-supplied test results on a givcn cir- tremely large and numerical overflow problems arise both
cuit breaker. The estimator developed was for the in fitting r(t) and calculating the inverse of (H'.H). After
Avdonin model. Since the steady arc voltage prior to arc repeated trials it was found that the samples should be
instability can be determined directly by inspection of the taken from a period of 20 ps before and 20 ps after the
test waveforms, the aim was to estimate the remaining current zero, for a total maximum sample span of 40 ps
constants A, B, a and p . during which the arc equation is valid.
 803

In the absence of field test data, the output from EMTP 3. Since measurement noise adversely affects the esti-
simulations was used as input to the estimator in an at- mation algorithm , the data will have to be processed
tempt to recreate the original parameters. To validate the before using the estimator. Some smoothing and fil-
estimation, the interruption limits for both sets of param- tering may be necessary in the pre-processing stage.
eters were evaluated from EMTP runs and these are listed The routine itself has to be added to the EMTP, and
in Table 5. Estimates were made for all three models ; in more rigorously tested.
the case of thc Urbanek and Kopplin models, the original
interruption limits are compared with those obtained from Application to HVDC Breaker Simulation
simulations using the equivalent Avdonin model parame-
ters. These parameters are estimated by using the Kopplin This example was chosen to illustrate the use of the new
or Urbanek model outputs as input to the estimator. It model in a situation which could not previously be simu-
can be seen that there is good agreement between the lated by EMTP. The principle of operation of the DC
various limits and the original values. breaker simulated here consists of inducing an oscillation
between the negative V-I characteristics of the arc and a
parallel L-C circuit, after the parting of the contacts. De-
pending on the system and breaker properties, the oscil-
LIMIT lations may increase in amplitude and induce an arc
current zero, leading to the interruption of the DC cur-
rent.
A 6.E-6 6,04E-6 Figure 7 shows the test system used, a multi-terminal
AVDONIN B 1,6E7 2,098E7 3-56 3.69 HVDC system taken from [7]. The simulation was per-
formed using the UBC (University of British Columbia)
ol -0,20 -0,1963
version of EMTP, which is much smaller and simpler to
a -0.60 -0,6266 use. The Kopplin model was used, since it's longer time
P 30.E3 A 2.87E-6 constant makes it suitable for lengthy simulations. It is
also similar to, though less complex than, the model used
URBANEK V 8000. B 1.046E7 in [8] for the same purpose. All cases described here cor-
e 46.E4 r< - 0 . 0 8 6 2 , 7 e 2.66 respond to the interruption of the rated load current of
\ 2.E-6 g -0,6679 450 A. The effects of varying L, C, K,and Kp and the line

I
inductance were studied.
4.E6 A 1.51E-6
It was found that there is a minimum shunt capacitance
KOPPLIN B 4.043E6 3 4 . 6 36.6 required to sustain the oscillation, and both a minimum
- 0 . 2 6 3 1 kV kV and a maximum shunt inductance. The smaller the
K 16.E-6 P -0.614 inductance, the larger the capacitance required to cause
the current zero. The effect of K,is inversely related to the
T I B U 6 : #ode1 Parameter E 8 t b ~ a t i O nb 8 U l t 8 interruption level ; smaller'the K, , the higher the current 1

that can be interrupted. As is to be expccted intuitively,

the power constant Kp and the current interruption level
The estimation routine was found to be extremely sensi- are in direct relation to one another.
tive to measurement noise ; a noise amplitude of 1 YObe-
ing sufficicnt to distort the estimates significantly, or in
some cases cause a failure to converge at all. This can be
remedied, as discussed later.

FUTURE RESEARCH AND APPLICATIONS

1
aa
There are several areas in which refinements can be made BREAKER
to the model and algorithm ; some of these are described

P
here.
1. Since test cases involving dielectric restrike and
chopping are not available, simulation of these phe-
nomena has not been attempted. In particular,
dielectric restrike presents some interesting problems,
mainly convergence-rclated or of a computational na- F i g u r e 7: T e a t HUDC Syrtern
ture. Tests of these are planned using the Urbanek
model, and field verification will bc attempted later
this year in a case involving reactor switching. Other
related issues include the modelling of multiple resist- Due to space constraints only a few cases are shown here.
ance switching, in order to damp the TRV, and invcs- Figure 8 shows the current with L = 0.25 mH and C =
tigation of the missing current zeroes in faults close to 1.0 pF. A current zero is induced, very quickly, and the
a synchronous machine. breaker cannot interrupt the current due to thermal fail-
ure. When the inductance is raised to 0.5 mH interruption
2. In the area of algorithm improvement the main focus occurs successfully, as shown in Figure 9. Figure 10 shows
is on increasing the minimum step-sizc that must be the effect of a much larger inductance, when no current
used for accurate simulations. At prcsent this is about zero is induced at all, with L = 20.0 mH. The results are
0.2 ps for the Avdonin model ; a value of about 1 ps very similar to those in [8],when the effect of the simple
would be desirable. Though convergence problems arc model used hcre is considered , and also to the exper-
have not bcen encountered yet, the possibility of these imental results rcportcd in [9].For example, the frequency
occurring increascs while simulating cases of restrike of oscillations in Figure 9 is about 8000 Hz, while the fre-
or chopping. fluency in Figure 9 of [8] is about 7800 Hz. Work is cur-
804

rently in progress to implement the more detailed model CONCLUSIONS

from [8] in EMTP to provide a more accurate tool for such
simulations. 1. An accurate and realistic circuit breaker model with
1-0 I I I I 1 a choice of three dynamic arc equations has been in-
corporated in EMTP. It was tested against previously
published results, and performed very well.
2. EMTP now has the capability to simulate phenomena
such as thermal and dielectric failure, current chop-
ping and interruption.
3. Default values for the arc model parameters are pro-
vided to overcome the difficulty in obtaining input
data. An estimation program has been written and
tested, which evaluates these parameters from user-
supplied test data.
Acknowledgements
This research, forming part of the M.S thesis of
V.Phaniraj at Virginia Polytechnic Institute, was sup-
2.5 ported by the EMTP Development Co-ordination Group
and the Electric Power Research Institute (EPRI/DCG).
F i g u r e 8: HUDC B r e a k e r Current
X: T i m e i n millimecondm.
- Came 1
REFERENCES
Y: Current i n amperem.
I. H.W. Dommel, “Digital Computer Solution of
Electromagnetic Transients in Single- and Multi-
phase Networks” ; IEEE Transactions on Power
Apparatus and Systems, April 1969, pp. 388-399.

2. A.G. Phadke (organizer), “ Digital Simulation of

Electrical Transient Phenomena ” ; IEEE Tutorial
Course, 1980.

3. H.W. Dommel, “Nonlinear and Time-Varying

Elements in Digital Simulation of Electromagnetic
Transients” ;IEEE Transactions on Power Appa-
ratus and Systems, November/December 1971, pp.
2561-2567.

4. G. St-Jean,R.F. Wang, “Equivalence between

Direct and Synthetic Interruption Tests on High
Voltage Circuit Breakers” ; IEEE Transactions on
Power Apparatus and Systems, July/August 1983,
pp. 22 16-2223.
F i g u r e 9: HUOC Breaker Current
X: T i m e i n millimecondm.
- Came 2
5. E. Thuries, P. Van Doan, J.Dayet, and B.
Joyeux-Bouillon, “Synthetic Testing Method for
Y : Current i n amperem.
Generator Circuit Breakers” ; IEEE Transactions
on Power Delivery, January 1986, pp. 179-184.

6. L. Blahous, “Derivation of Circuit Breaker Pa-

rameters by Means of Gaussian Approximation”
; IEEE Transactions on Power Apparatus and
-400 Systems, December 1982, pp. 461 1-4616.

7. W.F. Long, “ A Study of some Switching Aspects

of a Double Circuit HVDC Transmission Line” ;

300
tems, March/April 1973, pp. 734-74 I ,

8. K. Ragaller, A. Plessl, W. Hermann, and W. Egli

,“ Calculation Methods for the Arc Quenching
System of Gas Circuit Breakers ” ; CIGRE Paper
No. 13-03, Paris, 1984.
400- i.2 5
I
1.5
 805

9. B. Bachmann, G. Mauthe, E. RUOSS, H.P. Lips,

J. Porter, and J. Vithayathil, “ Development of a
500 kV Airblast HVDC Circuit Breaker” ;IEEE
Transactions on Power Apparatus and Systems,
September 1985, pp. 2460-2466.

INFORMATION FOR AUTHORS

In January 1986 IEEE TRANSACTIONS ON POWERAPPARATUS AND SYSTEMS was replaced with three new TRANS-
ACTIONS which are sponsored by the Power EngineeringSociety. The new publications will continue to be devoted to all
aspects of electric power generation, transmission. distribution, and utilization. Original contributionsof lasting value
to the profession are sought for publication in the appropriate TRANSACTIONS according to the subject matter as
follows.
IEEE TRANSACTIONS ON ENERGY CONVERSION: Research, development, design, application, construction, installa-
tion and operation of electric power generating facilities (along with their conventional,nuclear, or renewable sources)
for the safe, reliable and economic generation, conversion and control of electrical energy for general industrial, com-
mercial, public and domestic consumption.
IEEE TRANSACrlONSON POWER DEUVERY:Research, development, design. application, construction, installation
and operation of apparatus, equipment, structures, materials and systems for the safe, reliable and economic delivery
and control of electrical energy for general industrial, commercial, public and domestic consumption.
IEEE TRANSACXIONS ON POWER S Y ~ ~ Requirements,
MS planning, analysis, reliability, operation and economics
electrical generating, transmission and distribution systems for general industrial, commercial, public and domestic
consumption.
In general, papers are preprinted and presented at a technical meeting of the Institute before being published in the
appropriate Transactions; discussions by qualified specialists are sought and published along with a closure by the
author. All contributionsare reviewed by an appropriate technical committee of the IEEE Power Engineering Society
as listed on the inside back cover.
Following review, a decision is made as to publication. All papers accepted for publication will be published in full in
one of the TRANSACTIONS and a one page summary will be published in the IEEE POWERENGINEERING REVIEW.
COPYRIGHT: It is the policy of the IEEE to own the copyright to the technical contributionsit publishes on behalf
of the interests of the IEEE, its authors, and their employers,and to facilitate the appropriate reuse of this material by
others. To comply with the US.Copyright Law, authors are required to sign an IEEE copyright form before publica-
tion. This form, returns to authors and their employersfull rights to reuse their material for their own purposes. Authors
must submit a signed copy of this form with their manuscripts.
The original manuscript, typed on IEEE mats in accordance with the IEEE Power Engineering Society Publication
Guide for Power Engineers, November 1986, should be sent for review processing to the following address.
IEEE Headquarters
Society Special Services
345 East 47 Street
New York, NY 10017
IEEE P,ower Engineering Society Publication Guide for Power Engineers, November 1986, available on request
from the Society Special Services Department, at the above address, provides a guide to approved style and usage.
Detailed instructions, copyright forms, and standard mats for the preparation and handling of manuscripts can be
obtained from the IEEE PES Special Activities. Manuscripts should be limited to seven pages including illustrations.
Other editorial correspondence should be directed to the Editor.
Harold Gold
1037 North Primrose
Rialto, CA 92376