1

A Study of Electromagnetic Transient

Simulations Using IEEJs
West-10 Benchmark Power System Model
Taku Noda, Hiroshi Takizawa and Takayuki Nakajima

Abstract Electromagnetic-transient (EMT) simulations of

relatively-large power systems with generator mechanical dynamics have become quite common especially for studies of power
systems including power-electronics converters. The Power and
Energy Society of the Institute of Electrical Engineers of Japan
(IEEJ) prepared benchmark power system models for transient
stability (TS) simulations. Among those, the West-10 benchmark
power system model approximately represents the long radial
power system in the western part (60-Hz part) of Japan with ten
generators. In this paper, the West-10 benchmark power system
model is expanded and converted to an EMT model, and it is
shown that the results obtained by the EMT model agree well
with those obtained by the TS model in most cases. It is also
found that when the dc components of fault currents are large
the results obtained by those two simulation methods are different
due to current zero missing of circuit breakers.
Index Terms EMT simulations, IEEJs West-10 benchmark power system model, relatively-large power systems, and
stability-type simulations.

I. I NTRODUCTION

ODERN power systems utilize power-electronics converters for dc transmission, back-to-back system interconnection (frequency conversion between 50 and 60 Hz),
system stabilization, and so on. Since waveform-level calculations are essential to take into account switchings in
power-electronics converters, studies of power systems including power-electronics converters are now often carried out
by electromagnetic transient (EMT) simulations. This trend
gives further importance to EMT simulations in addition
to studies of conventional phenomena such as overvoltages,
inrush currents, and abnormal oscillations. Thanks to the
research and development of simulation methods and also
to the progress of simulation hardware (computers), EMT
simulations of relatively-large power systems with generator
mechanical dynamics have become quite common. The EMT
simulations related to power electronics converters mentioned
above and sub-synchronous resonance often involve relativelylarge power systems.

T. Noda is with Electric Power Engineering Research Laboratory, CRIEPI

(Central Research Institute of Electric Power Industry), 2-6-1 Nagasaka, Yokosuka, Kanagawa 240-0196, Japan (e-mail: takunoda@criepi.denken.or.jp).
H. Takizawa and T. Nakajima are with DCC (Denryoku Computing Center),
1-1-1 Manpuku-ji, Asao-ku, Kawasaki, Kanagawa 215-0004, Japan (e-mail:
{taki, nakaji}@dcc.co.jp).
Paper submitted to the International Conference on Power Systems
Transients (IPST2015) in Cavtat, Croatia, June 15-18, 2015.

The Power and Energy Society of the Institute of Electrical

Engineers of Japan (IEEJ) prepared benchmark power system
models for transient stability (TS) simulations. Among those,
the West-10 benchmark power system model approximately
represents the long radial power system in the western part
(60-Hz part) of Japan with ten generators.
In this paper, the West-10 benchmark power system model
is expanded and converted to an EMT model for the implementation in EMT simulation programs. The major differences
between the EMT and the TS simulation are as follows.
The EMT simulation calculates three-phase voltage and current waveforms, while the TS simulation calculates positivesequence r.m.s. values. In addition, voltages and currents in
the EMT simulation are in units of volts and amperes, while
those in the TS simulation are based on the per-unit system.
Therefore, all models have to be expanded and converted to
three-phase waveform-based ones, and appropriate nominal
voltages have to be given for all buses. Furthermore, an
appropriate winding connection, such as star-delta, star-stardelta, and so on, has to be assumed for each transformer, and
an appropriate grounding impedance has to be added to each
neutral point.
Using the developed EMT model of the West-10 benchmark
power system, simulations with various fault scenarios are
carried out. It is shown in this paper that the results obtained
by the EMT model agree well with those obtained by the
original TS model in most cases. It is also found that when the
dc components of fault currents are large, the results obtained
by those two simulation methods are different due to current
zero missing of circuit breakers. For the comparison with the
TS results, equivalents are used for the modeling of the
transmission lines, and they are then replaced with constantparameter line models for more practical simulation so as to
discuss the impact of line modeling.
II. IEEJ S W EST-10 B ENCHMARK P OWER
S YSTEM M ODEL
The 60-Hz power system in Japan interconnects several
electric power companies with 500-kV transmission lines and
extends more than 1,000 km in an east-west direction. The
system can roughly be considered a radial system. IEEJs
West-10 benchmark power system model, called the West-10
Benchmark System hereafter, approximates this radial system
with ten generators. Since only the summary of the West-10
Benchmark System is given here, ref. [1] should be consulted
for details.





$

)

-

*

*
























*

*

*

*

*





*



*

*


Fig. 1.

One-line diagram of the West-10 Benchmark System (IEEJs West-10 benchmark power system model).

TABLE II
G ENERATOR RATINGS .

TABLE I
G ENERATOR MODEL CONSTANTS .


GD[LVV\QFKURQRXVUHDFWDQFH[ >SX@
G




TD[LVV\QFKURQRXVUHDFWDQFH[T>SX@



GD[LVWUDQVLHQWUHDFWDQFH[G
>SX@



GD[LVVXEWUDQVLHQWUHDFWDQFH[G

>SX@


SHDNORDG
UDWLQJ >09$@

OLJKWORDG
UDWLQJ>09$@

*







***





TD[LVVXEWUDQVLHQWUHDFWDQFH[T

>SX@



*





GD[LVWUDQVLHQWWLPHFRQVWDQW7G
>V@



*





GD[LVVXEWUDQVLHQWWLPHFRQVWDQW7G

>V@



TD[LVVXEWUDQVLHQWWLPHFRQVWDQW7T

>V@



DUPDWXUHWLPHFRQVWDQW7D>V@



DUPDWXUHOHDNDJHUHDFWDQFH[O>SX@
J





JHQHUDWRU




($6

()6


($



All of the ten generators are represented by a dq-framebased synchronous generator model with one damper winding
for each of the d and the q frame [2]. The constants of the
generator models are set to the values shown in Table I. These
values are typical for large-scale thermal-plant generators, and
the reactance values are based on their own ratings shown
in Table II (see Section II-D for the peak-load and lightload conditions). The inertia constants of all generator models
are set to 7 s. Each generator model has an exciter (AVR:
automatic voltage regulator) model and a turbine-governor
model. The exciter model represents a rotating exciter, and
its control-block diagram is shown in Fig. 2 (a). The turbinegovernor model whose control-block diagram is shown in
Fig. 2 (b) represents a typical turbine governor used by thermal
and nuclear power plants. Each generator station is equipped
with a step-up transformer whose impedance is 0.14 p.u. based
on its own rating, and its tap ratio is unity.
B. Transmission Lines
The nominal voltage of all transmission lines is 500
kV. Since G1 and G10 equivalently represent interconnected
neighboring systems, they are considered as substations in the


V

()



V
V

Fig. 1 shows the one-line diagram of the West-10 Benchmark System. The system frequency is 60 Hz.
A. Generators


V


D


0


6J





V



PLQ





V


0


V


V



WXUELQH
RXWSXW




E
Fig. 2. Exciter and turbine-governor models. (a) Control-block diagram of
the rotating exciter model, where EA and EAS are the generator terminal
voltage and its initial value and EFS is the initial value of the field voltage.
(b) Control-block diagram of the turbine-governor model, where Sg is the
generator speed deviation. 65M and 77M are parameters related to governorfree operations and not used in the simulations in this paper.

discussion below. The transmission lines connecting between

substations are double-circuit and their length are 100 km.
The impedance of these lines is 0.0021 + j0.063 p.u. and a
half of their admittance is j0.122. Those connecting generator
stations to substations are also double-circuit. Except the one

G1

G10

P = 13.5
V = 1.03

3.5 + j0

3.5 + j0

3.5 + j0

3.5 + j0

5.5 +
j1.466

5.5 +
j1.439

5.5 +
j1.434

5.5 +
j1.437

12 +
j2.438

G2

G3

G4

3.5 + j0 5.25 + j0
5.5 +
j1.444

G5

5.5 +
j1.463

G6

28.3 +
j4.761

3.5 + j0
2.75 +
j0.823

G7

swing
node

5.5 +
j1.483

G8

G9

Based on
1,000 MVA
500 kV

P=9
P=9
P=9
P=9
P=9
P=9
P = 4.5
P=9
V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03

(a)
G1

G10

P = 6.75
V = 1.03

swing
node

1 + j0.01 1 + j0.01 1 + j0.01 1 + j0.01 1 + j0.01

2.6 +
j0.975

3.75 +
j1.397

G2

2.6 +
j0.913

G3

2.6 +
j0.903

G4

2.6 +
j0.913

2.6 +
j0.945

G5

G6

1.5 +
j0.015
2.2 +
j1.088

1 + j0.01
1.7 +
j0.555

G7

16.5 +
j2.822

2.6 +
j1.032

G8

G9

Based on
1,000 MVA
500 kV

P = 3.6 P = 3.6 P = 3.6 P = 3.6 P = 3.6 P = 3.6 P = 2.4 P = 3.6

V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03 V = 1.03

(b)
Fig. 3.

Power-flow conditions. (a) Peak (daytime) load condition. (b) Light (nighttime) load condition.

connecting Bus 18 to Bus 7, their lengths are 50 km, and

their impedance is 0.00105 + j0.0315 p.u. and a half of their
admittance is j0.061. The length of the transmission line
connecting Bus 18 to Bus 7 is 100 km, and its impedance
and a half of its admittance are 0.0021 + j0.063 p.u. and
j0.122 respectively. These line constants have been determined
by assuming a bundle of four TACSR-840-mm2 wires used
for each phase. All transmission lines are modeled by singlesection equivalents.
C. Loads
The models of the loads represent the following characteristics. When the voltage VL at the load is higher than or equal
to 0.7 p.u., the active power PL consumed by the load model
shows the constant-current characteristic given by
 

VL

1 + f
,
(1)
PL = P i
Vi
100
where Pi and Vi are initial active-power and load-voltage
values. A frequency dependence is represented with respect
to the frequency deviation f of the load voltage with the
parameter which is set to 3.33. When VL is smaller than
0.7 p.u., the active power shows the constant impedance
characteristic given by

 2 
VL

PL = Pi
1 + f
.
(2)
Vi
100
This equation also represents the same frequency dependence.
Regardless of the voltage value, the reactive power consumed
by the load model shows the the constant-impedance characteristic
 2
VL
QL = Qi
,
(3)
Vi

TABLE III
T RANSFORMER IMPEDANCE VALUES .


JHQHUDWRUVWDWLRQ
*

SHDNORDG

OLJKWORDG

M 

M 

***

M 

M 

*

M 

M 

M 

M 

*


where Qi is an initial reactive-power value.

D. Power-Flow Conditions and Transformer Impedances
As the power-flow condition from which a simulation is
started, the two cases, the peak (daytime) and the light (nighttime) load condition, are considered. For each case, the P V
and the P Q values are respectively given to the generators
and the loads as shown in Fig. 3.
The impedance values of the transformers in the generator
stations are shown in Table III in p.u. based on 1,000 MVA and
500 kV. The values vary with respect to power flow conditions.
III. EMT M ODELING
The simulations presented in this paper have been carried
out using the EMT analysis program XTAP [3]. It uses the
two-stage diagonally implicit Runge-Kutta (2S-DIRK) method
for the numerical integration of dynamic circuit elements such
as inductors and capacitors [4], [5]. The 2S-DIRK method
does not produce fictitious numerical oscillation when an
inductor current or a capacitor voltage is suddenly changed,
since it is a mathematically oscillation-free algorithm. For the

2S-DIRK

solution of nonlinear circuit equations, XTAP uses a robust and

efficient scheme which combines the conventional NewtonRaphson algorithm, its modified version named the biaxial
Newton-Raphson algorithm, and the Katzenelson method [6].
Using this scheme, XTAP obtains the solution at each time
step with a relatively small number of iterations. To start an
EMT simulation from a steady state, XTAP first calculates a
positive-sequence power-flow solution based on power-flow
conditions given by the user, then calculates a three-phase
steady-state solution based on the power-flow solution, and
finally starts an EMT simulation with initializing the dynamic
elements based on the preceding steady-state solution [7].
A. Generator Model

rotor
angle

d
Id

q-axis
equivalent

generator
mechanical
dynamics

q
Iq

Te
e.m.
torque

RK2

(a)

Jad(t)

(4)

Eda(t)

Jaq(t)

Edb(t)
b

Edc(t)
Jbd(t)

Jbq(t)
Eqa(t)

c
Jcd(t)

Eqb(t)

Jcq(t)

Eqc(t)
g

(b)

4
where k = 23 , a = , b = 2
3 , c = 3 , and is the
rotor angle given by the mechanical part of the model. The
current-controlled current sources that generate the abc-frame
currents are expressed by

Jad (t) = cos a Id (t) Jaq (t) = sin a Iq (t)

Jbd (t) = cos b Id (t) Jbq (t) = sin b Iq (t)
Jcd (t) = cos c Id (t) Jcq (t) = sin c Iq (t).

d
interb face
q
c

Fig. 4 (a) shows the framework of the generator model [8]

used in the EMT simulations in this paper. Since the generator
terminals in the abc frame are interfaced with the d- and the
q-axis equivalent circuit using the dependent sources shown in
Fig. 4 (b), there is no time delay in the interface between the
abc and the dq frame. The voltage-controlled voltage sources
that generate the dq-frame voltages are expressed by
Eda (t) = k cos a va (t) Eqa (t) = k sin a va (t)
Edb (t) = k cos b vb (t) Eqb (t) = k sin b vb (t)
Edc (t) = k cos c vc (t) Eqc (t) = k sin c vc (t),

d-axis
equivalent

(5)

The electromagnetic torque is obtained by the d- and qequivalent circuits and then passed to the generator mechanical
part as the input. The differential equations describing the
generator mechanical dynamics are solved using a secondorder explicit Runge-Kutta method (RK2) which gives the
solution at the present time step only from values at past time
steps. Thus, there is no time delay associated to pass the rotor
angle to the abcdq interface.
The nominal voltage of all generators are set to 22 kV,
and each of their neutral points is grounded through a 0.1 resistor representing the resistance of a generator-station
grounding mesh. The model used takes into account the
magnetic-flux saturation, but it is not represented in the West10 Benchmark System. The exciter model and the turbine
governor model shown in Fig. 2 are implemented as they are
using the control-block library of XTAP.
B. Transformer Model
The winding connections of all transformers are assumed
to be deltastar, where the primary side is delta and the
secondary is star. The nominal voltage of the primary side
is 22 kV, and it is connected to the generator. The secondary
side whose nominal voltage is 500 kV is connected to the
transmission network. The neutral point of the secondary

Fig. 4. Generator model used in the EMT simulations. (a) Model framework.
(b) abcdq interface utilizing dependent sources.

winding is grounded through a 0.1- resistor representing

a generator-station grounding mesh. Although more realistic
representations are possible, the leakage inductance is lumped
to the primary side, and the winding resistance, the magnetizing resistance, and the magnetizing inductance are ignored.
This is for comparison with TS results.
C. Line Model
Each transmission line is represented by three sets of a
single-section -equivalent circuit with positive-sequence line
constants. Again, this is for comparison with TS results.
For more realistic representation, the transmission lines are
represented by the constant-parameter line model [9], and its
impact is discussed in Section V.
D. Load Model
The load model used basically consists of star-connected
three-phase series RL branches. The resistance and the inductance value are controlled by the following algorithm. The
three-phase voltages of the load model are obtained and their
d- and q-axis components are calculated using the abc and
the dq transformation so as to obtain the r.m.s. voltage.
The frequency of the load voltage is obtained by a zerocrossing method. Using the r.m.s. voltage and the frequency,
the active power PL and the reactive power QL are calculated
using (1)(3). Then, the resistance and the inductance value of
the series RL branch are calculated so that it consumes one

EMT simulations of the West-10 Benchmark System have

been carried out for the following six cases. A three-lineto-ground (3LG) fault is considered at one of the points
designated by A, F, and J in Fig. 1, and the peak- and lightload conditions shown in Fig. 3 are considered as subcases for
each fault-point case. The 3LG fault is considered to occur on
one of the two circuits at t = 1 s, and the fault duration time
assumed is 70 ms.
Figs. 5, 6, and 7 show the simulation results when the 3LG
fault is considered at Point A, F, and J respectively. The rotor
angles of G1, G3, G5, G7, and G9 are shown considering
that of G10 as the phase-angle reference. The same simulation
cases have also been carried out by a TS simulation program,
and those results are superimposed using dashed lines in the
figures. For the TS simulations in this paper, CPAT (CRIEPIs
Power system Analysis Tools) which is widely used as the
de-facto standard program in Japan is used [2].
In the cases where the fault occurs at A and F, the EMT
simulation results agree well with the corresponding TS simulation results. In the cases of faults at J, however, the two

90
URWRUDQJOH>GHJ@

IV. S IMULATION R ESULTS

simulation methods do not give similar results. Especially in

the peak-load condition, the EMT result shows generator step

* *
*
*
*

90




WLPH>V@
D

*



(07

76

*

90

*

60
30

*

*




WLPH>V@
E



Fig. 6. EMT simulation results of the cases where a 3LG fault occurs at
F. The rotor angles of G1, G3, G5, G7, and G9 are shown. (a) Peak-load
condition. (b) Light-load condition. The TS simulation results for the same
cases are superimposed by dashed lines.

(07

600

200
0


360

76

400

200


WLPH>V@
D



WLPH>V@
D



270

(07

180

76
URWRUDQJOH>GHJ@

URWRUDQJOH>GHJ@

120

30


76

*
*

90
0

*
* *

90


150

URWRUDQJOH>GHJ@

URWRUDQJOH>GHJ@

(07

90

270

*
*

800

180

76

30

360
270

(07

* *
*

60

URWRUDQJOH>GHJ@

third of PL + jQL , and the calculated values are reflected to

the three-phase RL branches.
It is also possible to use three-phase parallel RL branches,
and the determination of the RL values may be even easier in
this case. However, the series RL branch is somewhat closer
to the physical structure of the power network, and thus, the
series branch is selected in this study.


WLPH>V@
E

180 *
*
90

Fig. 5. EMT simulation results of the cases where a 3LG fault occurs at
A. The rotor angles of G1, G3, G5, G7, and G9 are shown. (a) Peak-load
condition. (b) Light-load condition. The TS simulation results for the same
cases are superimposed by dashed lines.

76

*
*

0


(07

*



WLPH>V@
E



Fig. 7. EMT simulation results of the cases where a 3LG fault occurs at J. The
rotor angles of G1, G3, G5, G7, and G9 are shown. (a) Peak-load condition
(generators not identified since too crowded). (b) Light-load condition. The
TS simulation results for the same cases are superimposed by dashed lines.

fault current [kA]

300

14.5 m

200

100
0
b

100

11.2 m b'
PW

200
300
0.98

1.02

1.04 1.06
time [s]

1.08

1.1

200

11.6 m

a'

1.12
79.5 m

(a)

57.5 m

72.0 m

TACSR
410
6

12.0 m

43.0 m
50 cm

b
e = 300 m

100
0

Fig. 9. Wire arrangement of the 500-kV transmission lines assumed for

representation by the constant-parameter line model.

100
a
200
0.98

360

c
1

1.02

1.04 1.06
time [s]

1.08

1.1

1.12

(b)
Fig. 8. Fault-current waveforms in the cases where a 3LG fault occurs
at Point J. (a) Peak-load condition. (b) Light-load condition. Current zero
missing is marked by dashed circles.

URWRUDQJOH>GHJ@

fault current [kA]

GW
IACSR
150

c'

270
180

V. I MPACT OF L INE M ODELING

To investigate the impact of line modeling, all transmission
lines in the West-10 Benchmark System are now represented
by the constant-parameter line model [9]. It takes into account
traveling waves and coupling among wires. Since the wire
arrangement of the 500-kV transmission lines is not given in
the West-10 Benchmark System, the wire arrangement shown
in Fig. 9 is assumed to calculate the line constants.
This wire arrangement can be considered a typical one for
500-kV transmission lines in Japan. Since 500-kV transmission lines in Japan are generally not transposed, the line constants were calculated as an untransposed line. To obtain the
corresponding TS simulation results, each of the transmission
lines is represented by ten sections of cascaded equivalents
to gain accuracy. Since the line constants used in the EMT
simulations and those used in the TS simulations are not
exactly the same, a slight difference is observed in the initial
power-flow solutions in the results presented below.

76
* *
*

90
0

*
*

90


WLPH>V@
D

360
URWRUDQJOH>GHJ@

out, while the TS result does not. To investigate the cause

of this discrepancy, the fault-current waveforms of the EMT
results in these cases are shown in Fig. 8. In EMT simulations,
a circuit breaker cannot turn off if its current does not cross
zero due to a dc component. This phenomenon is called current
zero missing. In Fig. 8, current zero missing occurs as marked
by the dashed circles, since the currents of phase b do not cross
zero at around t = 1.07 s. Due to the current zero missing,
the fault duration time of phase b becomes longer than 70
ms in these EMT simulation cases, and this gives the results
different from those of the corresponding TS simulations.

(07

270

(07

180

76

*
*

90
0

*
* *

90



WLPH>V@
E

Fig. 10. EMT simulation results with the constant-parameter line model.
Cases where a 3LG fault occurs at A. The rotor angles of G1, G3, G5, G7,
and G9 are shown. (a) Peak-load condition. (b) Light-load condition. The TS
simulation results for the same cases are superimposed by dashed lines.

Figs. 10, 11, and 12 show the simulation results when the
3LG fault mentioned in Section IV occurs at A, F, and J,
shown in Fig. 1, respectively. Although a slight difference is
observed in the initial power flow, the EMT simulation results
agree well with the corresponding TS simulation results in the
cases where the fault occurs at A and F. When the fault occurs
at J in the peak-load condition, one of the generators steps out
in the EMT result, while there is no generator stepping out in
the TS result. Fig. 13 shows the fault-current waveforms of
the EMT results when the fault occurs at J, and current zero
missing is observed as marked by the dashed circle in the
peak-load condition due to a large dc component. Apparently,

300

*

(07

* *

60

76

fault current [kA]

URWRUDQJOH>GHJ@

90

30

0


*




WLPH>V@
D

100
0
100



200
300
0.98

*

200

c
1

1.02

1.04 1.06
time [s]

1.08

1.1

1.12

1.08

1.1

1.12

(a)

150

90

*

60

*

30

*

*

30


(07

76

fault current [kA]

URWRUDQJOH>GHJ@

200

*

120



WLPH>V@
E

URWRUDQJOH>GHJ@

800
(07

76

400
200
0

200




WLPH>V@
D



URWRUDQJOH>GHJ@

270

100
a
c
1

1.02

1.04 1.06
time [s]

(b)
Fig. 13. Fault-current waveforms in the cases where a 3LG fault occurs
at Point J, obtained with the constant-parameter line model. (a) Peak-load
condition. (b) Light-load condition. Current zero missing is marked by a
dashed circle.

the discrepancy comes from this current zero missing, and this
discussion is similar to that in Section IV.
It may be concluded from the simulation results shown
above that line modeling does not have a significant impact on
simulation results of system-wide studies, when balanced fault
conditions such as 3LG are applied. Further studies should
be undertaken to investigate the impact for unbalanced fault
conditions. Use of a frequency-dependent line model for more
realistic transmission-line representation may be an interesting
future study.
VI. C ONCLUSION

180 *
*
90

(07

*

76

*
*

0


0.98



Fig. 11. EMT simulation results with the constant-parameter line model.
Cases where a 3LG fault occurs at F. The rotor angles of G1, G3, G5, G7,
and G9 are shown. (a) Peak-load condition. (b) Light-load condition. The TS
simulation results for the same cases are superimposed by dashed lines.

600

100

200



WLPH>V@
E



Fig. 12. EMT simulation results with the constant-parameter line model.
Cases where a 3LG fault occurs at J. The rotor angles of G1, G3, G5, G7,
and G9 are shown. (a) Peak-load condition (generators not identified since too
crowded). (b) Light-load condition. The TS simulation results for the same
cases are superimposed by dashed lines.

In this paper, the West-10 benchmark power system model

prepared by the Institute of Electrical Engineers of Japan
(IEEJ) has been expanded and converted to an EMT model
for the implementation in electromagnetic transient (EMT)
simulation programs. Using the developed EMT model of the
West-10 benchmark power system, simulations with various
fault scenarios have been carried out. In most cases, the results
obtained by the EMT model agree well with those obtained by
the original transient stability (TS) model. It has been found
that when the dc components of fault currents are large, the
results obtained by those two simulation methods are different
due to current zero missing of circuit breakers. It should be
noted that we cannot conclude which method is correct, since a
circuit breaker may be able to interrupt a residual current fairly
close to zero (although current zero is missed). The impact
of line modeling for balanced fault conditions has also been
studied.

R EFERENCES
[1] Committee for Standardization of Benchmark Power System Models,
Standard benchmark power system models, IEEJ Technical Report,
no. 754, 1999 (title translated into English by the authors).
[2] Power System Stability Study Group, Integrated analysis software for
bulk power system stability, CRIEPI Report, no. T14, Apr. 1990.
[3] T. Noda, International and domestic development trends of electromagnetic transient analysis programs for power systems, IEEJ Trans., Power
and Energy, vol. 131, no. 11, pp. 872875, 2011.
[4] T. Noda, K. Takenaka, and T. Inoue, Numerical integration by the
2-stage diagonally implicit Runge-Kutta method for electromagnetic
transient simulations, IEEE Trans., Power Delivery, vol. 24, no. 1,
pp. 390399, Jan. 2009.
[5] T. Noda, T. Kikuma, and R. Yonezawa, Supplementary techniques for
2S-DIRK-based EMT simulations Electric Power Systems Research,
vol. 115, pp.8793, Oct. 2014.
[6] T. Noda and T. Kikuma, A robust and efficient iterative scheme for the
EMT simulations of nonlinear circuits, IEEE Trans., Power Delivery,
vol. 26, no. 2, pp. 10301038, Apr. 2011.
[7] T. Noda and K. Takenaka, A practical steady-state initialization method
for electromagnetic transient simulations, Proc., IPST 2011, paper # 99,
Delft, The Netherlands, 2011.
[8] O. Sakamoto, T. Nakajima, H. Takizawa, T. Noda, and K. Takenaka,
Development of a synchronous machine model considering simplified
saturation of the interlinkage magnetic flux behind the armature winding leakage inductance for XTAP, Proc., IEEJ Technical Meeting on
Rotating Machinery, paper # RM-13-118, Sendai, Japan, 2013.
[9] H. W. Dommel, Digital computer solution of electromagnetic transients
in single- and multi-phase networks, IEEE Trans., Power Apparatus and
Systems, vol. PAS88, no. 4, pp. 388399, Apr. 1969.