Documente Academic
Documente Profesional
Documente Cultură
Applied
ATP-EMTP
to
Highly-sophisticated
electric power
systems
Currents in cycloconverter
Eiichi Haginomori
August 2003
2
Preface
For power system engineers engaged in sophisticated electric power systems, such as with very
high power generations, with high capacity of EHV/UHV transmission lines, high density power
flows, combined with IPP systems, etc., more and more understanding in detail the various tran-
sient and complicated phenomena in the systems are obliged. The time regions of these are from
nanoseconds to several seconds, or even minutes.
As for the education of such engineers the best way is thought to be experiments and experiences
in actual power systems, any of which seems to be very difficult and costly in these fields. Also long
time interval of education is necessary.
Recently ATP-EMTP has made a significant development and became a more and more useful and
powerful tool to analyse various transients in power systems and, also, power apparatuses. The
program covers almost all of the transients and is applicable to PC, by which virtual experi-
ment/experience seems to be realistic. Therefore, the efficient usage of the program is beneficial for
cost and time saving.
In this text-book various kinds of transients together with the analyses with ATP-EMTP are written.
Also the principles of the phenomena and the usages of ATP-EMTP are involved as much. Data
files analysing such phenomena are attached.
Care should be taken that, beforehand, primary stage study of ATP-EMTP is strongly recom-
mended before reading this text book.
Contents
1. Transient currents in power systems
1.1 Short circuit current
1.2 Transformer inrush magnetizing current
1.3 Transformer magnetizing current under geo-magnetic storm condition
1.4 Inrush current in capacitive circuit
4. Transformer
Single-phase two winding transformer
Single-phase three winding transformer
Three-phase one core (three legs or five legs) transformer
A few examples
Appendix 4.1 Response to fast/very fast transient voltage (VFT)
7. Transient stability
7.1 Classic analysing method -------- Equi-area method
7.2 Time domain analysis ------- ATP-EMTP
Field excitation control
Back swing phenomena
Appendix 7.1 Modelling by Universal Machine
Current in inductance is to be continuous, therefore irrespective of the point on wave short circuiting
timing, the current starts from zero. So for compensating the instantaneous a.c. value to zero for the
short-circuiting time, d.c. component exists. Appling ATP-EMTP, an example is shown in Fig. 1.2,
also see the attached data file, where each current starts from zero value and equal a.c. component
waves irrespective of the d.c. and short circuiting timings are shown.
In practical system circuits, where the circuits are mostly three phases, both positive/negative and
zero sequence parameters are to be considered in three phase circuits. Also for practical systems,
discharging currents from parallel capacitances, such as transmission lines, cables, or shunt ca-
pacitor banks, are occasionally not insignificant. Such discharging currents have components of
several hundred Hz and
mostly decay after a few
tens ms from the
short-circuiting time. But
in special cases, they
may not be negligible
after several tens ms
from the short circuit ini-
tiation when the currents
are to be interrupted by
circuit breakers. A typical
example, corresponding
to an extremely high
density network near a
megalopolis, is shown in
Fig. 1.3, which is a case
of EHV substation
bus-bar is short circuited,
Fig. 1.2 Short-circuit current in Fig. 1.1
where extremely high
capacitances such as
EHV cables via certain
length of overhead line (20km) and high capacity of shunt capacitors in the tertiary winding side of
the transformer are connected. Care should be taken in such calculation regarding the damping of
the transient current frequency by capacitance discharging current. The frequency of the transient
is in the order of several hundreds Hz, so the losses in transformers, transmission lines, cables, etc.
are to be based on that frequency range.
The calculation was done for a 550 /300 kV sub-station, the capacity of which transformer is (in
2
current. Therefore the current goes gradually to symmetrical one, down to less than 1 A. For correct
calculation as for the damping of the current, care is to be paid for the correct resistance value(s) in
the circuit.
Notes:
- In the attached data calculating Fig. 1.5, the magnetizing characteristic is modelled by only 3
segments for one polarity. If correct value(s) for 10 ---- several 1000 A current value(s) is wished,
more accurate modelling for the current range may be necessary.
- Three non-linear inductance menus are available:
Type 98 Pseudo non-linear reactor : Most simple and useful for general usage but initial (residual)
flux is not applicable. Type 93 True non-linear inductance : Initial flux is applicable. Calculation is
only a little bit slower. Care should be paid when applying initial flux, current dose not starts from
zero but a certain value relevant to the flux value on the magnetizing curve.
- Type 96 Pseudo-non-linear hysteretic reactor : Care should be paid when up going and down
going, the current/flux locus traces the same line for each, i.e., the width between the two lines is
constant.
The second equation is the integration of the first one. The third one is just modification of the
second one, which shows the average current value is just the DC current value applied.
Fig. 1.7 Circuit diagram for geomagnetic Fig. 1.8 ATP-EMTP calculation of transformer
storm condition magnetizing current under geomagnetic
storm
ATP-EMTP calculation result is shown in Fig. 1.8, also see attached data file for the circuit pa-
rameter details. In the calculation to attain shorter time interval to the steady state, the series re-
sistor values are intentionally enhanced. Otherwise, the time to steady state is to be several ten
seconds for the actual circuit parameters. The calculation was done applying both AC and DC
voltages to the transformer without initial residual flux.
Note :
- Geomagnetic storm condition lasts several ten minutes, while the thermal time constant of a
transformer is in the order of one --- several hours. Higher current lasting less than one minute
such as inrush current is of no importance as for the thermal phenomena. Electrical time constant
around a transformer is far less than one minute. Therefore, electrically steady state phenomena
are of importance regarding geomagnetic storm.
- In the attached data file calculating the phenomena, introducing initially residual flux and/or an
other timing of source voltage, different current wave shape is obtained only for the initial time
iterval. After some time interval, current reaches to the same steady state value.
Capacitive circuits such as high capacity shunt capacitor banks or EHV under ground cable sys-
tems, when closing by relevant switching facilities, i.e. circuit breakers, may create very high inrush
current up to the order of the short circuit one. The frequency of the transient is in the order of a few
hundred Hz ---- several kHz. The transient last generally short time interval, so the contact con-
sumption of the relevant switching facility (circuit breaker) is the most concern. Also facilities are
influenced electro magnetic forces. Some examples of circuit diagrams are shown in Fig. 1.9 in
single phase ones. Most actual circuits are in three phases, so in calculating three phases model-
ling is necessary for most cases.
a): Single capacitor bank circuit in most simplified representation. The highest current is easily
calculated by V (voltage), C (capacitance) and L (total series inductance). R (series resistance)
only influences on the damping of the transient current.
b): Ditto, but with series reactor intentionally added. The practice is very common in Japan to sup-
press the current and also harmonic current component.: So called back to back capacitor bank
circuit, where, if the series reactance values are low, very high inrush current flow. Special care
should be paid for circuit breaker application.
c): So called back to back shunt capacitor bank arrangement. When a capacitor bank is switched
on while another one is previously energized, true inrush current may flow. The series reac-
tances are to be carefully calculated.
d): In high capacity of sub-stations, capacitor banks are installed in the transformer’s tertiary wind-
ing circuits. Thus some amount of series reactances are automatically introduced.
e): In EHV under ground cable systems where, especially, plural circuits are connected to the bus
5
EMTP calculations themselves are thought to be not so, so complicated, so no example is shown
here. Care should be taken for also the damping elements (resistances) in the circuits. The values
depend on the relevant (the inrush current) frequencies, such as transmission lines, cables, trans-
formers, etc. It may be necessary to preliminarily calculate the inrush current frequency, and to
re-calculate each damping element. The followings are general ideas for the damping, which might
be of your help unless otherwise obtained. Also see as for details in the following relevant sections.
* Overhead transmission line and underground cable: Parameters are to be calculated based
on the relevant transient frequency. For underground cable, dielectric loss (tan), which can be
neglected in power frequency, might be necessary to be counted in. See section 3.
* Capacitor bank:: Appropriate dielectric loss in the relevant frequency range is to be considered.
The loss of the series reactor, if any, is in the order of 0.05% of the capacitor bank capacity in
power frequency. About 60% of which is copper loss which can be represented by a constant
value of series resistor irrespective the frequency. Iron loss (ca. 25% in power frequency) is
represented by a constant value of resistor connected in parallel due the fact that the loss de-
pends on the 2nd power of the voltage irrespective of the frequency. Stray loss (ca 15% in power
frequency) is proportional to 2nd power of current and 1.5th power of frequency.
* Power transformer: Typical losses of a high capacity of transformer is, ------ Iron loss is ca.
0.03% of the capacity, which can be represented by a constant resistor connected in parallel.
Load dependent loss is 0.15 --- 0.2%, 85% of which is copper loss and 15% is stray loss. Like
capacitor bank’s series reactor, the relevant losses are applied.
For circuit breakers or other switching facilities, transient voltages just after the current interruptions are
of great concern with successful current breakings, as the phenomena relate to the competition be-
tween the insulation recoveries and transient voltages across the contacts. The transient voltage is
called as “Transient recovery voltage, (TRV).” For some typical cases in power systems, calculation
principles are shown in this chapter.
For circuit breakers short circuit current breaking is the most important performance to fulfil. Just after
the interruption of high current the insulation across the contacts is to withstand against TRV of rela-
tively or even very rapid recovery. Applying EMTP, TRVs can be straightforwardly calculated. Never-
theless, for calculating TRVs in large power systems, simplified and effective calculation processes are
wished. For the purpose, to bear current injection principle in mind is strongly recommended. The
principle is shown in Fig. 2.1.
As shown in the upper figure, the inverse polarity of current ( -I ) is superimposed to the originally
flowing short circuit current ( I ) from
the switch terminals. Then the total
current is to be zero, corresponding to
current interruption. The phenomenon
is represented by (a), (b) and (c), i.e.
the total phenomenon is (a), which can
be replaced by (b) + (c) by superim-
posing principle. (b) corresponds to
short circuited as up to the time, so no
TRV appears. TRV is produced only in
(c). Therefore, we are to calculate for
(c) to obtain TRV. The feature of this
method is :
- As for source, only the breaking
current is to be considered and
some sources in the system are
eliminated.
- Only the circuit parameters seen
from the switching facility terminals
are to be considered. Therefore,
parameters remote from the ter-
minals seem to be not so, so im-
portant.
- In (c), any initial condition is ex-
cluded, so we can apply only dead
circuit and injecting current to cal-
culate TRV, thus simplified and
easier consideration is applicable.
- In the calculation, ramp current in-
stead of sinusoidal one is mostly
Fig. 2.1 Current injection principle applicable due to relatively short
In current breaking time interval of TRV duration time
interval, compared to a loop of
power frequency current.
As a whole, the method suggests possibility of simplification in TRV calculation. Accurate parameters
are necessary only circuit locations close to the switching device concerned.
2
Great care is to be taken the principle is applicable only for TRV across the switching facility terminals.
For other variables, e.g. voltages to ground, the original process ((a) in Fig. 2.1)is to be taken.
Actual complicated systems are com-
posed of simple elements, so firstly to
study responses by simple circuit ele-
ments seems, hopefully, to be useful and
beneficial. In Fig. 2.2 some circuit ele-
ments frequently applied to represent ac-
tual systems are shown.
(6) ---- (9) are often used in practical short
circuit test plant circuits. In Fig. 2.3
ATP-EMTP calculation results injecting
ramp currents to the elements (1) ---- (9)
are shown. As for actual numerical values
of the parameters applied, see the at-
tached data file.
The followings are noted :
- In digital calculations finite values are
Fig. 2.2 Some circuit elements to represent systems to be applied, even for the initial part
of the ramp current, yielding aston-
ishing results for cases (2) and (4).
Small capacitances are to be con-
nected in parallel to the inductances.
- (6) and (7), also (8) and (9) yield simi-
lar results respectively, see Fig.
2.3.(b). But in the enlargements of the
very initial parts, significant differ-
ences are found in Fig. 2.3 (c). The
differences are often of certain im-
(a) (1) ----- (5) in Fig. 2.2 portance in actual short circuit test
circuits. The differences are origi-
nated by capacitances directly con-
nected to the switching facility termi-
nals.
- Hand calculations (analytical) are not
so, so difficult, so the reader is
strongly recommended to try, at least
once, for better understanding these
phenomena.
(b) (6) ---- (9) in Fig. 2.2 Then, actual application to calculate
power system TRVs will be shown next.
Let’s introduce Fig. 2.4 (single-phase
circuit diagram) as an example. For three
phase circuits, basic principle will be
shown later.
In Fig. 2.4, a 300kV system around a
sub-station is shown. Transformer’s an-
other side system is simply represented
(c) Enlargement of the initial part in (b) by voltage source via inductance equiva-
lent to transformer and the system short
Fig. 2.3 Voltages by injecting ramp currents to circuit circuit reactance. Applying what are men-
elements in Fig. 2.2 tioned before, transmission lines near the
circuit breaker are relatively accurately
3
represented by distrib-
uted parameter lines,
while; remote systems
are simply represented
by lump elements such
as capacitors and in-
ductors. For the details
of the parameters ap-
plied, see the attached
data file. The short circuit
capacity at the bus bar is
approx. 50kA, repre-
senting a sub-station in a
relatively high capacity
of system. “X” is the
connection bus bar re-
Fig. 2.4 One line diagram around 300 kV sub-station lating to ITRV, to be
to calculate TRVs in short circuit clearings shown later.
Fault points are F1 ----
F4. F1 is so called ter-
minal fault, and F2 ---- F4
are line faults. Especially,
fault at F2 is called short
line fault (SLF), which,
due to relatively high
breaking current and very
high rate of rise of TRV,
is of importance for cer-
tain type of circuit-
breakers.
EMTP calculation result
as for F3 fault current
breaking in Fig. 2.4 is
shown in Fig. 2.5. In (a)
(a) Overall voltages and current overall phenomena are
shown. Before current
interruption, a portion of
voltage exists at the bus
corresponding to the dis-
tribution along the line
(10km). After interruption,
bus side voltage recovers
to the source value with
some transients. Line
side one goes to zero
also with some transients.
The details are clear in
(b) as the zooming ex-
pression around the cur-
rent interruption.
Note:
(b) Enlargement around current interruption
- Breaking means
overall phenomena
Fig. 2.5 Fault current breaking, fault at F3 in Fig. 2.4
4
including initiation of movement of circuit- breaker, contact separation, arcing, quenching of arc and
current interruption, TRV appearing, withstanding against TRV and power frequency recovery
voltage. While interruption means just end of arcing current.
Applying current injection principle before mentioned, TRVs in line fault breakings are understood as:
- Current injections from the circuit breaker are to be --- one current to the bus direction and the other
of opposite polarity to the line direction, value of which correspond to F3 fault current.
- Line side TRV follows the principle (8) in Fig. 2.2 and Fig. 2.3
- Bus side TRV, at the first step before the reflection waves coming, follows (3) in Fig. 2.2 and 2.3, as
overall surge impedance Z, which is equivalent to resistance Z before the reflection coming, and
Inductance corresponding to Transformer, etc. are connected in parallel. Then afterwards, arrivings
of reflection waves in some lines, overall voltage change is such like as (5), (6) or (7) in Fig. 2.2 and
2.3. These phenomena are well explained in ANSI C37.06.
- TRV across the terminals of the circuit breaker is the difference of the two TRVs, as shown in Fig.
2.5 (b).
- Be aware, as written before, current injection is valid for across terminals TRV. By each side current
injection, only voltage change appears.
Applying current injection principle, for example, dv/dt (rate of rise of recovery voltage, RRRV) and re-
flection time are easily obtained as [di/dt times surge impedance] and [line length divided travelling
speed respectively, thus overall conception of TRV is easily obtained. “X” in Fig. 2.4 corresponds to a
connection line between the circuit breaker and the bus, the length of which is in the order of several
ten meters. The connection line yields TRV similar to the line side one (SLF), but due to shorter length,
of lower amplitude. This is called Initial Transient Recovery Voltage (ITRV). This may be of importance
for certain type of circuit breakers, especially breaking higher current such as F1 or F2 fault in Fig. 2.4.
A distributed parameter line of the relevant length models the connection line.
For calculation TRV introducing very short connection line by EMTP, very short step time is required by
EMTP. In EMTP step time of calculation shall be shorter than the minimum travelling time of the dis-
tributed parameter line in the relevant circuit. Therefore, huge number of steps is necessary, as usually
several ten ms of calculation time interval is necessary for calculating breaking phenomena, mainly due
to initialisation technique in EMTP. On the other hands, time interval of ITRV concerned is very short,
such as, several microseconds. Introducing current injection principle in also EMTP calculation, effi-
cient calculation is possible. An example is shown in the attached data file, where 50m of connection
line and 0.1 microsecond of step time are introduced, while the total calculation time interval is 20 mi-
croseconds.
Three-phase circuit
Like a single-phase circuit, current injection principle is applicable to also a three-phase circuit in a
power system. The main concept of current injection in TRV calculation is:
TRV = [Injection current] times [Impedance looked through circuit breaker terminal]
For three phase circuits, the following equations are introduced. :
where Z0, Z1, and Z2 are respective sequence impedances looked through circuit breaker terminals,
and e0, e1 and e2 are voltages appearing across the terminals of the circuit breaker. Notations u, v, and
w relate to phases. In most cases in power transmission systems, Z1 = Z2.
Such equations, based on symmetrical component principle, were originally introduced for phenomena
in power frequency domain. But introducing Fourier series spectrums for some voltage/current wave
shapes, equations are thought to be valid for any voltage/current wave shape including transient one.
5
Fist example is to calculate first pole to clear impedance for three-phase fault. In the case, assuming
phase “u” is the first pole to clear, then, ev = ew = 0. From the equations shown before, :
For phase “u”, (eu / iu) is thought to be the equivalent impedance for first pole to clear. From the equa-
tion above, :
Likewise for second and third pole to clear, the followings are introduced respectively. :
Also for three phase circuits, TRVs are conceptually considered as products of injection currents and
equivalent impedances. Therefore from these impedance values, TRVs in three phase circuits could be
guessed, at least for relative values or qualitatively. For quantitatively accurate values, of cause, EMTP
calculations are inevitable.
Note:
- Modelings of power system elements such as transformers, overhead transmission lines, under
ground cables, etc. will be explained in the following chapters. Care should be taken that the models
depend on the frequency of the relevant phenomena. The best way is models for power frequency
are applied up to the current interruption, and ones for the frequency of the transient phenomena are
applied for the following phenomena. In actual cases compromise is to be necessary.
By re-strike, significant over voltage and greate shock due to the impulse discharging current are cre-
ated in the circuit. So, for modern sophisticated power systems with reduced insulation revel, re-strike
free is an earnest requirement.
In Fig. 2.7, 550kV no-load overhead line’s ca-
pacitive charging current breaking is shown in
simplified manner. The line is represented in
symmetrically transposed condition, 150km of
length. The source side is much simplified; still
general trend is well represented. Details are
shown in the attached data file and a).
b) shows voltage changes in normal breaking,
i.e. currents in three phases are interrupted in
order at each current zero. c) shows delaying of
current interruption in the second pole to clear.
a) System layout Due to the electro-static coupling, the first
pole’s line side voltage is much influenced, so
the recovery voltage of the pole is enhanced
much. If the scattering of the contact separation
timing is more than one 6th of one cycle time
interval (2.7ms for 60Hz and 3.3ms for 50Hz),
such possibility exists.
More accurate line and system modelling will be
explained in the following chapters. In most
overhead transmission line systems, so called
“Rapid auto-re-closing” is applied. In the sound
b) Normal breaking phase during the operation, the circuit pole may
close against the residual voltage of the inverse
polarity of the source voltage. The most severe
case’s result of circuit diagram in Fig. 2.7 is
shown in Fig. 2.8 a), where each pole closed at
each maximum voltage timing. The highest
over voltage at line end terminal is approx. 4 p.u.
of the system voltage. Pre-insertion resistor’s
effect is significant as shown in b), where 1000
ohm of resistors are inserted approx. 10ms in
three poles. For details of the system and op-
b) Delaying in 2nd pole to clear eration sequence parameters, see the attached
Fig. 2.7 No-load overhead line data files.
Capacitive current breaking
Note :
- In calculating re-closing over voltages, accurate transmission line modelling is necessitated due to
the wide range of frequency voltage components included. Damping of the line that is dominant for
over voltage value is dependent on the frequency. See the following chapters.
- In the case above shown solidly earthed neutral source circuit is applied. For non-solidly earthed
conditions, some examples will be shown in the following as mainly for a cable system.
ing.
Note:
- As relatively high frequency of oscillation is created by the series connected reactor, in calculation by
ATP-EMTP, sufficiently low value of step time is to be used.
Inductive current means shunt reactor (SHR), no-load transformer magnetizing or stalled motor ener-
gizing current. Due to low current value, the interruption itself is of little problem. While breaking by
usual circuit breakers such as air blast, SF6 or vacuum ones, the current tends to be chopped (forced
interruption) before its prospective (natural) current zero.
Note:
- Physical chopping phenomena by circuit breaker arc with negative v-i characteristic in conjunction
with circuit parameters will be explained in the following chapter.
Fortunately, in ATP-EMTP, current chopping (forced current interruption before current zero) is easily
introduced by time controlled usual switch. As an example, 300kV, 150MVA shunt reactor current
breaking is explained in Fig. 2.11. “a)” shows one phase of the circuit diagram in simplified modelling.
The connection bus inductances are to be introduced adjacent to the circuit breaker. “b)” shows
voltage changes around current interruption.
Note:
- For details of the circuit parameters in a), see the attached data file, where, for the purpose of
calculation stabilizing, several additional elements such as series connected resistors are intro-
duced. Also, “chopping --- re-ignition --- re-interruption” are represented by three switches, which
shall not directly be connected forming a ring. Small resistors are to be introduced in between.
At t1, the current is interrupted with chopping (by 5A). When chopping, as for the energy in the reactor
(magnetic) and capacitor connected in parallel, the next equation is introduced:
where: V: Maximum voltage across reactor terminals after chopping, when all energy is transferred to
C
ic: Chopped current revel
V0: Source voltage peak (approx. equal to the voltage at the chopping)
L: Reactor’s inductance
9
C: Reactor’s capacitance
Then the maximum voltages across the reactor terminals and circuit breaker are easily calculated. In
Fig. 2.11 b), after t1, the SHR terminal voltage goes to the maximum, and then goes down with the
across circuit breaker voltage recovers (B-SHR).At t2, the circuit breaker re-ignites and very high fre-
quency of voltage change at the SHR terminal appears together with high frequency and amplitude (up
to a few thousand A) of re-ignition current flowing. The current is re-interrupted after approx. 0.05ms
and re-establishment of TRV (B-SHR) appears. What is to be noted, after the second interruption, due
to higher trapped magnetic energy in the SHR winding, the voltage recovery is steeper than the first
one. So, also the second re-ignition might occur. Such is called as “Multiple re-ignitions” which may
corresponds to extremely severe over voltage condition to the reactor insulation.
Note:
- The second current interruption mostly occurs at current zero of the circuit breaker, where the
current is composed with initially very high but then mostly damped re-ignition high frequency cur-
rent through capacitances adjacent to the circuit breaker (so called second parallel oscillation cir-
cuit) and combined with the AC current in the SHR winding.
- What is most serious as in inductive current breaking for shunt reactor is, very rapid change of
voltage at winding terminals by re-ignition. Voltage stress of the winding, especially at the entrance
part, is generally very severe by high frequency of voltage stress. Some will be explained in the
following chapter.
where
tL: time to peak without capacitance
ω: angular frequency of the breaking current
I : breaking current peak
Z: surge impedance
S: Laplace operator
The equation is valid for 0 δ t δ 2tL. If the TRV for t > 2tL is required, the equation (1) is to be replaced by the
following:
TRV ( s ) = ωIZ
s2
(1 − 2e − t L s + 2e −2t L s + ............... ) (1a )
−t L s
In order to introduce damping of the wave, the term 2e in equation (1) should be replaced
−tL s
by, 2ke , where k < 1,0.
The TRV can be represented by the product of the breaking (= injection) current and the impedance, also in
the Laplace domain. The injection current in the Laplace domain can be approximated such as (due to very
short time interval concerned):
ωI
s2 (corresponding to the current = ωIt in the time domain)
Then the impedance of the distributed parameter line in Laplace domain is: (for t < 2tL)
The capacitance value includes both the lumped capacitance at the circuit-breaker terminal side producing
the inherent tdL of the line and the additional capacitance, if any.
Connecting the two impedances represented by (2) and (3) in parallel, the following equation is obtained for
the total impedance in the Laplace domain:
(4)
where t dL = ZC
tdL is also applicable for conditions with additional parallel capacitances.
The product of the injection current ( ωI ) and the impedance (4) is the TRV with parallel capacitance in
s2
the Laplace domain:
11
The second part of the equation (5) is valid for tL δ t δ 2tL only.
By reversal Laplace transformation process, SLF TRV with parallel capacitance in time domain is calculated
as follows:
For 0 δ t δ tL:
for tL δ t δ 2tL
(7)
with t' = t - tL
Using equations (6) and (7), the correct wave shapes of SLF TRVs with line inherent tdL and for conditions
with additional parallel capacitance can be calculated.
For cases t > 2tL, (1a) instead of (1) should be applied. When damping is introduced, 2ke − t L s should be
−tL s
used instead of 2e in equation (1) as mentioned before. The total calculation process is then slightly
modified.
For every case, with or without parallel capacitance, the peak value of the TRV is quasi equal to ωIZt L , i.e.
no significant damping to the peak value is introduced. Dividing equations (6) and (7) by ωIZt L , gives the
following equations.
12
The TRV wave shape given by equations (6a) and (7a) can be normalised such that the peak value is unity
and time unit is in tdL. The parameter is tL/tdL. Fig. 2A.1 shows the results of a calculation for tL/tdL =1.0 --- 15.
Multiplying the Y-axis value by ωIZt L and X-axis value by tdL, the actual wave shape is obtained. The peak
Appendix 2.2 4-armed shunt reactor for suppressing secondary arc in single pole
rapid re-closing
As the first step of studying switching phenomena in systems with 4-armed shunt reactors, math-
ematic study seems to be beneficial to grasp the outline. In single pole rapid re-closing, where only
the faulted phase of a transmission line is opened, the faulting arc is to quench during the re-closing
time interval. By electro static coupling with the sound phases, a certain level of arc current tends to
continue without quenching. As higher the system voltage is and as longer the transmission line is,
the tendency increases.
For eliminating the arc current (secondary arc current) aiming successful re-closing, 4-armed shunt
reactor where the neutral is earthed by means of another reactor is applicable. Fig.2A.2 shows the
circuit layout.
“a)” shows system layout during one phase line to ground faulting, where both ends of the phase are
open. Secondary arc may exist. “b)” shows the equivalent circuit at the faulting point, where, assuming
voltages along the phase v and w lines are quasi uniform, voltages are applied from the point, instead
of both ends, i.e. eu=0 (faulting), ev and ew. “iu” is the secondary arc current. Z0, Z1 and Z2 are sequence
component reactances of the line section (capacitances and inductances of 4-armed shunt reactor
shown in “c)” connected in parallel). The following equations are obtained. :
Except for rotating machines, in transmission systems, Z1 = Z2. In a transmission line with 4-armed
shunt reactor, parameters other than neutral reactor’s are fixed by the relevant system condition. So,
adjusting the neutral reactor reactance value, we can have:
Z0 = Z1 = Z2
Introducing this condition, then we can have iu = 0 applying the above shown equations, i.e. the sec-
ondary arc current can be suppressed.
Note:
- During switching of such transmission line, due to the non-linearity of the reactors as usually iron
cores are used, certain value of transient voltages appear at the neutral point and some insulation
failures have been reported. For sophisticated insulation design especially around the neutral point,
accurate analysis introducing every detailed parameters of the system including the non-linear
characteristics of iron cores is recommended.
14
Table 3.1 Overhead transmission line calculated PCH files (three kinds)
For transposed line, only zero sequence and positive sequence parameters are given. Mode transla-
tion is fixed, so no matrix is necessary. The first and second lines correspond to zero and posi-
tive/negative sequence parameters respectively. For transmission line positive sequence parameters is
identical to negative ones.
As an example typically showing the differences between these three models, power frequency COS
wave shape voltage, the amplitude of which is 100V, is applied from one side of each model transmis-
sion line and the other end voltage is calculated. The initial step of the voltage involves high frequency
components; so examining wide range of frequency response is expected.
3
current is to be injected.
Now, let’s step in to Frequency dependent model.
As the most popularly and widely used modelling for overhead transmission lines, JMARTI (Jose Marti)
set up routine is took up. Also, in this case, calculations for transmission line shown in Fig. 3.1, 10km in
length, are done. In modelling calcula-
tion, various parameters are calcu-
lated, based on “frequency by fre-
quency” in mode domain. One exam-
ple is shown in Fig. 3.4, where propa-
gation constants of various modes of
propagating waves vs. frequency are
shown. Mode 6, representing great
damping in high frequency range,
mainly corresponds to zero sequence
mode. Also surge impedances and
travelling times are calculated.
Note:
- Such figures are easily obtained by
PARAM command in GTPPLOT. For
details, see HLP file and
Fig.3.4 Propagation constants of various GTPPLOT.PDF file attached to
mode of waves (damping) GTPPLOT.
In the set up routine, three kinds of
frequencies are input, i.e. for mode
matrix calculation, for steady state calculation and the lower limit frequency of the frequency range. Of
cause upper limit, also, is input. For mode matrix calculation, only one frequency is applied. In principle,
the mode matrix depends on frequency so, for other than the specified frequency, errors might be in-
troduced. As shown before, the matrix dose not varies so much depend on the frequency, the errors
are kept within permissible range, so experiences show.
Five kinds of frequency sets are applied as for fm (mode matrix), fs (steady state) & fl (lower limit).:
- No. 1 line fm = 50Hz fs = 50Hz fl = 10Hz
- No. 2 line 500Hz 50Hz 10Hz
- No. 3 line 5000Hz 50Hz 10Hz
- No. 4 line 50000Hz 50Hz 10Hz
- No. X line 1MHz 100kHz 0.1Hz (for wide range up to 10MHz)
For details of the input data, see attached data files, DATA3-06 ----- DATA3-09, and DATA3-0X. The
output PCH files (line parameters in JMARTI models) by these data files are calculated which should
be checked by your self-calculations. The PCH files are directly used in transient calculation data files
alike by LINE CONSTANTS.
For some very simple cases, examples are shown
next. In Fig. 3.5, step and cosine shape voltage is
injected from one end of the lines and the other
open ended terminal voltages are calculated. The
condition is almost the same as in Fig. 3.2, where
LINE CONSTANTS model is applied The end
terminal voltage wave shapes seem quite appro-
priate, in values, damping, rounded shapes con-
trasting to angular ones in LINE CONSTANTS,
shown in Fig. 3.2, finite front steepness. De-
pending on frequencies, which are the base of
characteristics, differences are negligible. Calcu-
Fig. 3.5 Injecting step-&-cosine wave voltage lating high frequency of voltage phenomena,
5
In d) El. magnetically induced currents in cross-bonded cable sheathes are shown, the values of which
are dramatically reduced, i.e. from ca. 1000A to less than 100mA.
In e) El statically induced currents in cross bonded cable sheathes are shown, where due to the both
end terminal earthing, the three phases’ capacitive currents are only partially cancelled, and the values
a) b) c)
d) e)
Especially for analysing disturbances in control cables by, e.g. lightning stroke to station conductors or
re-striking impulse current, both overhead and under ground conductors are to be included in one
analysis domain. In CABLE PARAMETERS, such option is applicable. In Benchmark DC27.DAT, one
example is shown, where an overhead and an underground conductors are modelled.
In the present CABLE PARAMETERS in ATP-EMTP, system consists of only one kind of conductors is
applicable. So, modelling system with coaxial cable(s), special idea(s) is to be introduced. Let us in-
troduce conductor layout, the cross-sectional view of which
is shown in Fig. 3A.1, where one overhead conductor, one
cable and four earthing mesh conductors exist. In the
system, due to the restriction of present CABLE PA-
RAMETERS, the cable is to be modelled as the same
conductor. So, modelling coaxial cable, another coaxial
cable in the another domain should be introduced, the
sheath of which is electrically connected to the relevant
conductor at proper points. Due to negligible penetration
through the coaxial cable sheath, this modelling is thought
to be appropriate, see Fig. 3A.2. As for the detailed data
coding in the modelling, see the attached data files. 60m of
the conductor system (segmented to three sections, for
obtaining intermediate information if any) and 100m of
under ground cable connection are introduced. The ground
Fig. 3A.1 Conductor layout mesh is earthed at both ends and the sheath of the cable
conductor is connected to the ground mesh via appropriate
impedance. An impulse voltage (1 MV in crest value,
enormous case) is applied to the overhead conductor and
various part voltages are calculated. The layout image in
longitudinal direction cross sectional view is shown in Fig.
3A.3
The calculated result summing up is shown in Fig. 3A.4, :
a) 1MV of impulse voltage is applied from a terminal of the
overhead conductor. By transmission and reflection, some
ripples appear and higher voltages appear.
b) The underground cable sheath voltage, based on the
Fig. 3A.2 Imaging underground cable very deep underground, reached to ca. 15kV.
c) The voltage difference between the cable sheath and
the grounding mesh is, lower than several kV.
Therefore,
d) The voltage between the underground cable core and
the sheath is negligibly low (several mV).
As the conclusion of this case, provide all facilities are
earthed to the electrical station underground mesh, the
maximum induced voltage in the underground secondary
circuits is several kV by 1MV of over-voltage in the outdoor
conductor.
Fig. 3A.3 Longitudinal cross section Great care should be taken in modelling.
view
10
CABLE PARAMETERS accept up to two-layer coaxial insulation cable. As shown in the previous ap-
pendix, due to non-penetration of the transmission wave through the sheath, multi-layer coaxial insula-
tion seems to be represented multiple coaxial cables of
respective radii in each domain, the sheathes and
cores are connected each other. Fig. 3A.5 shows the
concept.
As a typical example, let us take up a capacitance
graded bushing in a gas-insulated bus-duct. Due to the
difference of length between insulation layers, i.e.
travelling time, something may happen in dielectric
stress in very fast transient voltage. Fig. 3A.6 shows
sectional view in such modeling layout, though very
arbitrary. Hopefully we can obtain general tendency in
such simplified model. To model cone type bushing,
core elements in coaxial insulation with different length
are to be introduced. Calculating coaxial insulation
Fig. 3A.5 Multi-layer coaxial insulation elements, overhead coaxial cable in very high position
representation from the ground surface is applicable both for gas in-
sulated bus-duct and bushing core. For detailed di-
mensions/parameters, see attached data files.
In the surge voltage travelling calculation, VFT voltage
is applied from one side of the layout in Fig. 3A.6, and
the other side is connected long bus duct. Voltage dis-
tribution along the bushing cone surface is calculated.
Typical calculated results are shown in Fig. 3A.7. In a)
four sections’ cone type surface voltage distribution is
shown. In the first part, due, most probably, to reflec-
tions, slight voltage enhancement appears. In b),
highest stressed part, i.e. conductor surface part, de-
tailed voltages are shown. Out-going side voltage
Fig. 3A.6 Modelling capacitance graded shows higher enhancement, which is higher than the
bushing applied one (gas insulation part) by approx. 30%.
The calculation shows that cone type capacitance
graded bushing in gas insulated bus duct may produce more or less voltage stress concentration by
VFT surge coming.
1
4. Transformer
Note:
The features of the latter two (XFORMER, BCTRAN) are,
- From general test data of transformers, data to be applied to transient calculations are directly
calculated.
- The data are in the form of “mutually coupled inductors” (actually in “PI-EQUIVALENTS” form,
the function of which perfectly covers “Mutually-coupled R-L Elements”, see Rule Book)
For example, three-phase two winding and one core transformer is represented by six mutually
coupled inductors.
- Saturation characteristics are not covered within the scope. So, another saturable non-linear
element(s) is to be introduced out side. The influence by the saturation on the self/mutual in-
ductance(s) can never be considered.
- Magnetising inductance corresponds to self-inductance of the relevant winding, but leakage in-
ductance cannot directly represented.
- Total procedure is “BLACK BOX” like one.
dt dt
Replacing
n1φ 1 = l1i1 , n1φ 0 = L0 i0 n 2φ 2 = l 2 i 2
it is understood that Fig. 4.2 represent the same contents, so Fig. 4.2 can be a perfect equivalent
circuit of a two windings transformer. The “Saturable transformer component” menu in ATP is base
on the principle of this circuit.
2
Notes:
- In any of transformer menu, each winding is considered as one lump element, i.e. the current in
one winding is uniform. This seems to be correct up to the fundamental inherent frequency of the
winding. The value is in the order of a few --- several kHz for most power transformers.
- In calculating TRV, for example, the prospective frequency of the TRV is dominated by both the
transformer elements and system ones such as lines, cables, etc. So, the frequency is never
higher than the transformer’s inherent one. Therefore, transformer menu in this chapter is appli-
cable.
- For much higher frequency of phenomena such as by lightning surges or VFT switching surges,
another models are required. Depending on the frequency concerned, respective models are to be
considered. In most cases, considering the physical constructions, parameters are to be calcu-
lated. “Mutually coupled R-L elements” and additional capacitors are mostly applied.
A few examples
ⅰ) 537kV/287.5kV Auto-transformer, inrush magnetising current
The first example is as for inrush
magnetising current in an
auto-transformer. (The maximum
tap voltages of the transformer are
shown in the figure. The maximum
system voltages are 550kV and
300kV.) As shown in Fig. 4.3, each
phase has a common winding
(287.5kV) and a branch winding
(537kV – 287.5kV). Also
delta-connected tertiary winding
Fig. 4.3 537kV Auto-transformer (74kV) exists. For details of the
G0 = I 2 N 0 G = G0 (1 + θ s )
These equations are easily introduced to TACS in ATP-EMTP, and the result G can be the arc
conductance, the reciprocal of which is introduced as the circuit-breaker into the system circuit.
For details, see the attached data files.
Mayr arc model is suitably applied to low current (< several tens A ) of arc, or even post arc current
arc.
Though fault current of a cir-
cuit-breaker is very high, cur-
rent interrupting is phenom-
ena around current zero, so
Mayr arc model is suitably
applicable. Also the model is
applicable to “arc”, i.e. high
temperature gas predominant
time region, therefore, so
called “Inter-action interval” is
Fig. 5.1 Short line fault breaking circuit a typical application.
300kV, 50kA, L90 condition Note:
- After the Inter-action interval,
the time interval is called as “High-voltage interval”, where dielectric phenomena are significant, so
another model(s) is to be applied.
As the first example, short line fault current breaking is taken up. Fig. 5.1 shows basic (simplified)
circuit according to IEC standard for 300kV, 50kA circuit-breaker’s L90 (SLF current is 90% of
terminal fault one) rated condition. For the circuit-breaker, Mayer arc model is applied,
where, θ(time constant) = 1μs, N0(arc loss) = 293kW
The main part of current breaking phenomena in this case lasts several micro seconds, and, also
very short length of distributed parameter line for ITRV is involved. Therefore very short step time
2
Fig. 5.5 shows an alternative circuit, which is approved in IEC standard as equivalent to Fig. 5.1,
main purpose of which is making
easier the breaking test. For de-
tails, see IEC standard
62271-100.
In the circuit, ITRV circuit is ex-
cluded and, instead, the ramp
capacitance at the line side ter-
minal is lowered. Then the TRV
across terminals is mostly
equivalent to the original one.
Fig. 5.5 Alternative circuit to Fig. 5.1 By the circuit, introducing
θ= 1.0 μs
N0= 300kW
critical condition is obtained.
Much fortunately by quasi equal
arc parameters, the critical con-
dition appears and both circuits
are evaluated to be equivalent.
Nevertheless, in Fig. 5.6, some
differences in post arc current
and TRV damping are observed.
The circuit in Fig. 5.5 produces
much post arc current and much
damping of TRV.
Fig. 5.6 Breaking currents and TRVs by Fig. 5.1 and 5.5 By circuit-breakers with another
arc parameters ( θ, N0), the re-
sults may be different.
modern SF6 gas circuit-breaker. For detailed modelling, also see the attached data file.
Calculated circuit-breaker current
is shown in Fig. 5.8. Before ap-
proaching the prospective current
zero, at ca. 8A, oscillation begins
and by the oscillating current zero,
i.e. before the prospective inter-
ruption time, the current is inter-
rupted. Roughly looking, the cur-
rent is chopped at ca. 8A. There-
fore the phenomena is called as
current chopping.
In the figure, the first part oscilla-
tion is created by switching for the
calculation purpose This may help
Fig. 5.8 Current chopping by shunt reactor current to introduce disturbance to create
breaking oscillation.
By doubling N0, corresponding to
longer arcing time (longer contact
gap), higher gas pressure, or
multi-break circuit-breaker, the
calculated result is shown in Fig.
5.9, where oscillation initiating
current is almost doubled from the
above case. Such phenomena is
know, i.e. by stronger quenching
force, chopping current is higher.
In some literatures, chopping cur-
rent is reported as approximately
proportional to square root of
breaking point number. Factor
Fig. 5.9 Ditto, but doubled N0 other than N0 may be dominant by
increasing breaking point.
The next example, shown in Fig.
5.10, is by doubled parallel ca-
pacitance to the reactor winding.
In the relevant oscillation circuit,
i.e. 0.03 μ F, 50 μ H , cir-
cuit-breaker and the relevant ca-
pacitor, which is called as “second
parallel oscillation circuit,” actually
the capacitance value is doubled.
Then as mentioned in some lit-
eratures, the chopping current is
enhanced approximately propor-
Fig. 5.10 Same as Fig. 5.8, but doubled reactor parallel tional to the capacitance value.
capacitor from 5nF to 10nF
5
G0 = G 2
(G = G0 ) G=I E G0 =
I2
E0
2
1
1+θs
R =1 G
These equations are, likewise as Mayr model easily introduced to TACS. For details, see the at-
tached data file.
For steady state, i.e. d/dt = 0, arc voltage E equal constant E0. Therefore, as the equation is to be
applied to relatively long time interval of high current region, to introduce just appropriate E0 value
is important. As an example, so called zero skipping current breaking near a synchronous gen-
erator is taken up.
In Fig. 5.11 a), the
generator is sup-
plying transmis-
sion line charging
current via the
step-up trans-
former. Close to
the 550kV bus bar,
three-phase
grounding faults
occurs in one of
the two circuits.
Non simultaneous
faulting is intro-
duced, i.e.
three-phase point
on wave individual
Fig. 5.11 Circuit for zero skipping current breaking calculation timing of faulting
to create max of
current zero skipping. Actually, the faulting timing is, 14ms, 8.3ms and 8.3ms for phase A, B and C
respectively from the voltage crest in phase A. The timing was gotten by trial and error procedure.
Therefore such high rate of zero skipping of current could seldom be created.
For introducing circuit-breaker’s dynamic arc characteristics calculated by TACS, circuit diagram in
Fig. 5.11 b) is used. The main reason is TACS is active only after time 0, so for initialisation pur-
pose, i.e. for t < 0, the switch connected in parallel is required. For more details, see the attached
data files. Some important generator parameters applied are shown in the figure. Also for other
details of modelling synchronous generator, see the following chapter(s).
In Fig. 5.12, three-phase short circuit current is shown, where in phases A and C apparently no
6
In the middle of eighties, the present Type 59 synchronous machine model program was implemented
and put into practical use in EMTP. In the first half of nineties, also Type 58 model, which has
significant improvement from the former, was put into practical use in ATP-EMTP. Most sources of AC
power systems are synchronous generators, so the dynamics of the machines are of great interest,
especially regarding relatively short time interval of phenomena. Only time domain analysis is
applicable to such fast phenomena as down to sub millisecond. In such circumstances EMTP is a
significantly useful tool in power system dynamics analyses. As the special feature of Type 58,
calculations are stable especially in asymmetrical circuit conditions such as non-transposed over-head
lines, which are mostly applied in today’s power systems.
It should be noted the present type 59 involves a great bug calculation of saturation in magnetising.
The usage is mostly common by both, excepting write “58” or “59”. In this chapter, therefore, mostly
Type 58 is explained.
j,k : a, b, c, F, G, KD, KQ
Ljk : time varying functions, depending on
the angle between Rotor and Stator
As for torque,
Note:
- In the 2nd and 3rd lines, only 58 and node names are to be written. Voltages, frequencies and an-
gles are automatically introduced as symmetrical three phase AC.
- < 2.0 of PARAMETER FITTING corresponds to open circuit time constants are to be used.
> 2.1 of PARAMETER FITTING corresponds to short circuit time constants are to be used.
- “1” in col. 7 of 5th line corresponds to metric unit mechanical constants are to be used.
- For R and X, p.u. values (machine rating bases) are to be applied.
- For time constants, “second” is to be used as the unit.
- If XCAN (Caney reactance) can be applied, transient rotor coil currents such as during short-circuiting
are more correctly calculated. For armature currents, little influence is introduced. Without introduc-
ing the value, XL value is automatically introduced as XCAN.
- To write 11, 21, 31, and 51 in Output ordering cards yields full out put for one mass machine case and
generally recommended.
- For initialising Type 58 machine, CAO LOAD FLOW option is applicable, which may introduce better
results especially asymmetrical circuit cases. The usage of which is identical to FIX SOURCE, see
data files in the following example case.
are “a”, “b” and “c” from the top in one side, and “c”, “b” and “a” in the other side for obtaining as better
symmetry. Details are shown in Chapter 03 of this text. As for step-up transformers, details are shown
a) Generator terminal voltages and currents b) Line voltages and charging currents
- Max allowable step time was approx. 100 μs in the case. By longer step time, diverge may arise.
The critical value depends on also the circuit parameters (transmission line, load circuit, etc.).
4
- Generally, imaging actual systems, numerous times of “Try and error” process is inevitable for op-
timum calculation even such simple cases.
Some results are shown in Fig. 6.3. The HV line charging voltages and currents are shown in b) and
the generator terminal voltages and outgoing currents are in a). In HV side, asymmetry is not significant
due to the phase line location crossing. In generator side, on the other hand, some asymmetry exists in
the currents. The cause seems to be “Delta-Wye connection” of the step up transformer, i.e., the cur-
rents out of Delta connected coils are the subtraction of the coil currents corresponding to the HV side
currents. Subtraction often introduces higher asymmetry. The steady state rotor winding currents in
symmetrical condition are to be constant, i.e. the rotor and the armature flux rotate in an equal speed.
Nevertheless, Fig. 6.3 c) shows some fluctuations in the rated frequency. The cause seems the
asymmetry of the load (transmission line) circuit. In the torque (Fig. 6.3 d)), also ripple of doubled fre-
quency exists.
a) Gen. terminal voltages & currents b) HV bus voltage & Line charging current
a) Generator terminal voltages & currents b) Bus voltages & fed currents
c) Voltages along the transmission line d) Generator torques & d-axis angles
Fig. 6.5 Two generators and infinitive bus case ---- Supplying full active power
6
In the next example, two machines are feeding load power. All of components in Fig. 6.2 are connected
and both generators are feeding full active powers while the reactive powers are automatically fixed by
CAO LOAD FLOW, details are shown in Data6-04.dat. Some results are shown in Fig. 6.5. In a), gen-
erator terminal voltages and feeding currents are shown in comparison, G1 terminal voltage phase
angle is a little advanced compared to G2, while the current phase angels are respective. G1 and G2
feeding apparent powers are calculated to be 1185MVA and 620MVA respectively.
HV side voltages and currents of the plants are shown in b). The phase angles are advance by 30 de-
grees compared to the lower voltage side due to Delta – Wye connection, and the wave shape rela-
tionships are same in both sides. The voltage phase angle relation along the transmission line is shown
in Fig. 6.5 – c). Angle differences in Bus21 – Bus31 – “Inf.-bus” are apparently higher by higher active
power transmission, compared to the one generator case in Fig. 6.4 – c). As for the voltage amplitudes,
the middle part is the lowest. This will be explained in Appendix 6.3.
d) shows air gap torques and rotor’s d-axis angles of two generators. Ripples in both generator torques
are not yet thoroughly clarified.
Sudden short-circuit
Next calculation is for sudden short-circuit, i.e. to calculate transients when short circuit occurs. In the
first example, assuming three phase simultaneous short circuit near Bus 11 and clearing after several
tens milli-seconds. As for details of the data coding, see Data6-05.dat attached. Some results are
shown in Fig. 6.6.
Three phase fault current is shown in a). What is notable is the AC component amplitude gradually
damps during short circuiting. The cause will be shown in d) and e). The fault current is the sum from
G1, G2 and infinitive bus through relevant reactances (transmission lines, etc.).
b) shows the generator side currents during the fault. The maximum ac amplitude of G1 is approxi-
mately twice of the rated current which is limited by the generator’s sub-transient and transformer re-
actances.
c) shows bus voltage of before, during and after the faulting. At the instant of the fault clearing, the ac
component of the voltage (recovery voltage) shows damping by ca. 15% compared to the pre-faulting
value. Directly this corresponds to the damping of armature coil linkage flux. (See e)) This shows
(transient) recovery voltage immediately after fault clearing near generator(s) is more or less damped.
d) shows rotor coil currents. During steady state, only the field coil currents flows (steady DC). But
during the transient, some value of AC and DC flow with respective damping. These damping are the
cause of the damping in short circuit current and recovery voltage.
e) is the d-q plain representation of armature d-q domain coil flux linkages (See Fig. 6.1). In steady
state, both coils flux linkages are steady, i.e. the initial point during steady state is upper right position.
During faulting via transformer impedance, the flux trace draws smaller circle. If short circuit occurs at
the generator terminal, i.e. the voltage equal to zero, the full value of the flux is trapped, then the trace
is to be a big circle with a radius of the initial flux value. Details will be explained in the Appendix 6.1.
(Flux linkage output is applicable using TACS, see the same Appendix.)
f) shows rotor d-axis position and the rotating velocity change. During faulting, due to mainly reactive
power output only, the generator more or less accelerates and this acceleration in the most important
cause of transient stability failure. The phenomena is well analysed by EMTP simulation. In the next
chapter, details will be shown.
In chapter 05 fault current zero skipping is briefly discussed. Also in this chapter, the subject is to be
surveyed introducing a few kinds of faulting condition.
7
c) Fault current from No. 2 plant d) d-axis angle & velocity change in G1
Fig. 6.7 (continued)
From the survey in this chapter as for zero-skipping of fault current, the following are clarified. :
9
Appendix 6.1 Synchronous machine sudden short circuit --- description on d-q
co-ordinate plane
In Type 59 synchronous machine program, calculation is made as d-q-0 domain variables. Though
some demerits exist in Type 59 compared to Type 58, phenomena within the machine are well repre-
sented by Type 59. In this appendix, such inside phenomena in sudden short circuit will be discussed
for better understanding of synchronous machine dynamics. Therefore, Type 59 is better to be applied.
As for d-q-0 domain calculation method,
we can find some materials/literatures
easily, so little is necessary to explain. In
the first example, we take up the follow-
ing perfectly symmetrical (non-salient)
rotor machine. The machine parameters
are, :
Ratings: 3.3kV, 1MVA, 2P, 50Hz
Main constants: Xd = Xq = 1.5pu, Xd’ =
Xq’ = 0.2pu, Td’ = Tq’ = 0.3sec.
The machine has no damper coil and is
of winded rotor induction motor type one, a) 3 phase SC current b) Armature flux linkage
and some of such ones are applied as
variable speed generators. Again, going
back to Fig. 6.1, assuming three phase
armature coils are suddenly and simul-
taneously short-circuited. The flux link-
age in the armature is trapped at the ini-
tial value with damping concerned. Fig.
6A.1 a) shows three phase armature
currents during short circuit. In two
phases, high value of DC component
c) Rotor flux linkage d) Armature current
exists. b) shows armature flux linkage in
d-q domain plane, based on the rotor
position. As the flux is fixed on the ar-
mature, its rotes in d-q plane, drawing a
spiral. c) shows rotor flux which is also
trapped with damping due to short circuit
of the coils (voltage source of field is of
zero impedance.).
Assuming two coils, coupled each other,
the next equations are easily obtained.
e) Rotor current f) SC current without DC
Winding turn ratio is assumed to 1.0,
then we have,
Note
For drawing X-Y plot graphs applied in this Appendix, PlotXY.exe is best applicable. Also
GTPPLOT.exe is applicable for the purpose.
12
Synchronous generators/motors are occasionally started as induction motors, where the field coils are
short-circuited. The principle is identical to induction machine. Type 58/59 synchronous machine pro-
gram in ATP-EMTP was originally developed as in service in near power frequency range. Therefore,
for calculating quasi zero speed of phenomena, special care is to be taken.
a) Velocity change, origin corresponds to 1Hz. b) Air gap torque when starting
c) Voltage enlargement
7. Transient stability
In AC power system, each generator is to keep phase relationship according to the relevant power flow,
i.e. for a certain reactance X, the both terminal voltages V1and V2, and phase angle difference θ, the
active power flow P through the reactance is calculated as:
P = V1 • V2 sin θ (7.1)
X
In the previous chapter, such relationships are calculated for some steady state system conditions. By
disturbances such as short-circuit, sudden load rejection, switching transmission line, etc. each gen-
erator may accelerate/ decelerate due to the probable unbalance between the driving and load torques.
The angle θ before shown may swing and such phenomena are called as “Transient stability”.
In this chapter the phenomena are explained mainly applying time domain analysis, contrasting with
conventional process “Power frequency phasor domain analysis” or “Equi-area method”, and also the
counter measure to enhance the stability will be explained.
Fig. 7.5 Generator terminal voltage Fig. 7.6 Field exciting current
In Fig. 7.3, the maximum rotor swing is approx. 20 degree for 4 pole machine, i.e. electrically 40 degree.
In Fig. 7.2, the value is 36 degree. The difference depends on the flux movement on the rotor shown in
3
Fig. 7.4. In classic method, such as equi-area method, source side voltage is an imagined one. In time
domain analysis, every actual value is calculated. Also actual voltage amplitude is changeable, which
can not be involved in classic method.
A little more details are shown in Fig.
7.5 and 7.6, the former of which is
generator terminal voltage and the
latter field exciting current. Both
change during the time interval. In Fig.
7.4 also the flux angle and amplitude
changes are shown. These show the
variables of the generator are
changeable during the transient,
As the next example, increasing the
transmission active power by 5%, the
rotor angle swing is shown in Fig. 7.7
in comparison with the former case.
Fig. 7.7 Rotor angle by over-loading The amplitude increases, furthermore,
the recovery delays a lot. The condi-
tion seems to be critical. Actually, not
shown, by more 5% of power increase, the generator losses the synchronism (out-of-phase).
In the next example, line
faulting result in one circuit
of the double-circuited line
opening is taken up. Two
cases are introduced.:
Case 1: excluding interme-
diate switching station in
Fig. 7.8, faulting (F) is
Fig. 7.8 Power system diagram, 3LG fault --- 1 circuit clearing
cleared by CB11 and CB31,
i.e. whole length of the line
is to be single-circuited.
Case 2: introducing inter-
mediate switching station,
the fault is cleared by CB11
and CB21, i.e. only half of
the line is single-circuited.
The results are shown in
Fig. 7.9 as the rotor angle
swing. In case 1, 1.17GW of
transmission power results
in out-of-phase, so as the
critical value, 1.13GW is
introduced. In case 2,
where only the half of the
line is single circuited, due
Fig. 7.9 Rotor angle swing by 3LG --- 1 circuit clearing to less enhancement of the
line impedance, significant
improvement is apparent as for transient stability enhancement.
In the next case, second
generator unit is con-
nected to the intermedi-
ate bus, the rating of
which is 600MVA with 2
pole, feeding 0.6GW.
Faulting and clearing
sequence is the same as
the former.
The result (rotor angle
swing) is shown in Fig.
7.11. For easy
Fig. 7.10 Two generator vs. inf. bus case
4
Fig. 7.17 Two Gen. Vs. infinitive bus case Fig. 7.18 G2 El. Torque
Fig. 7.19 8 generator vs. inf. bus system Fig. 7.20 Generator swing angle
as applying No. 19 synchronous machine model to be introduced partly, might be necessary. According
to the Rule Book of ATP-EMTP, No. 19 synchronous machine model seems to be more stable for espe-
cially multi-machine case. Therefore machines distant from the disturbance, i.e. short-circuiting, might be
suitable to be modelled as No. 19 machine.
a) Fault current with full DC b) Fault current with less DC c) Torque under full/less DC
In Fig. 7.21 fault current and generator’s air gap torque for both full of and less DC component are shown.
DC component condition depends on the point on wave fault initiation timing. Significant difference be-
tween the two conditions is found in c). Nevertheless, difference in the swing angle is not so significant
as shown in Fig. 7.22. Also for full DC condition, which is possible only in time domain analysis (EMTP),
corresponds to less severe condition under full load condition such as the present case.
7
For both modelling initialisation by FIX SOUCES / CAW LOAD FLOW is not suitably applicable. To the
author’s experience, giving the generator terminal voltages and phase angles to all generators has been
suitably applied. For No. 58/59 machines and U.M. in type-59 format, values are directly typed. In U.M.
by standard format, as automatic initialisation of U.M. the values are also applicable. For details, see
Data7-14, -16 and 17, also Rule Book. Obtaining these initial values (terminal voltages and phase an-
gles), another power flow calculation is necessary, where, No. 14 alternating current voltage sources
instead of generators and FIX SOURCES menu are best applicable.
In Fig. 7A.1 --- 3, calculation results by three type of S.M. modelling for G2 are shown where in the
graphs three kind of results are drawn superimposed. Difference between the modelling methods is not
significant, so any kind seems to be applicable depending on your choice.
1
No. 19 Universal machine input data is based on T-type equivalent circuit. Inputting these catalogue
values, but 500Hp instead of 667Hp, the following machine parameters are output. As for the details of
inputting and outputting data, see the attached files and/or text file attached to the program.
(As induction machine is of non-salient configuration, common values are applied to both d- and q-axis.
Also, in No. 19 UM, rotor winding is modelled in the same turn number as the armature.)
- If test/measured values are available, the values are most preferably to be applied.
- For magnetising branch, saturation characteristic is applicable in No. 19 program, if necessary. For
leakage inductance saturation, if necessary, TACS/MODELS controlled additional inductances are to
be connected outside of the model.
Also the Indmot.exe program produces PCH file, the format of which is directly applicable to EMTP
calculation. Using the PCH file, the machine’s starting was calculated. For detail of the program data,
see the attached file.
Note:
- In the attached data file, spe-
cial initialising technique is ap-
plied, which is applicable also
for multi-machine case. Detail
will be explained later.
In the program, mechanical system is represented as electric circuit, i.e. Torque (N-m) is in Current (A),
2
and Velocity (radian/sec) is in Voltage (V). Therefore, Moment of inertia (kg-m ) is by Capacitor (F) and
Viscosity (N-m/radian/sec) is by Resistor (ohm).
A few examples are shown next regarding these.
4
Fig. 8.3 Sudden rated torque application under no-load operation condition
In the first case (Fig. 8.3), mechanical torque is suddenly increased to almost the rated value. Torque is
dominated by the current source in the equivalent electric circuit. (See the attached data file.) Velocity
decreases rather gradually and also active/reactive/apparent power enhances. Active power component
is far higher than reactive one in the on-load region.
In the last case (Fig. 8.4), mechanical load with certain moment of inertia and friction loss is connected
via clutch and torque converter. In the electric equivalent circuit, clutch and torque converter can be
represented by a switch and a non-linear resistor respectively as shown in Fig. 8.4 d).
In a) (Velocity and torque) the load smoothly accelerates and at the initial stage highest torque is trans-
ferred from the moment of inertia of the motor. By decrease of the velocity motor torque increases and
then the velocity recovers. At the initial and the final stage, the difference of torque between the motor
(air gap) and the load corresponds to the motor friction loss. The velocity decrease is rather high (over
loading condition), so reactive power increase is higher.
Note:
5
Fig. 8.5 Two-motor case, one is under operation and the other starting
In Fig. 8.5 b) ---- e) calculated results are shown. In b), M1’s velocity downs by approx. 0.6% during M2’s
staring due to the terminal voltage down. Nevertheless, in c), the torque is kept quasi constant which is
the given condition. In d) M1’s powers are kept approx. constant and, in e), the current is enhanced a
little.
Note:
- Applying the principle given in the attached data file, up to several tens of machine case can be
calculated.
induction motor seems to be a constant power load for the system irrespective of the applied voltage.
Great care should be taken of the over current by the voltage drop.
The phenomena might be of great importance as voltage stability (voltage collapse) of power system, as
voltage stability is quite vulnerable to ruin by constant power load.
VVVF starting
Owe to development in
power electronics, any
kind of voltage/frequency
source can easily be ob-
tained. As shown before,
when starting induction
machine by direct
switching on, extremely
high current/power flows.
But applying suitable
power electronics con-
trolled source, more so-
a) Equivalent circuit b) V1 vs. E0 phisticated starting is
realised.
Fig. 8.7 Induction motor equivalent circuit & voltages Fig. 7a) shows an induc-
tion motor equivalent
circuit in simplified one-phase representation. The following equations are introduced by the circuit.
E (R2 ω s ) 2 + (M + L 2 )2 E0
2
R2 ω s
I1 = 0 T =3
ωM ( R 2 ω s )2 + L 2 2 ω 2
(R2 ω s )2 + L 2 2
Then, keeping E0/ω and ωs to be constant respectively, constant current and constant torque driving is
realised, where ωs is slip frequency in angular velocity.
7
So, supplying linearly rising voltage with linearly rising frequency, starting with constant torque, i.e. ac-
celeration, and constant current is expected. This is the most simplified VVVF controlling. For more
sophisticated controlling, turning of V1 (terminal voltage) as shown in Fig. 8.7 b) is to be applied.
In the following example, TACS controlled voltage source is applied, where the voltage wave shape is
mathematically created in the TACS. For details, see the attached data file.
Actually such voltage source is realised by power electronics circuit, which will be explained in the fol-
lowing chapter (chapter 10).
Fig. 8.8 shows calculated results, where the next features exist. :
- Equations before shown correspond to steady state. Actually another transients may be introduced.
So, for smoother controlling, mitigated originally rising signal, shown in Fig. 8.8 a) is applied. In
8
side, the slip frequency of current (0.05Hz) is supplied. Small plus value of air gap torque exists for each
case, i.e. generator mode. The mechanical torque is supplied from outside corresponding to the current
source supply.
In the next example, let’s introduce dynamic controlling. As shown in Fig. 8.9, specifying I2 vector, i.e.
secondary side current, the machine is controlled.
Note:
- For various controlling mode of the machine, details will be shown in the Appendix.
10
If the machine output is specified as generator without changing the mechanical torque, depending on
the air gap torque and moment of inertia, the machine decelerates. By motor mode, it accelerates.
Based on the vector diagram in Fig. 8.9, the secondary side current vector can be calculated by the chart
shown in Fig. 8.11. For the details of the vector calculation equations, see the attached data file, where
vector calculation is represented in FORTRAN equations in the TACS part.
The left side flow shows
calculation of the second
side command current seen
from the primary side. The
chart can be perfectly
based on Fig. 8.9.
The right side flow is for
calculating the rotor position
angle. The angle can be
obtained from the calcu-
lated primary side view of
the secondary side drive
current and the actual sec-
ondary side drive current.
For both side vectors, ma-
chine parameters and/or
measured constants are
applicable. Then, without
rotor position sensor the
rotor position angle is ob-
tained which is to be ap-
Fig. 8.11 Calculation of secondary side commanding current plied to calculate the sec-
Based on the rotor position ondary side (rotor) drive
V1: Primary terminal voltage PP, QQ: Target output (Act./React) current.
X1: Primary leakage reactance X0: Magnetising reactance In the example case where
I1: Primary target current E1: Internally induced voltage the machine is in operation
I0: Magnetising current I2: Primary side view of command as a fly-wheel generator,
I10: Current primary current I20: Current secondary current the machine is primarily in
I21: Current calculated secondary current based on Primary operation by 110% of speed
under mostly no-load condi-
tion. Then, approx. 150% of
output (generator mode) power value is applied. The machine decelerates. It should be noted that the
magnitude of the current in the mechanical network, i.e. torque is in the order of MN-m. So the con-
necting resistance in the circuit is to be very low, otherwise abnormal velocity violence may be created.
Active, reactive and apparent powers of both primary and secondary sides are shown. It should be noted
that the secondary side power depends on the slip velocity, i.e. mostly in proportion to the slip values.
The secondary power source is TACS controlled current source. See Fig. 8.12 for details.
As shown in Fig. 8.12 e) and f), the secondary power source is to be in proportion to the slip value for
constant primary side output, high value of mechanical moment of inertia corresponds to lower secon-
dary side power source capacity due less velocity change from the synchronous velocity.
Note:
Lately, during editing this chapter, WINDSYN.EXE was released which seems to be a well developed
INDSYNW.EXE mainly regarding GUI (Graphic User Interface) for calculating machine parameters and
creating PCH files. Details will, hopefully, be shown elsewhere.
- Data8-08.dat: Starting by VVVF source (TACS controlled) for constant torque/current, minimum
power loss.
- Data8-11.dat: Doubly fed machine, approx. no-load condition, voltage source controlled at the sec-
ondary side.
- Data8-12.dat: Ditto, but current source controlled at the secondary side.
- Data8-15.dat: Ditto, but the output is controlled by sensor less vector controlling principle.
- DFED.DA: INDSYNW.EXE input and output data for induction machine.
- DFED.PCH: Ditto PCH file applicable to case calculation data file.
Appendix 8.1 Doubly fed machine vector diagram as fly-wheel generator operation
For a few cases of doubly fed machine operations, vector diagram are drawn as follows.
Fig. 9.1 HVDC converter/inverter circuit layout ---- Plus polarity only
a) Upper side phase current and voltage b) Lower side phase/transformer currents
c) 50Hz system side phase current and voltage c) 50Hz current Fourier spectrum
(Fig. 9.2 Converter circuit variables ------ continuing to the next page.)
2
Fig. 9.2 Converter circuit variables ----- continued from the previous page
Fig. 9.2.
As shown in the circuit diagram, valve switching side is constructed by two groups, the phase voltage
angle difference between two groups is 30 degree by means of Delta/Star transformer winding connec-
tions. By such arrangement, higher DC voltage is easily obtained, and at the same time, harmonics in
the AC side current is diminished. Fig. 9.2 a) --- d) show such effect. Each top or bottom side bridge
phase current is square formed one. But, combining both side currents, AC side current is well formed.
The effect is clearly shown in the Fourier spectrum (d), where low value of (12n + 1)’ order harmonics
exist. Filter’s capacity for eliminating such high frequency of harmonics can be rather low compared to
lower frequency of harmonics.
Valve gate signals are to be based on the phase-to-phase voltage’s phase angle, which is shown in Fig.
9.2 e). In the case, so called ignition delay angle (Alpha) is 18 degree. In f) output DC voltage is shown,
together with the case of ignition delay (Alpha) equal to 30 degree. Theoretically, by primitive estimation,
the output voltage is proportional to cosine of the delay angle (cosα). For further eliminating the ripple in
DC, higher value of DC reactor can be applied.
Next, let’s study externally communicated inverter circuit simulation. For higher capacity of systems such
as power utility ones, due to economical and efficiency point of view, externally communicated systems
are exclusively applied, where relatively cheep and high capacity of thyristers can be applied.
Connecting DC source to the right side of Fig. 9.2 b), where for easier understanding the directions of the
a) Upper side phase voltage and current b) Down side phase current
c) 60Hz AC side voltage and current d) Upper side valve gate signals vs. voltage
Fig. 9.3 Inverter variables ----- DC --- 60Hz AC
3
thyristers are reversed, and 60Hz of AC source to the left side, the performance is analysed. Some re-
sults are shown in Fig. 9.3.
It should be noted that the gate signal timing (for phase “a”) is advanced from the zero point of the ap-
plied voltage (Va – Vc) as shown in d), resulting in normal externally communicating inverter perform-
ance. This is called advancing angle (Beta or β). Phase currents for both upper and down sides are
square wave formed ones like in converter, and combining these in AC side via the Delta – Wye con-
nected transformer, well formed AC current (60Hz) is produced. The current value is controlled by both
DC voltage and advancing angle.
Finally, connecting the right side of the converter to the right side of the inverter, excluding DC source
a) AC side voltage wave shape (50Hz/60Hz) b) AC side current wave shape (50Hz/60Hz)
Some calculated results are shown in Fig. 9.6. SVC controlling is based on α (Alpha)---- ignition delay
angle, by which the current flowing window is controlled, together with the crest value.
a) Phase to phase voltage and inductor current connected in between by α = 30 degree. The current is
no more sinusoidal.
b) Phase current and within delta current at the top of the delta. The phase current wave shape is well
formed. Fourier spectrum is shown in f).
c) Ditto but α = 45 degree.
d) Ditto but α = 55 degree
e) Reactive power calculated for α = 5, 30, 45, 55 degree. The calculation basis is 3-phase balanced
sinusoidal wave shape, so the absolute values may be questionable. For the correct values, calcu-
lations based on the fundamental component of the Fourier spectrum are to be performed.
f) Fourier spectrum of the phase current in a)
5
Some calculated results are shown in Fig. 9.9. a) shows control signals in TACS by which gate signals to
switch elements in the inverter are created. b) shows actual applied voltage wave shape to phase A to B.
At approx. 40ms, the control signal VA0 – VB0 in a) is maximum. At that timing in b), pulse width is
6
maximum, i.e. the pulse width is well controlled, proportional to the crest value of the voltage. c) shows
Fourier spectrum of the voltage in b), where harmonics of the carrier wave frequency and its integral
numbers are significant. As inductively reactive components are involved in the load circuit, harmonics in
the load current is not significant as shown in d) and e).
9.4 Cycloconverter
For relatively low frequency of power source such as 10 --- 20 Hz, cycloconverters have been widely
applied, the special feature of which is that high power and relatively low price thyrister is applicable as
the switching valve element, and the efficiency is high due to direct frequency converting.
In Fig. 9.12 one-phase of cycloconverter circuit is shown, three sets of which compose a three-phase
cycloconverter. In a three-phase cycloconverter, minimum 36 arms of switching elements are involved
7
c) Through bridge & load currents d) Fourier spectrum of the load current
f) shows the system (60 Hz) phase to phase voltage based saw teeth waves (three phase) and cosine of
the target voltage basis (15 Hz), by which the ignition delay angles (alpha’s) are calculated in TACS.
a) shows across-bridge voltages of both polarity converter bridges. Each bridge produces one polarity of
voltage but due to the connection each other, both polarity of voltages are induced on the terminals.
b) shows the Fourier spectrum of the voltage, where relatively high order of harmonics are involved. Six
times of switching per cycle are performed in the converter bridge and the out-put frequency is one forth
of the system frequency. Therefore the number of harmonic orders are around 6 X 4 = 24.
c) shows both bridges’ currents and the load circuit one. Around the load current zero time interval, cir-
culating current through both bridges are observed, producing continuity of the load current at the zero.
d) shows the Fourier spectrum of the load current, which involves less harmonics, i.e. little distortion of
the wave shape.
e) shows the load voltage and the current wave shapes. The instantaneous voltage in the power system
including, especially, around the current zero time interval, are high. This corresponds to very low power
factor of load current in the power system. This is typical demerit in cyclocoverter. Details will be shown
in the next chapter regarding rotating machine driving application.
parallel to the switching elements. Current regulating principle is shown in Fig. 9.14 b). The actual cur-
rent shows zigzag wave shape within the upper and lower limit band, the centre of which is the target
reference current wave shape. Care should be taken that the switching frequency of the switching ele-
ment is higher by narrower limit band and higher DC voltage. The minimum DC source voltage in Fig.
9.14 a) depends on the load impedance, the highest output current, internally induced voltage if any and
the frequency. The gate signals for the switching elements can be composed in TACS as shown in the
data file. See the attached data file. Dr./Prof. Ned Mohan (University of Minnesota), in the text book for
ATP-Exercise, introduced an excellent algorism for the purpose, the principle of which is introduced also
in the data file.
Some calculation results are shown below, where minimum possible DC voltage is applied. :
Initialisation
Firstly, the initialisation technique in ATP-EMTP for such system is to be established. In Fig. 10.1 a) the
basic simplified system layout is shown where power to two induction machines are supplied by a syn-
chronous generator. Assuming one induction motor is in full loading and the other is in very light loading
conditions, the following mode of initialisation is applied as in the attached data file (Dat10-01.dat). :
- As all universal machines are to be initialised in uniform initialisation mode, slip conditions are given
to the two machines.
- Voltage amplitude and phase angle are given to the synchronous generator terminal.
- No Fix Source nor Cao Load Flow option is specified, as these are not suitably applicable to uni-
versal machine by the author’s experiences.
- In Fig. 10.1 a), AVR is not in service and the switch for IM2 is permanently closed.
Some calculation results are shown in Fig. 10.2, where, for convenient comparison purpose, 2P and 4P
machines’ variables are shown in common graphs in relevant respective scale sizes. In a) each ma-
chine’s velocity is a little by little increasing during the transient calculating time interval. In b) the torques
of two machines is mostly steady especially after 0.2s. A little bit of miss matching in the initialisation
seems to be introduced by the calculation. Especially by the initial stage of the torques (in Fig. 10.2 b))
such phenomenon is expected. Nevertheless, the initialisation method seems to be suitably applicable to
transient calculations, especially, of time interval of up to several seconds.
Fig. 10.3 One motor is starting during the other is in full load operation
applied to the motor, and 100% of initial slip value is given to the motor. Then the starting switch is closed.
For details see the attached data file (Dat10-02.dat).
Calculated results are shown in Fig. 10.3. Shortly speaking the results show typical “voltage collapse.” In
a) the bus voltage is being collapsed gradually during starting of MOT2. By the starting current of MOT2,
the generator’s terminal voltage drops, so, for keeping MOT1’s torque constant, MOT1’s current in-
creases as shown in d). The bus voltage furthermore drops. MOT1’s velocity can never been kept as
shown in b), while acceleration of MOT2 is very low. By the voltage drop, the generator supplies less
power, i.e. less air gap torque as shown in c), the generator accelerate gradually as in b) due to the
constant mechanical input torque. As the result, MOT2 can never start appropriately in the system.
3
Application of AVR
The most important requirement in the system above is to keep the voltage. AVR is the first priority for
the purpose. So, let us introduce AVR to the generator. In chapter 7 (See Fig. 7.12) AVR was discussed.
In this chapter, the same AVR (but without PSS) is introduced. Chapter 7 relates to very high capacity of
generators, but for simplification, the same one is applied also to relatively low capacity of generator in
this chapter. For details, see the attached data file (Dat10-03.dat).
Fig. 10.4 Induction motor starting in a system supplied by an AVR furnished generator
Calculated results are summarized in Fig. 10.4. a) shows generator terminal voltage, where, though
voltage drop of short time interval appears at the initial time, the voltage is kept approx. constant value.
The generator exiting voltage, which is the output of AVR shows (in b)) the initial steep enhancement
and the following approx. constant value of 250% of the original one during the motor starting time in-
terval. After the start has been established, the value comes back to approx. the original value. The ex-
iting current (in c)) shows the similar variation.
d) shows the motor (MOT 2) started normally. But the system frequency, i.e. the synchronous genera-
tor’s velocity, lowered a little. The generator air gap torque, shown in e), due to the enhancement of the
field exiting voltage, enhanced a lot during the starting, whereas the mechanical input torque is kept
4
Fig. 10.5 Variables in earthed neutral system in both motor and system sides
a) Reference voltage wave --- initial part b) Reference voltage & tri-angler wave
in TACS in TACS
Fig. 10.6 ------ continue to the next page -----
With significantly high value of zero sequence reactance in the reactor between the source and the rec-
tifier bridge in Fig. 10.1, and with, also, high impedance for the capacitor neutral earthing (at CAPN),
5
c) Motor terminal phase-to-phase voltage d) Same as the left but enlarged time resolution
some calculation results are shown in Fig. 10.6. For details of the circuit parameters, compare the at-
tached data files (Dat10-04.dat and Dat10-05.dat. Dat10-06.dat is only for fine time resolution output.).
The followings seem to be noted for each figure. :
a) 3-phase reference voltage wave shape is shown for the first 1 second, the amplitude and frequency
of which are linearly rising to the certain specified values.
b) Tri-angular carrier and 3-phase reference voltage wave shapes in TACS are shown for the inter-
mediate time. By the comparison of the waves the gate signals to the inverter valves are produced
in TACS for producing correct phase-to-phase voltage to the motor. For details, see sub-chapter
9.3.
c) Motor terminal phase-to-phase PWM inverter output voltage is shown for the first 2 second.
d) Same as c), but in very fine time resolution representation for the intermediate time interval.
e) Generator and motor velocity changes by VVVF starting are shown in comparison with direct start-
ing ones. By similar starting performances of both cases, VVVF brings far less influence to the
system, i.e. less descending in synchronous machine velocity/frequency.
6
f) Time-integration of the starting motor’s torque, i.e. the area below the torque-vs.-time curve is to be
equal for both cases. Nevertheless, generator’s torque curves shown great difference between two.
Significant electrical loss is to be produced, most provably by the winding’s joule loss.
g) Instead of the mostly similar starting characteristics by both, the starting motor currents show great
difference each other. This is the most typical feature of VVVF starting of induction machine, i.e.
highly efficient starting.
h) Active, inductively reactive and apparent powers during starting by both direct and VVVF startings
are compared in the figure. (For the output of VVVF starting, due to non-symmetrical three-phase
variables creating lot of fluctuations, the outputs are smoothed in TACS.) Great energy saving in
VVVF, especially for starting is significant. Also little reactive power is consumed. Within the inverter
circuit, reactive power can be produced.
Fig. 10.7 shows generator’s field exiting
voltages by both direct and VVVF starting,
both are controlled by AVR. By VVVF start-
ing, high response of AVR seems to be al-
most un-necessary.
Fig. 10.8 shown circuit layout applied. The followings are to be noted. :
- In sub-chapter 9.4, detail of one-phase cyclo-converter is explained. Three of the same converter
systems are applied for driving 3-phase synchronous machine, the ratings of which are 3.3kV, 15Hz,
1MVA, 6P, etc.
- Transformer secondary side (converter valve side) is to be non-solidly earthed condition in each
phase. Therefore, 2 sets of high-ohmic resistor earthed star windings are applied for each phase as
shown in the figure. The primary side could be a common one set of 3-phase winding. In the case,
for simplification, three sets of star connected windings are applied for 3-phase.
- 3-phase reference voltages are to be given to 3-phase converter controlling (in TACS).
- For initialisation very fine tuning is required, especially between the reference voltage and the initial
machine terminal voltage, regarding the amplitude and phase angle.
The first example is to apply sudden mechanical load to the rotating motor in almost no-load condition.
The initialisation and transient calculation process applied is (as the most simplified one):
- Initially the motor is disconnected from the cyclo-converter.
- Automatic initialisation is highly recommended for synchronous machine. The motor is rotating in
very lightly loaded generator mode, i.e. giving the terminal voltage with the relevant frequency, and
high-ohmic resister is connected to the terminal of the machine for the purpose of easy and proper
initialisation.
- As the next step, the machine is connecter to the cyclo-converter source for motor operation.
- Then afterwards, sudden mechanical load is applied such like in rolling machine by means of TACS.
For details, see the attached data file (Dat10-12.dat).
Fig. 10.9 Cyclo-converter driven synchronous motor ----- continuing to the next page
amount of high-frequency component. The current amplitude changes a lot depending on the load
condition.
c) Fourier spectrum of the current clarifies less high-frequency component.
9
i) Current & voltage --- highly loaded j) Current & voltage --- load ejecting
d) Suddenly applied mechanical torque (TACS controlled) and calculated air gap torque are shown.
Due to sudden application and ejection of the torque, significant swing of the air gap torque is pro-
duced.
e) By the swing, the motor current (in total time range) changes a lot.
f) The swing is observed in, also, rotor position angle and velocity.
g) As shown in chapter 9, cyclo-converter driven system consumes a lot of reactive power. In many
cases, compensation facilities (capacitor bank) are required.
h) Positive and negative polarity converter bridge work well at turning over.
i) At highly loaded instant, the power factor seems to be high. Fig. g) clarifies this.
j) At the load eject instant, the current drastically changes, especially, in phase angle. Turning over
between the positive and negative bridges seems to be suitable from the current wave shape.
In the next example, quick starting of cyclo-converter driven synchronous motor is taken up.
In chapter 6 --- Appendix 6.2, synchronous machine starting as induction machine is demonstrated,
where, to represent short-circuited field coil, very low voltage is generated while initialisation. Now, in this
case, properly exited synchronous machine is to be driven by cyclo-converter. Then, the following
initialisation process is to be applied. :
- The machine’s initial velocity is to be as low as possible within restriction of ATP-EMTP synchronous
machine initialisation menu. In this case, 0.5 Hz is applied.
- The motor-internally generation voltage is to be proportional to the velocity, i.e. 0.5 / 15 = 0.033 times
of the rated voltage where the rated frequency of the motor is 15Hz.
- The applied voltage is to correspond to the machine induced voltage. So, cyclo-converter output
voltage is to be linearly rising frequency and amplitude one, corresponding to linearly rising velocity.
For details of the input data file, see attached data file (Dat10-13.dat)
stability study seems to be necessary. Powerful feed back control system may be effective.
h) Active power, reactive power, apparent power and power factor measured at the power frequency
inputting point are shown. At t = 0.9s, when the motor torque is highest, the power factor is still not
so high, i.e. around 0.3p.u. For such reason, compensation facilities (capacitor bank) are installed
by some cyclo-converter systems.
11
Comparison with inverter driven system will be shown in the following cases.
Firstly, “sudden mechanical load application” is introduced to inverter driven system. In the inverter
driven system, mostly equal to one in the previous sub-chapter is applied, i.e. the inverter circuit layout in
Fig. 10.1 c) is applied. As for the detail, see attached data file (Dat10-15.dat). The reference voltage to
control inverter output is identical to one in the cyclo-converter.
a) Inverter output voltage & current b) Mechanical & air gap torques
In the next, quick starting by inverter driven is taken up. The reference velocity/voltage condition to
control the inverter is identical to cyclo-converter case. Great care should be taken that, by inverter,
especially for very low velocity of condition, the total system tends to be unstable. The phenomenon is
observed in Fig. 10.12 d). The cause of the un-stable seems to be mechanical and / or electric circuit
condition of the system. Detail has not been clarified by the author. In the following exercise, the resis-
tance value in the reactor between the inverter and the motor is doubled compared to the case by
cyclo-converter. For detail, see attached data file (Dat10-15.dat). Otherwise, the motor performs
un-stableness. The reader should try the case by “Dat10-1X.dat”, where the resistance value is equal to
12
c) Inverter voltage and motor current d) Air gap torques by both kind of source
a) Mostly identical starting characteristics are obtained by both (inverter and cyclo-converter) source
driven systems, though in inverter, the connection reactor’s resistance value is doubled.
b) Motor input currents are compared between both cases. By identical reference voltage conditions
by both, the actual and effective output voltage may more or less differ each other, as details of the
voltage wave shapes, including high frequency components, are different by each other.
c) When the maximum torque is created, the motor current is of very high power factor such like in
cyclo-converter driven case. After establishing the velocity (not shown), however, the motor current
power factor is low. Still the power factor of the power from the power frequency source system is
high as shown later.
d) From the air gap torque characteristics, the following is clear.
Inverter driven tends to be unstable in very low velocity region but stable in higher velocity region.
Cyclo-converter driven tends to be opposite, i.e. stable in low velocity and un-stable in higher ve-
locity. As the countermeasure for mitigation, powerful feed back system could be applied.
e) Active, (inductively) reactive and apparent powers are compared between both systems. Inverter
system’s high power factor feature is remarkable.
f) In power factor graph, also, the tendency is clearly shown. By inverter, power frequency source side
13
power factor is always approx. 100%. While, by cyclo-converter, the power factor is only more or
less than 20%. By inverter, though the motor current power factor is not always high, the power fre-
quency side one can always be high. This means inverter can provide reactive power by itself. On
the other hand, by cyclo-converter, even though the load current power factor is high, power fre-
quency side power factor can never be high enough due to the frequency converting principle.
As for cyclo-converter system with relatively low const, low power factor in consuming power, necessity
of capacitor bank, limitation in output frequency, and as for inverter system with higher cost, high power
factor, non-necessity of compensation, possibly higher output frequency, not only qualitative but also
quantitative comparison could be exercised by ATP-EMTP simulation, such as shown above.
10.3 Fly-wheel generator ---- Doubly fed machine application for transient stability enhancement
As shown in chapter 8 (in the final part), doubly fed machine can produce, though for relatively short time
interval, both active and reactive powers. In chapter 7 transient stability phenomena is explained with
relation to energy balance in the relevant power system. From these combined, applying doubly fed
machine as fly-wheel generator to power system, transient stability enhancement effect is expected.
In Fig. 10.13 shows power system layout for analysing such effect in single line diagram.
Fig. 10.13 Single line diagram of fly-wheel generator equipped one generator vs. infinitive
bus system for analising transient stability enhancement
In the figure, one generator vs. infinitive bus power system is identical with one in chapter 7 (Fig. 7.1).
The doubly fed machine as a fly-wheel generator is identical with one in chapter 8. The current regulated
inverter to energise the doubly fed machine rotor is identical with one in chapter 9.
In the figure, infinitive DC voltage source is applied for the power source of the current regulated inverter.
In actual cases, the DC source energy is to be supplied from the main power system. So, the actual ef-
fect of the fly-wheel generator may be increased/decreased depending on the velocity of the doubly fed
machine. Please refer Fig. 8.9 and Appendix 8.1 in chapter 8. It should be noted that the DC supplying
system for the inverter is to be bi-directional, i.e. rectifying/re-generation system, as in certain operating
state the doubly fed machine rotor supplies energy towards the inverter side.
For controlling the fly-wheel generator to absorb/exhaust energy, usage of the information from the as-
sociated bus voltage is thought to be realistic. So in the case, frequency change of the bus voltage is
pick up and applied to control the inverter, i.e. the primary side absorbing power (both active and ca-
pacitively reactive) is set to be proportional to delta F. The detailed controlling algorism is shown in Fig.
8.11. This output is applied as the reference current of the current regulated inverter.
Initialising
Though each component in Fig. 10.13 could be appropriately initialised in each respective mode, com-
bining plural components, an unified mode of initialisation is to be applied according to the restriction of
ATP-EMTP. In the case, “step by step” and/or “try and error” procedures seem to be convenient.
Synchronous generator: In chapter 7 where only synchronous generators are applied, CAO LOAD
FLOW option is quite appropriately applied for each case initialisation. However, in the case with also
universal machine(s), the option has not been successfully applied. Therefore, in this case, another
mode, which is compatible with also universal machine, is to be applied to the synchronous generator
initialisation. As shown in sub-chapter 10.1, where synchronous machine and cage rotor induction ma-
chine (universal machine) exist in a common system, inputting terminal voltage amplitude and phase
angle for the synchronous machine, and slip value for the cage-rotor machine produces appropriate ini-
tialisation result. So, in principle, let’s try the same mode of initialisation.
In the first step, the one generator vs. infinitive bus system (without fly-wheel generator) is automatically
initialised applying CAO LOAD FLOW option. For details, see Dat10-21.dat attached. The precise initial
generator terminal voltage amplitude and phase angle can be obtained.
15
opening).
Applying FREQUENCY METER to the bus
voltage where the fly-wheel generator is
connected to, and after some mitigating
processes in the calculation, Fig. 10.18’s
result is obtained. For details of the calcula-
tion, see Dat10-26.dat attached.
Due to the sudden change in the voltage
during 0.3s --- 0.4s (3LG, Fault clearing and
1 circuit opening), the rapid frequency
change in this time interval may be better to
be excluded. However, rather steady fre-
quency change output is obtained and suit-
able application to control the flywheel gen-
Fig. 10.18 Bus voltage frequency change erator is expected.
during 1LG --- 1cct opening
Fly-wheel activity in transient stability
enhancement
In the first trial, both active and reactive power outputs of the fly-wheel generator are set to be equal and
proportional to the frequency change of the HV bus voltage, i.e. by increase of the frequency, the ab-
sorption of the active power and capacitively reactive power increase. Thus, the synchronous genera-
tor’s acceleration is expected to be damped. The maximum power of the fly-wheel generator is set to be
approx. 200MVA, i.e. 200% loading due to short time interval. Some results are shown in Fig. 10.19.
a) Except the violent transient interval (3LG & clearing), fly-wheel rotor (secondary coil) current is well
appropriately supplied from the inverter.
b) According to the bus voltage frequency change (Fig. 10.18) and along the vector control algorism,
the fly-wheel is driven to absorb power, enhancing the velocity, which is shown in Fig. 10.19b.
c) The figure shows the powers are well controlled. The same value of active and reactive powers is
shown.
d) The synchronous generator’s d-axis angle (swing) during the transient is damped by the function of
the fly-wheel. Due to the relatively low power output of the fly-wheel generator (as for active power,
approx. 13% of the synchronous generator’s), the damping rate is limited.
e) The figure shows top (plus side) and bottom (minus side) valve currents in a certain phase of the
inverter. For certain time interval only one side valve is ON. This means the DC source voltage is
critical and can never be lower.
f) The HV bus voltage is lower during the swing. Therefore, according to equation 7.1 in chapter 7, the
transmitting power is lower, resulting in excess source side energy. Absorbing higher capacitively
reactive power to enhance the voltage, further damping of the swing may be expected.
c) Synchronous generator swing comparison d) Velocity and air gap torque comparison
Fig. 10.20 Effects by solo-active power and jointing with reactive power
active and reactive powers, the condition is easily introduced. For details, see Dat10-28.dat attached.
Comparing to the previous case, some results are shown in Fig. 10.20.
a) Rotor currents from the inverter are compared between two cases, where due to approx. equal
apparent powers, the maximum crest values are approx. equal by two. On the other hand the phase
angle is shifted relevantly. These can be explained by Fig. 8.9, Fig. 8.11 and Appendix 8.1.
b) Active, reactive and apparent powers of the fly-wheel are shown. Due to the almost zero reactive
19
power, the value of the apparent power is equal to the active one, which is approx. equal to the
previous case’s apparent one (Fig. 10.19c).
c) Effects on the synchronous generator’s swing (d-axis angle, which is the representative of the
transient stability) are compared. In spite of by approx. equal apparent powers, solo-active power
control is not so effective compared to by both active and reactive power control. Discussion will be
shown later.
d) By solo-active power, velocity change and air gap torque are higher, though the effect is less.
Discussion
In chapter 7, transient stability in the identical one
generator vs. infinitive bus system is explained.
Also significant effect of AVR/PSS to enhance
stability is shown. In Fig. 10.21a the effects by
fly-wheel generator and AVR/PSS are compared
under similar initial load flow conditions. The ef-
fect by AVR/PSS is apparently superior. By
AVR/PSS the synchronous machine’s exciting is
controlled, yielding enhancement of the trans-
mission voltage. Therefore, by the increase of the
transmission power according to equation 7.1,
the air gap torque rises as shown in Fig. 10.21b.
a) Synchronous generator’s swing The maximum torque is 135% of the initial, and
the difference from the initial one, due to the
constant mechanical input torque, acts to damp
the swing. By the fly-wheel, the maximum torque
is 118% which is limited by the fly-wheel rating
(over loading included). Moreover, the rise by
AVR/PSS is fur more quick. As the result,
AVR/PSS is more effective in this case.
Nevertheless, depending on the system layout,
synchronous generator’s and fly-wheel genera-
tor’s ratings, fault conditions, etc., various results
are obtained, with possible superiority in
b) Synchronous generator’s air-gap torque fly-wheel.