1.
A NOVEL SETTING-FREE POWER SWING
DETECTOR
Gerhard Koch
Siemens AG, Nuremberg, Germany
gerhard.koch@siemens.com
SUMMARY
Power swing detectors prevent distance relays from maloperation under power system out-of-step conditions.
The traditional swing detection approach is to measure
the time, which the traveling swing impedance trajectory
needs to pass through two impedance thresholds, e.g.
blinders.
This method has deficiencies with power swings and
simultaneous power system faults as well as with high
swing frequencies. Apart from these difficulties proper
swing detection settings require comprehensive power
system studies. The power system is dynamic, it changes
its parameters with the system configuration, generation
and system load.
This paper describes a novel adaptive out-of-step
detection algorithm based on dissimilar measurement
approaches. Two superimposed measurements enable
the algorithm to detect fast as well as slow power
swings. Both methods recognize power swings right
from the beginning of the swing impedance movement.
The first principle is based on continuous rate-of-change
measurement of the traveling swing impedance. The
second method continuously monitors the speed and the
shape of the swing impedance trajectory.
The algorithm detects extremely high swing frequencies
of up to 7 Hz and maintains functionality under singlepole open conditions, i.e. auto-reclose dead time. The
associated distance relay detects all kinds of internal
faults, e.g. 3-phase, 1-phase and 2-phase also in the
presence of power swings.
Contrary to the traditional approach, the power swing
detector does not need any set points or any power
system stability calculation. Thus it also contributes to
simpler and setting-free relaying.
KEY WORDS
Power swing blocking - Power swing tripping - Out-ofstep - Power swing - Distance relay.
1. POWER SYSTEM SWINGS
A sudden change of load in the power system caused by
a fault, by a disconnection of line load or auto-reclosure
forces the generators to adjust to new load conditions.
Due to the generator inertia, the adjustment may not
happen instantly, but rather as transient oscillation. This
causes the current and the voltage of the power system
to change, with oscillation of amplitude and phase.
With stable systems, this oscillation will be damped and
the generators remain in service. In extreme cases the
power swing may be so large that it causes unstable out
of step conditions, in which case the grid needs to be
separated.
2. IMPACT OF POWER SWINGS ON DISTANCE
MEASUREMENT
The fluctuations of voltage and current in case of power
swings make it difficult for the relay to discriminate
between 3-phase faults and power swings. Impedance
calculations based on these measurement quantities
suggest similar conditions as under system faults. Thus,
the swing impedance trajectory may enter the fault
detection zone or even the instantaneous impedance
zone of distance relays, Fig. 1.
With stable swings, the swing impedance trajectory
returns to the actual load impedance locus. Thus all
distance relays in the power system subject to swings
need to be securely blocked for the time the swing
impedance remains within the distance relay
characteristic in order not to disrupt the power system
integrity.
A controlled trip in a out of step situation can be
desirable at the electrical center of the power system
during unstable swings. The relay may then divide the
power system into separate stable subsystems. The logic
of the relay in the electrical centre distinguishes between
stable power swings where the system recovers and an
unstable out of step conditions where the grid needs to
be separated.
3. TRADITIONAL POWER SWING DETECTION
The swing impedance moving along its trajectory needs
some time to travel through e.g. the two blinders in
Fig.1. Its trajectory speed is slow compared to the
sudden impedance jump when faults occur. With faults,
the impedance jumps instantly from load to fault
impedance.
Traditional power swing detection is based on the time
Δt that elapses as the traveling swing impedance
trajectory enters and leaves two thresholds (circles or
blinders). If the time, the swing impedance requires to
pass through the two impedance set points is longer than
a set time Δt, the swing detector will block the distance
relay’s impedance zones.
Blinder
B
Line ΔZ
X
In conclusion, new swing detectors and their associated
distance relays must
Blinder A


Instant jump
from load to
fault impedance

Power swing impedance
trajectory
trajectory
Δt
R
Load
Prior to swing
Figure 1: Fault and swing impedance trajectory

cover extremely fast swings frequencies
maintain operation during under open pole
conditions, i.e. single-pole auto-reclose dead
time
clear all kinds of internal fault during power
swings
be virtually setting free
All these requirements cannot be met by one single
measurement approach. Only an adaptive approach can
cope with these numerous and changing measurement
conditions the relay is exposed to.
5. ENHANCED SWING DETECTION
4. NEW CHALLENGES
Interconnected power systems cause large geographic
areas to be affected by power swing conditions. With
further expansion the system fault probability also
increases and the number of simultaneous faults, which
make the grid system vulnerable to wide-area outages.
Particularly grid systems interconnected by long
transmission lines and fed by huge hydro power stations
are exposed to power swings having extremely short
frequencies down to some 7Hz, i.e. 150 ms.
The present technology for power swing detection is not
capable of fully coping with all the protection
requirements of large grid systems.
The common predictive method to determine loss of
synchronism is the Equal-Area Criterion. It assumes that
the power system behaves like a two-machine model
where one area oscillates against the rest of the system.
In reality a power system is more complex and changes
its parameters over time. The time Δt as criterion does
not fully cover all possible situations. The set points are
fixed and do not adapt to power system changes.
Finding proper settings for traditional swing detectors is
not simple and often requires comprehensive grid
studies. If the study does not consider worst case
conditions, then the relays may lack security.
When single pole auto-reclosure is applied, the swing
detector may not assume symmetrical conditions during
the auto-reclose dead time. A more comprehensive logic
is required to cope with open pole conditions in the
presence of power swings.
Another challenge is the clearance of line faults during
power swings. The swing detector blocks the ordinary
distance processing under those conditions. Thus, the
distance relay itself requires extra provisions to cover
faults during power swings. E.g. for all unbalanced
faults the negative or zero sequence current or voltage,
which are not affected by power swings, may be used for
line fault calculation.
Three phase fault detection is more challenging.
Nevertheless there are differences between swing and
fault quantities for fault detection. Swing quantities
oscillate, while fault quantities remain constant as long
as the fault prevails in the power system.
The enhanced approach utilizes virtually all criteria that
distinguish between fault and swing conditions. It is
based on two dissimilar principles and a number of
plausibility checks. Each one is ideal for its specific
conditions, i.e. for fast or slowly moving swings along
the impedance trajectories.
5.1 Rate-of-change in Swing Impedance
The speed of the resistance dR (k) moving along the
fault or swing trajectory is one criterion to distinguish
between fault and swing conditions, Figure 2. The swing
impedance movement is relatively slow when compared
with the instant jump from load to fault impedance
during system faults.
x
Fault Impedance
Jump from load to
fault impedance
dR(k-n)
dX(k-n)
swing trajectory
dR(k)
dX(k)
Load impedance
R
Figure 2: Rate-of-change in swing impedance
X
The continuity and monotony monitors detect power
swings 30 ms after the impedance begins to travel, even
before the moving swing impedance trajectory enters the
relay polygon PPOL, Fig. 4.
0°
estimated centre
+90°
-90°
x
PPOL
unstable
area
Swing Trajectory
-180°
+180°
R
reversal on stable
power swing
R
Figure 3: Steady state instability range
The continuous dR/dt rate-of-change measurement along
the trajectory covers slow swing impedances with high
accuracy. If a number of consecutive measurements
exceed a threshold a power swing is detected. This
method is optimized for detection of slow swing
impedance movements < 5 Ohm/s.
5.2 Space Vector Estimation
The space vector estimation is based on speed
estimation and a shape analysis of the impedance
trajectories
derived
from
some
consecutive
measurements fig. 4. Under swing conditions the
impedance vectors describe an elliptical trajectory. By
analyzing this ellipse with its estimated centre, one can
also distinguish between stable and unstable swings.
The swing trajectory and its associated centre are
continuously assessed with measurements at consecutive
points in time. Thus, any change in the trajectory shape
and swing speed is recognized. This method enables
detection of high slip frequencies of up to 7 Hz.
Swing polygon
set > Fault detector
°°° instants of measurement
Figure 4: Trajectory shape and speed estimation as well
as monotony check
When the swing impedance enters the relay polygon, the
detector blocks the distance relay as per set selection. If
for two consecutive calculations no continuity prevails,
a system fault is assumed. Thus, one can also detect
power system faults occurring simultaneously with
power swings.
I/A
5.0
2.5
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.0
1.1
1.2
1.3
t/s
0.0
-2.5
5.3 Plausibility monitoring
-5.0
Strom iL1
Strom iL2
Strom iL3
Strom iE
U/V
75
50
For plausibility, the impedances of three fault loops are
continuously monitored for continuity, monotony,
symmetry and stability.
25
0
0.1
0.2
0.3
0.4
0.6
0.7
0.8
0.9
t/s
-50
-75
Spannung uL1
Continuity means, the phenomena must prevail for at
least 6 consecutive measurements; monotony prevails
when the impedance trajectory does not change its sign,
i.e. direction. Monotony only need to be checked as long
as the impedance has not yet entered the relay polygon.
Inside the polygon the impedance trajectory may
reverse, if the power swing returns to normal service
conditions.
0.5
-25
Spannung uL2
Spannung uL3
Spannung uen
>Meldung 1
>Meldung 2
>Meldung 3
>Meldung 4
Ger.Anr. L1
Ger.Anr. L2
Ger.Anr. L3
Ger.Anr. E
Ger.AUS L1
Ger.AUS L2
Ger.AUS L3
EF Anr 75%Ie
Not G-Anr
Res G-Anr
WE nicht ber.
WE EIN-Kom
Dist.Anr L1
Dist.Anr L2
Dist.Anr L3
Dist.Anr E
Dis Anr vorw.
Dis Anr rück.
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
t/s
Figure 5: Three phase fault detection under swing
conditions, lower trace indicate relay tripping
6. SYSTEM FAULTS SIMULTANEOUS WITH
POWER SWINGS
Trajectory stop
Figure 6: Trajectory-continuity, three phase fault
detection under swing conditions
In addition to symmetrical component evaluation, phasesegregated detectors evaluate all 3 phase loops
separately. If there is no fault in the system, these three
calculated impedances must be symmetrical, if one pole
is open; only the remaining 2 loop impedances must be
symmetrical.
Trajectory stability- When the impedance trajectory
enters the relay’s swing polygon, the power system must
be at “steady state instability”. Thus, the swing detector
estimates the swing ellipse center and checks whether
the calculated impedance at the point of entry is closer
to the origin than the center, Fig.3.
Once the power swing is detected it remains picked-up
until the swing impedance vector leaves the power
swing polygon (PPOL). This is unless a fault occurs
during this time. The detection of a jump in the
trajectory or non-symmetry of the trajectories will reset
the power swing blocking condition.
The power system becomes particularly vulnerable, if
system faults occur simultaneously with power swings.
The operation of traditional relays subjected to power
swings is blocked in this very instant. 3-phase, 2-phase
and ground faults, such as cross country faults or
intersystem faults on multi-circuit lines may cause
system instabilities. This applies to long fault clearance
times and particularly to breaker failures, where the
breaker fail relay or the up- and downstream relays and
breakers must remove the fault from the system. Thus
modern relays must have provisions for dependable fault
clearance even under these unusual conditions.
6.1 Ground faults
The distance relay already has a special ground fault
detection operative during the single pole dead time.
This measurement is based on sequence component
currents. The ground fault detector also modifies the
swing detector during single pole open conditions.
5.4 Open Pole Conditions
Open pole situations prevail within the dead time of
auto-reclose cycles of the associated, and also parallel or
adjacent lines. Only two of the three phases show a
swing impedance trajectory. The swing detector must
also consider that load may also be transferred via the
zero sequence system.
On single pole open conditions the detector reverts to
the single pole dead time mode. It distinguishes between
power swings and two phase to ground faults elsewhere
in the system during the dead time.
A comprehensive logic and measurement secures
reliable swing and fault detection under these complex
conditions. For one, the detector distinguishes between
symmetrical (load, three phase faults and power swings)
and non-symmetrical conditions with unbalanced faults,
by symmetrical component evaluation. Phase symmetry
is a prerequisite for the release of the swing detector.
During open pole conditions this applies for the two
remaining phases.
Figure 7: Power swing detection during one pole open
situation
6.2 Directional polarization with phase-to-phase
faults
Under swing conditions directional polarization of
distance relays with healthy phase or memorized
voltages is not always applicable. The angle shift after
swing inception is not predictable and may assume any
value from 0 to 360 degrees.
I/A
2 .5
0 .0
0 .0 0
0 .2 5
0 .5 0
0 .7 5
1 .0 0
1 .2 5
1 .5 0
1 .7 5
2 .0 0
2 .2 5
2 .5 0
2 .7 5
3 .0 0
2 .2 5
2 .5 0
2 .7 5
3 .0 0
t/s
- 2 .5
- 5 .0
S tro m iL 1
S tro m iL 2
S tro m iL 3
S tro m iE
U /V
100
50
0 .0 0
0 .2 5
0 .5 0
0 .7 5
1 .0 0
1 .2 5
1 .5 0
1 .7 5
2 .0 0
t/s
0
-50
-100
-150
S pannung uL1
Spannung uL2
Spannung uL3
S pannung uen
> M e ld u n g 1
> M e ld u n g 2
> M e ld u n g 3
> M e ld u n g 4
G e r.A n r. L 1
G e r.A n r. L 2
G e r.A n r. L 3
G e r.A n r. E
G e r.A U S L 1
G e r.A U S L 2
G e r.A U S L 3
E F A n r 7 5 % Ie
N o t G -A n r
R e s G -A n r
W E n ic h t b e r.
W E E IN -K o m
D is t.A n r L 1
D is t.A n r L 2
D is t.A n r L 3
D is t.A n r E
D is A n r v o rw .
D is A n r rü c k .
Fig. 9 shows an example of a correctly cleared close in
single-phase to ground fault during a power swing (with
CVT transient). The distance relay remained stable on
reverse bus faults. The fault current was smaller than the
swing current although this was a close in fault.
The negative sequence direction measurement does not
allow phase-segregated measurement of the fault loops.
Thus it is only applied under these difficult conditions
where the fault voltage is too small for a secure direction
determination.
7. CONCLUSIONS
0 .0 0
0 .2 5
0 .5 0
0 .7 5
1 .0 0
1 .2 5
1 .5 0
1 .7 5
2 .0 0
2 .2 5
2 .5 0
2 .7 5
3 .0 0
t/s
Figure 8: Swing detection during one pole open situation
The phase angle is predominantly dependant on the
relay’s location relative to the “electrical centre” of the
power system. The actual faulted loop voltage returns a
correct direction decision. On close in forward faults
and reverse bus faults this voltage is zero which again
renders it unsuitable for a direction decision.
To cope with these situations, phase currents and phase
to ground voltages are used for negative sequence
current I2 computation and further for the negative
sequence impedance Z2 = V2 / I2.
The negative sequence impedance is not affected by
power swings. It indicates the relative position of the
relay to the fault locus. On reverse bus faults, the
downstream negative sequence source impedance is
seen by the relay, while it sees on forward faults the
upstream negative sequence source impedance.
i L 1 /A
0
-2
1,85
1 ,9 0
1,95
2,00
2,05
2,10
2 ,1 5
2,20
2 ,2 5
2,30
2 ,3 5
1,85
1 ,9 0
1,95
2,00
2,05
2,10
2 ,1 5
2,20
2 ,2 5
2,30
2 ,3 5
1,85
1 ,9 0
1,95
2,00
2,05
2,10
2 ,1 5
2,20
2 ,2 5
2,30
2 ,3 5
t/s
i E /A
0 ,5
t/s
0 ,0
-0 ,5
u L 1 /V
50
0
-5 0
t/s
-1 0 0
Figure 9: Single phase-ground fault during power
swings, top = phase L1 current, middle = E/F current,
bottom = L1-E voltage
Because of economy in system design and operation,
many systems today have smaller margins than ever
before between normal operating conditions and
transient stability limits. Thus, there is a need for a
reappraisal of to-day’s out-of-step relaying practices.
The fact remains that it is complex to assess all
combinations of events which may cause an out-of-step
and simultaneous fault conditions.
This paper has presented a novel patented power swing
algorithm which detects swings and even simultaneous
power system faults under extreme conditions. The
adaptive approach of the new generation of out-of-step
relays utilizes more dissimilar measurements and
plausibility checks to cope with all system changes and
peculiar measurement conditions.
8. REFERENCES
[1] CIGRE Working Group SC34-WG04, Application
Guide on Protection of Complex Transmission Network
Configurations, Paris, 1991
[2] 7SA52 Relay Instruction Manual, Siemens AG,
Nuremberg, 2002
[3] Ziegler, G., Numerical Distance Protection: Berlin
and Munich, 1999
[4] Steynberg, G., Power Swing Detection, Siemens AG,
Nuremberg, 2001
[5] Holbach, J., New Out-of-step Blocking Algorithm,
30th Annual Western Protective Relay Conference
Spokane, Washington
[5] Koch, G., The Contribution of Relaying towards
Power System Stability, Southern African Power System
Protection Conference, Johannesburg, 2002