I A1 j ( 120º )
I w1 I B1 Ie
I w1 I B1 Current in relay from primary winding
I C1 I C1 Ie j (120º )
Ia 2 Then:
I w2 I b 2 Current in relay from secondary winding
I c 2 Ie j k aa k ab k ac Ie j 0 º
Depending of the transformer connection group it is I w2c Ie j ( 120 º ) k ba k bb k bc Ie j ( 120 º ) (4)
needed to compensate in magnitude and angle one of Ie j ( 120 º ) k ca k cb k cc Ie j (120 º )
the two currents, in order to get a balance before to
solve the differential algorithm. The regular practice
is to choose one winding as reference and the Being the phasor M ( ) :
compensation is made on the other winding. In this
particular case, we select primary winding as
reference and compensation is made on the e j0º e j e j
secondary winding currents. Doing this, we have the j ( 120º) j ( 120º) j j ( 120º )
M e M e M e e
general equation e j (120º ) e j ( 120º ) e j e j ( 120º )
I w 2c M ( ) I w 2 (3a) k aa k ab k ac e j 0 º
kba kbb kbc e j ( 120 º ) (5)
where:
k ca k cb k cc e j (120 º )
I w 2 c compensated currents from secondary winding
k aa k ab kac
With M ( ) kba kbb kbc ; We developed the equations:
kca kcb kcc
e j k aa e j 0 º k ab e j ( 120 º ) k bc e j (120 º )
I a 2 c k aa k ab k ac I a 2 j j ( 120 º ) (6)
M e e k ba e k bb e
j0º j ( 120 º )
k bc e j (120 º )
I w 2 c I b 2c k ba k bb kbc I b 2 ( 3b ) e j e j ( 120 º ) k ca e j 0 º k cb e j ( 120 º ) k cc e j (120 º )
I c 2 c k ca k cb k cc I c 2 e j k aa e j 0 º k ab e j ( 120º ) k bc e j (120º )
(7)
M e j k ba e j (120º ) k bb e j 0 º k bc e j ( 120º )
Being I w1 1 and I w 2 2 , the phase shift of e j k ca e j ( 120º ) k cb e j (120º ) k cc e j 0 º
primary and secondary currents is 1 2 with
winding 1 as reference. Then the non-compensated From (7) we can demonstrate that:
currents of winding 2, must be rotated by the angle
of to be in phase with currents from winding 1.
k aa k bb k cc k xx
k ab k bc k ca k xy ( 8)
k k k k
ac ba cb yx
e j k xx k xy k yx e j 0 º
M e j ( 120 º ) k yx k xx k xy e j ( 120 º ) (9)
Fig. 2 Phase compensation angle within
e j ( 120 º ) k xy k yx k xx e j (120 º )
-300 and -360 deg sector
From [1] the reference positive and negative currents k xx
1
k xy
2
k yx
form sections of 60º, then each current in the 0º , 60º cos sin 0 sin ( 12 )
3 3
secondary winding will be expressed in function of 60º , 120º 0 cos
1
sin cos
1
sin
3 3
the closer reference current. 120º , 180º cos
1
sin
2
sin 0
3 3
1 2
In case of phase A for sequence ABC 180º , 240º cos sin 0 sin
3 3
1 1
e j0º 240º , 300º 0 cos sin cos sin
3 3
M e j k xx k xy k yx
e j ( 120 º ) 300º , 360º cos
1
3
sin
2
3
sin 0
e j (120 º)
In summary, all compensated currents are obtained
1 3 1 3
from equation (3) using the coefficients from (12).
cos j sin k xx k xy .( j. ) k yx .( j. )
2 2 2 2
Note that if θ = 0º, then M(θ) is the identity
where,
matrix
1 Where:
k xx cos sin
3
I Bwi Base current for winding i
2 ( 11a )
k xy sin S B Base apparent power
3
k yx 0 U wi Nominal phase-to-phase voltage in winding i
I d M 1 (30º ) I w1 M 2 ( ) I w 2 ( 23 )
Figure 3. Some relay settings. Besides, (6) states that if the Y star-point is not
grounded, (i.e., no I0 removal is required):
When set in “Auto”, the chosen winding will be that
with lowest CT margin. The CT margin is defined 1 1
as follows: 0
3 3
I CT 1 1
I m arg ( 22 ) M 2 (30º ) 0
3
I nom 1
3
1
0
Where: 3 3
ICT = primary nominal CT current.
Inom = primary nominal protected transformer current
On the other hand, if the Y star-point is not
(based on the nominal voltage and nominal power
grounded, according to (9):
parameters set before).
In case of manual reference selection, SB in (13) M0(30º) = M(30º) ( 24 )
will be the nominal power configured in the
winding selected as magnitude reference.
Then, if θ = 30º, the in-zone grounding makes no
Grounding: for each winding, shall be indicated difference. This is valid for compensation angles
whether the winding is grounded within the of 90º, 150º, 210º, 270º and 330º (θ = (2n-1)∙30º,
protection zone or not. Those windings with IZG where n is any integer).
(In Zone Grounding) will require I0 removal by
The kind of flexibility described above is provided
software (i.e., matrix M(θ) will become M0(θ)).
by most of leading manufacturers with inbuilt
Phase shift angle respect to reference winding: algorithms where they follow unique
when configuring this IED, it is possible to assign characteristics. The equipment utilized to test these
arbitrarily the winding order, or indexes. Phase functionalities should be capable of adapting to any
shifts are expressed as negative angles (0º a -359º) algorithm.
measured with respect to winding 1 currents
although, internally, the phase compensation III. Test Conditions
algorithm (if enabled) selects as phase reference the
first delta winding. Once described the differential current calculation
For example, suppose an Yd11 transformer, with Y algorithm and the configuration parameters
winding configured as winding 1. Of course, angle involved, it is convenient to establish general
for winding 1 will be 0º, while for winding 2 (the guidelines to test percent restrain and instantaneous
delta winding) shall be set at -330º (equivalent to differential protection functions.
+30º). Internally, (user won't see it), phase
The test conditions referred in this section are those
reference will be winding 2. Then, currents in
to address where adjusting the injection currents to
winding 1 must be compensated for the -30º phase
emulate all faults types that should be cleared by
shift respect to winding 2. According to (2)
the relay so that, when analyzing the oscillography,
the event log and fault reports generated during
testing, to verify the correct relay performance
during faults.
For that, it is necessary to specify, for every fault
type, the injection conditions applicable in the test,
being the most common type of faults for testing:
single-phase, two-phase and three-phase faults.
In general, the goal of testing is the reproduction of
the characteristic differential curve adjusted. Then,
it will be necessary to inject currents in order to
produce currents Irest for each zone of the Irest - Id
curve, with values of differential current just below
and above the differential curve, checking non-
operation in the first case and trip in the second a) WYE Side
one.
So, the question to answer is: what relations must
be satisfied between currents on either side of the
transformer in each type of fault?: The answer will
depend on the compensation angle (θ), as stated in
(18).
As we demonstrate in previous paragraph, the
answer to this question will be derived using
equations (20) and (21). Of course, to make them
useful, reasonable criteria should be established for
determining the currents to be injected, addressing
testing conditions that provides useful information
to end-user.
To illustrate the process, we have made fault
simulations using EMTP software, using a DY1
200 MVA, 66&66 kV Power transformer as b) Delta side
example.
Figure 5. External single phase-to-ground fault
a) WYE side
I restF MAX I w1cF ,..., I wNcF ( 25 )
Where:
F subscript = phase (A, B or C).
|Iw1cF|,...,|IwNcF| =F phase compensated current
magnitudes in windings 1...N, respectively.
According to sections I y II,
• Magnitudes are compensated referred to the
winding (manually or automatically) selected
as magnitude reference.
• Angles are compensated referred to the
winding selected (by algorithm) as phase
b) Delta side
reference. Note that both references can be
Red (Phase A). Green (Phase B), Blue (Phase C) at different windings.
Figure 6. External phase-to-phase fault As a starting point, balanced load at both sides is
assumed, with magnitudes that ensures the desired
A. Differential Currents (Id)
Irest. Later, it will be determined the most
convenient way to modify the values, in order of
Id, made up of IdA, IdB and IdC, will depend on the
obtaining the desired differential current.
fault type to test and the characteristic curve to be
tested. Then, for the magnitude reference winding;
For single-phase faults, a simple criterion to define
differential currents is enunciated as follows:
I wXA I wXB I wXC I rest ( 26 )
For the faulted phase, the differential current In (26), X subscript applies to magnitude reference
must be slightly over or slightly below the winding's currents (this nomenclature will remain
characteristic curve, while differential currents below). Currents will be injected shifted 120º,
are maintained in 0 for other phases. according the phase sequence considered (ABC or
ACB).
IdA, IdB and IdC are phasor expressions, then
differential currents with arbitrary directions may It might seem that (26) is a general definition,
be obtained by varying their real and imaginary where it is needed to consider I0 removal, if
values. The angle, or direction, of these currents needed, in the magnitude reference winding. In
will be relative to the current phase reference that case, compensated phase current's magnitude
winding. will be reduced to two-thirds of injected current. It
means that Irest will be two-thirds of injected
This approach is extendible to multi-phase faults currents.
(bi-phase and three-phase ones). In these cases,
more than one differential currents must be It is possible to overcome this issue reformulating
different to zero. (26) as follows:
B. Line Currents 3
I wXF I rest ( 27 )
2
Less obvious is the selection of Iw1A, Iw1B, e Iw1C
(Iw1) and Iw2a, Iw2b, e Iw2c (Iw2). I d M 1 (0º ) I w1 M 2 ( ) I w 2 ( 28 )
The first addressed constraint is obtaining the Irest F can be replaced by A, B or C, as needed.
for every provided test point.
Phase and magnitude compensated currents of
For this, it is necessary to know the relationship other windings must match Irest magnitudes,
between phase currents and restrain currents for according to equation (13). These currents will be
each winding. In case of T60 relay, equation (25) 120º shifted from each other plus the angle θ
defines that relationship respect to the currents of phase reference winding.
Under this circumstance, differential current is 0, If we apply ΔI to only one phase, it will be
and restrain current is the desired Irest. generated differential currents in all three phases
(positive two-thirds of ΔI for the same phase, and
The next question to answer is: what magnitude or
negative one-third for other two phases).
magnitudes should be used with respect to balanced
condition, with the restrain current previously Summarizing: there exist quite simple phasor
supposed, to get the desired Id (values slightly expressions: ΔI1A, ΔI1B and ΔI1C , relating single-
above the curve to test trip, and slightly below the phase current variations to resulting differential
curve to test non-trip)?. currents.
To answer this, we must return to equations (20) If, by contrast, the current variations are performed
and (21). in phases of non-reference winding (winding 2 in
equations of section I), the resulting differential
According to (20), it is possible obtaining a IdA ≠ 0 currents could be obtained by (21). As M2 depends
varying only phase R current in phase reference
on θ, this relation won't be as intuitive as those
winding. Of course, it is possible for the other two obtained for variations in phase currents in phase
phases also (B and C). reference winding (winding 1 in equations of
It is important to note in (20), that the differential section I).
current obtained when only one phase current
In other words; the expressions relating injection
magnitude in the phase reference winding is currents variations to resulting differential currents
modified, is not dependent on θ, since M1 is not are simpler and easier to interpret when those
dependent on θ.
variations are performed in phase reference
Conversely, when only one non-reference winding winding.
phase current is modified, according to (21), Id will
Based on the above observations, we can conclude
be dependent on θ, since M2 is dependent on θ. that the most convenient way to generate
The second observation about (20) is related to differential currents is: starting from balanced
alternatives for M1. As stated in section I, M1 is the load state, apply decrement variations over one
identity if phase reference winding does not require or more phase currents in the phase reference
I0 removal, and M1 is equal to M0(0º) when I0 winding, as Irest must remain unchanged
removal is required in the phase reference winding.
Then, always according to (20), when no I0
removal is required, the differential current in a
phase will match with the ΔI applied to that phase
(this is valid for A, B and C). In this case,
differential current on phases with no ΔI will be
0. Generalizing (20):
I dA 1 0 0 I1 A I1 A
I d I dB 0 1 0 I1B I1B ( 29 )
a) Settings for differential algorithm
I dC 0 0 1 I1C I1C
I dA 2 1 1 I 1 A
1
I d I dB 1 2 1 I 1B
3
I dC 1 1 2 I 1C
( 30 )
2I1 A I1B I1T
1
I1 A 2I 1B I 1T
3 b) Curve characteristic
I1 A I1B 2I 1C
Figure 7. Settings in T60 (GE) relay
IV. Proposed test procedure for T60 relay In this case, differential currents in other phases
will appear, but its magnitudes will be half of
IdF. Since IdF must be very close to the
In III.B, simple criteria for injected currents
characteristic curve, (e.g., 5 or 10% above or
determination was stated and justified. below it) with the Irest previously established, it
The goal for this section is to define a basic general is reasonable to expect that the differential
protocol for differential-characteristic-curve-survey currents in other phases will not cause any trip.
tests, applicable to both types of faults (single- Then, if I0 removal is required in phase
phase and multi-phase). reference winding, each test point for pure
single-phase faults can be tested reducing the
Those steps are as follows: faulted phase current by ΔI1F, where ΔI1F will
1. Define/Determine reference windings. As be obtained from (31). Currents must be
stated in section II: reduced, since Irest must remain unchanged.
• Magnitude reference: manually defined by In contrast, If I1F is augmented, according to
configuration, or automatically determined (25) IrestF will be modified, since IrestF is equal to
(that winding with the largest CT margin). the largest compensated current for phase F,
From now on, this winding will be referred as then, the testing point is no longer the expected
X. before.
• Phase reference: It is delta (D) winding For two-phase faults with I0 removal in phase
with the lowest index, or winding 1 in case of reference winding, the equation (30) is usefu. If
no delta or zig-zag windings are present in two phases have variations of equal magnitude
transformer. (ΔI) but 180º shifted, it will appear differential
2. Irest definition and balanced load currents currents in that phases, with magnitudes equal
calculation (Id=0). to ΔI.
For currents in the X winding: For example, for an A-B fault, we apply a
• If I0 removal is required, use (26) to reduction for IA by ΔI (i.e., ΔIA= ΔI<180º),
calculate injected currents. while the applied ΔIB will be ΔI<0º (180º
• Otherwise, use (14). shifted from ΔIA):
For other windings, injected currents should
be calculated according to (13) (i.e. must be I dA 2I 180º I 1B 0º
magnitude-compensated referred to currents in I 1 I 180º 2I 0º
dB 3 1B
X winding). I dC I 1 A 180º I 1B 0º
Current angles: ( 32 )
• For currents in phase reference winding 2I 180º I 1B 180º
1
(winding 1 in equations of section I), phase I 0º 2I 1B 0º
currents shall be shifted 120º, according the 3
I 1 A 0º I 1B 0º
expected phase sequence. No additional phase
shift is needed. I dA I 180º
• For currents in other windings, besides the I 1 I 0º ( 33 )
120º shift between phase currents, an additional dB 3
I dC 0
angle of -θN must be applied to all phase
currents. θN is the compensation angle for Optionally, the variations IA and IB currents
winding N. could be:
3. Use (29) or (30) to generate the desired Id. I dA I 240º
As noted in section III.B, when I0 removal is I 1 I 60º ( 34 )
required in phase reference winding, equation dB 3
(30) applies. Then, to simulate single-phase I dC 0
faults it still be enough to modify only the phase I dA 2I 240º I 1B 240º I 240º
current in the faulted phase, but according to I 1 I 60º 2I 60º I 60º ( 35 )
dB 3 1B
(31) (i.e. IdF is not equal to ΔI): I dC I1 A 60º I 1B 60º 0
Then, injected values proposed in the procedure of As an example, if load is capacitive, with Ψ=30º,
section IV, are consistent with resistive phase-to- then α=0º, and the result is the same as the one
ground faults inside the protection zone, on the described in the first example of two-phase faults in
phase reference winding (expressed by (29) when section IV.
no I0 removal is required, or by (30) otherwise). It
Finally, three-phase faults can be seen as a three-
is assumed short-circuit power at both buses very
phase symmetric load at the fault point. Due to its
high (ideally infinite), supposition that enables the
symmetry, it doesn't matter if fault is grounded or
assumption of constant voltage and external load
not.
respect of initial balanced state.
Modifying fault impedance, it is possible to obtain VI. Procedure application example
every ΔI (and Id) provided in test points.
Faults in another phase follows similar criteria. To clarify all definitions and concepts presented
before, we will include in this section, a complete
For two-phase faults, injection values might be example of procedure defined in IV.
consistent with ungrounded phase-to-phase faults,
as we show on figure 9. The information referred to the modeled
transformer was taken from real field tests.
Next table summarizes transformer data and
settings needed or used by 87T function algorithm,
as stated before.
The transformer is a 132/34.5/13.8kV,
30/30/30MW, Yy0d11.
Primary is 132kV winding, and 13.8kV is delta-
Figure 9. Phase-to-phase fault connected, being the reference. To clarify the
analysis, it was decided to use 13.8 kV side as reference.
winding 1.
2. CT's margins are calculated and presented
Table I below.
Mw1=0.69
# of Windings 3 ―
Mw2=0.69
Winding 1 13.8 kV
Mw3=0.88
Unom Winding 2 34.5 kV If automatic selection option for phase
Winding 3 132 kV reference were used, magnitude reference
Winding 1 30 MVA would be winding 1 (132kV).
Snom Winding 2 30 MVA 3. As criteria for selection, a good approach is
Winding 3 30 MVA to subdivide every zone in the curve in an
Winding 1 Delta / No ― arbitrary number of segments, which is
Connection / IZG Winding 2 Wye / Yes ―
equivalent to define a number of points
enough to verify that the actually
Winding 3 Wye / Yes ―
characteristic curve fits the desired one.
Winding 1 1500:1 A:A
CT Winding 2 600:1 A:A
Iprim:Isec The table II, summarizes the selected points
Winding 3 200:1 A:A
provided for this test, detailing Irest, Id, Iw1, Iw2 and
Winding 1 0 º Iw3. Magnitudes of Iw1, Iw2 and Iw3 are those
Phase shift Winding 2 -30 º corresponding to balanced state for windings 1, 2, 3
Winding 3 -210 º respectively; Id is the differential current
Threshold 0.2 p.u. established by characteristic curve, depending on
Breaking Pt. 1 1.4 p.u.
Irest. Z column specifies the zone of the
Breaking Pt. 2 6.7 p.u.
characteristic curve where each point lies (1:
87T minimum threshold zone; 2: slope 1 zone; 3: cubic-
Slope 1 25 %
spline transition zone; 4: slope 2 zone; 5:
Slope 2 80 %
instantaneous 87T zone). Phase reference
Instant. Id 9 p.u.
winding does not require I0 removal.
2). This is justified by CT's polarity: 132kV CT's 3 1 0.7 0.2 0.7 0.7 0.55
are pointing toward the transformer, while 34.5kV 4 2 0.9 0.225 0.9 0.9 0.71
and 13.8kV CT's are pointing toward the buses. In 5 2 1.1 0.275 1.1 1.1 0.86
that case, according to (18), it is needed to invert 6 2 1.3 0.325 1.3 1.3 1.02
the currents in winding 3 (shifting them by 180º). 7 3 1.5 0.378 1.5 1.5 1.18
Then, total phase shift for winding 3 is 8 3 2 0.598 2 2 1.57
9 3 4 2.433 4 4 3.14
-30º - 180º = -210º.
10 3 6 4.719 6 6 4.7
The injection equipment used is an Omicron® 11 4 6.7 5.36 6.7 6.7 5.25
CMC-256, with 6 current sources to facilitate test 12 4 7 5.6 7 7 5.49
of differential currents between 2 windings at a
13 4 7.5 6 7.5 7.5 5.88
time.
14 4 8 6.4 8 8 6.27
Once general settings and conditions were 15 4 9 7.2 9 9 7.06
established, we perform the steps proposed in IV: 16 4 10 8 10 10 7.84
17 4 11 8.8 11 11 8.63
1. Phase Reference: winding 1.
18 5 11.5 9 11.5 11.5 9.02
Magnitude Reference: winding 1 (13.8kV)
was manually selected as magnitude 19 5 12 9 12 12 9.41
Currents Iw1, Iw2 e Iw3 are expressed as secondary encountered problems, during 87T characteristic
per-unit values. Then, these values need an curve tests, when some injection devices which
additional correction, beyond primary magnitude incorporates control software packages, based in
compensation according to (13). Then, assuming standard connection groups, are used.
ki as the CT ratio for winding i, secondary base
Table III
current will be:
SB Irest Id Iw1- Iw1+
k i i Bwi I Bwi i Bwi ( 39 ) # Z
p.u. p.u. p.u. p.u.
3 k i U wi 1 1 0.3 0.2 0.3 0.3
Nominal apparent power for all windings is 30 2 1 0.5 0.2 0.5 0.5
MVA. Then, SB=30 MVA. Furthermore, under 3 1 0.7 0.2 0.7 0.7
IX. References
Figure 10. Typical test results (CMC Omicron test [1] Lubomir Sevov, Zhiying Zhang, Ilia Voloh, and Jorge
Cardenas: “Differential Protection for Power
equipment)
Transformers with Non-Standard Phase Shifts”
As result of all the above considerations, and General Electric Digital Energy – Multilin.
since the injected currents in “healthy” phases [2] GE Grid Solutions: “T60 Transformer Protection
will be lower than current injected in faulted System”. Manual edited on 2107.
[3] Omicron: “CMC Test equipment”
phase, then small differential currents could
produce (undesired) trip conditions on
“healthy” phases.
X. Vita
Two situations may arise:
César Galli received his Engineering degree from Universidad
• If test point is a Iw1- as stated in VI (i.e., points Nacional de La Pampa (Argentine) in 2006. Since 2007 he
slightly below the curve, implying non-trip), works in Autotrol's (Argentine) Energy Division, as a Protection
and Integration Engineer, combining field and laboratory work
relay might operate by faults in “healthy” in many Utilities and Industry projects.
phases.
Jorge Cardenas received his Engineering degree from the
• If test point is a Iw1+ as stated in VI, (i.e., points Universidad de Ingenieria (Peru) in 1977 and his MBA from the
slightly above the curve) for single-phase Universidad Politecnica de Madrid (Spain) in 1998. Jorge began
(implying trip presence only in one phase), his career with the Utility Electroperu (Peru) in 1979, as a
Protection & Control Engineer, and in 1987 he moved to ABB
relay may operate also in the other phases. (Spain) as a HV equipment Sales Engineer, and then promoted
This cannot be interpreted as relay's to a Control Design Engineer. In 1989 he joined GE, where he
has held several positions. Currently Jorge works as EMEA-
maloperation, because currents are not realistic Application Manager with GE Digital Energy-Multilin. He
ones. authored and co-authored more than 60 papers presented on
protective relay conferences around the world. He is a member
It must be noted that, according to (12), (17), (18), of the CIGRÉ WG B5.31 and the WG B5.43 working groups.
(30) and (21), if θ≠0, a current variation of only
one phase on any winding, might produce
differential currents in two phases (this was