Transfguide transformer protection application guide

TransformerProtection
ApplicationGuide

About the Author
George Rockefeller is President of Rockefeller Associates, Inc. He has a BS in EE from Lehigh University,
a MS from New Jersey Institute of Technology, and a MBA from Fairleigh Dickinson
University. Mr. Rockefeller is a Fellow of IEEE and Past Chairman of IEEE Power Systems Relaying
Committee. He holds nine U.S. Patents and is co-author of Applied Protective Relaying (1st Edition).
Mr. Rockefeller worked for Westinghouse Electric Corporation for twenty-one years in application and
system design of protective relaying systems. He worked for Consolidated Edison Company for ten years
as a System Engineer. He has also served as a private consultant since 1982.
About the Guide
This guide contains a summary of information for the protection of various types of electrical equipment.
Neither Basler Electric Company nor anyone acting on its behalf makes any warranty or representation,
express or implied, as to the accuracy or completeness of the information contained herein, nor assumes
any responsibility or liability for the use or consequences of use of any of this information.
Original issue date 05/96
Revised 05/99, John Boyle; small updates
Revised 08/03, Larry Lawhead; small updates
Revised 04/07, John Horak; extensive rewrite
Revised 06/07, John Horak; minor typographical and editorial corrections

1
Transformer Protection
Application Guide
This guide focuses primarily on application of
protective relays for the protection of power
transformers, with an emphasis on the most
prevalentprotectionschemesandtransformers.
Principles are emphasized. Setting procedures
are only discussed in a general nature in the
material to follow. Refer to specific instruction
manuals for your relay. The references provide a
source for additional theory and application
guidance.
The engineer must balance the expense of
applying a particular protection scheme against
the consequences of relying on other protection
or sacrificing the transformer. Allowing a pro-
tracted fault would increase the damage to the
transformer and the possibility of tank rupture
with a consequent oil fire and consequent
personnel safety risks. There is no rule that says
what specific protection scheme is appropriate
for a given transformer application. There is
some tendency to tie protection schemes to the
MVA and primary kV of a transformer. While
there is some validity to this approach, there are
many other issues to be considered. Issues to
be considered include:
•The severity of personnel safety concerns
and the possibility that a given protection
scheme can reduce these risks.
•The danger to nearby structures and pro-
cesses if a transformer fails catastrophically
and the possibility that a given protection
scheme can reduce the possibility of such a
failure.
•An overall view of the economic impact of a
transformer failure and what can be done to
reduce the risk, including:
· The direct economic impact of repairing
or replacing the transformer.
· The indirect economic impact due to
production loss.
· Repair time vs. complete replacement
time.
· The availability of backup power feed or
emergencyreplacementtransformers,and
the cost of each option.
· The possibility that a given protection
scheme can reduce the damage and
resultant repair time, or that it can change
a replacement into a repair.
Some specific applications that affect protection
are: A tap changer flashover can ordinarily be
repaired in the field, but if this fault is allowed to
evolve into a winding fault, the transformer will
need to be shipped to a repair facility; hence,
protection that can rapidly sense a tap change
fault is desirable. A high magnitude through fault
(external fault fed by the transformer) shakes and
heats a transformer winding, and the longer the
through fault lasts, the greater the risk of it
evolving into an internal transformer fault; hence,
fast clearing for close-in external faults is part of
the transformer protection scheme. Some
transformers are considered disposable and
readily replaced, reducing the need for ad-
vancedprotectionschemes.Transformerprotec-
tion commonly includes some coverage of

2
external bus and cable, and faults in these zones
may expose personnel to arc flash hazards. Slow
clearing protection schemes may be unaccept-
able from an arc flash exposure perspective.
Fires in an indoor transformer may have high risk
of catastrophic facility damage and even higher
personnel safety risks, increasing the need for
advanced high speed protection. The proximity
of flammable process chemicals increases a
need for protection schemes that reduce the risk
of a tank fire. The failure of a transformer used in
a large base load unit-connected generator may
causeextendedgeneration-replacementcosts;
even the loss of a small station service trans-
former can cause a notable disruption of genera-
tion and high economic consequences. Similar
economic impacts may also exist at industrial
sites. Some transformers are custom designs
that may have long lead times, increasing the
need for advanced protection schemes.
1. Failure Statistics
Table I lists failures for six categories of faults
(IEEE C37.90, “Guide for Protective Relay
Applications to Power Transformers, Ref. 1).
Winding and tap changers account for 70% of
failures. Loose connections are included as the
initiating event, as well as insulation failures. The
miscellaneous category includes CT failure,
external faults, overloads, and damage in
shipment. An undisclosed number of failures
starts as incipient insulation breakdown prob-
lems. These failures can be detected by sophis-
ticated online monitoring devices (e.g. gas-in-oil
analyzer) before a serious event occurs.
Table I - Failure Rates, Ref. 1.
1955-1965 1975-1982 1983-1988
Percent of Percent of Percent of
Number Total Number Total Number Total
Winding failures 134 51 615 55 144 37
Tap changer failures 49 19 231 21 85 22
Bushing failures 41 15 114 10 42 11
Terminal board failures 19 7 71 6 13 3
Core failures 7 3 24 2 4 1
Miscellaneous failures 12 5 72 6 101 26
TOTAL 262 100 1127 100 389 100
2. Protection Example and General Concepts
The reader interested in additional information,
advanced or unusual application advice, and
detailed settings guidance should refer to Ref. 1.
Thisdocumentincludesextensivereferencesand
bibliographies. Also, Ref. 2 and 3, textbooks on
protective relaying, contain chapters on trans-
former protection, and Ref. 4, another IEEE
standard, includes good overall protection
schemes where a transformer is the interface
point between a utility and an industrial customer.
There are three general categories of protective
relay technology that arise in the discussions to
follow:
•Electromechanical:usesmagneticflux
created from current and voltage to create
torques on movable disks and relays, which
is the source of the term “relay.” Usually
single device number functionality.
•Solid State: uses low voltage analog signals
created from sensed currents and voltages;
uses discrete electronics and basic logic
circuits; may contain a basic microprocessor
for logic and some math. Usually single or
dual device number functionality.
•Numeric:amultifunction,programmable
logic relay; digitizes sensed current and
voltage, then calculates an RMS or phasor
equivalent value; uses a high-end micropro-
cessor. Usually incorporates many device
numberfunctions.
All Basler Electric relays are solid state or
numeric.

3
Table II lists some common ANSI device num-
bers associated with transformer protection. A
numeric relay generally contains many imple-
mentations of these devices within its program-
ming, and each instance of a device is referred
to, herein, as an “element” in the relay. For
example, while the Basler BE1-CDS220 is
primarily a transformer differential relay (hence,
includes the 87 device in elements named 87P
and 87N), it also includes nine independent
implementations of the 51 overcurrent device,
called the 51P, 151P, 251P, 51Q, 151Q, 251Q,
51N, 151N, and 251N elements as well as many
other device functions.
Figure 1 shows extensive use of relays that
would be representative of a large industrial
load. This will be used for discussions in some of
the material that follows. There are two 115 kV
feeds to two 30 MVA transformers that are
resistance grounded on the 13 kV side, limiting
ground fault current to about 400A from each
transformer. In other applications, a reactor is
used, and in some applications, the ground fault
current is limited to less than 10A. In a typical
utility application, transformers are connected
directly to ground, but occasionally a small
reactor is placed in the transformer neutral that
limits ground fault current to approximately the
same level as three phase faults. In this example
system, the protection scheme described
applies to solidly grounded (as well as imped-
ance grounded) systems, except the effect of
ground impedance results in the addition of
protection functions not required on a solidly
groundedsystem.
The phase and ground differential (87P and 87N,
Section 4) and sudden pressure relay (63,
Section 6) provide the primary transformer fault
protection. The suite of overcurrent elements
(51, Section 8) is generally considered backup
transformer protection, or for protection of the
bus and backup protection for the feeder relays.
These elements are part of the transformer
protection in that they limit the accumulated
damage that occurs from a transformer feeding
high current into downstream faults. The 67N
relay offers an alternative to the 87N function.
Hot spot monitoring (49, Section 9) is indicated,
but is likely an alarm only scheme.
If there is a possibility of over voltage on the
units due to local generation or a transformer
being placed at the end of a long line (the
“Feranti” effect), voltage relays (24 and 59,
Section 4.4.4) may be included. Another pos-
sible backup protection scheme is low voltage
(27) or unbalanced voltage detection (47). If
there is local generation, to help detect
islanding conditions an over/under frequency
(81, O/U) relay may be installed, though an 81
may not be considered a transformer protection
element.
Directional overcurrent relays and directional
power (67/50, 67/51, and 32, respectively,
Section 8.4) respond to load current circulating
through the 13 kV buses when the 115kV
breaker A is open and the 13.8kV tie is closed.
The elements may also respond to faults in the
transformer near the secondary bushings. If the
transformers can be operated in parallel, the
elements also provide a means to sense tap
changers that have become out of step with one
another. If there is generation in the 13.8kV
system, sensitive 67 elements can sense a small
generator backfeeding a 115kV fault.
The primary and secondary relaying would
sometimes be configured to feed their own
lockout relays (86) to help ensure that protection
is available even with a failure of one 86 relay or
its dc feed.
The protection scheme in Fig.1 does not utilize
fuses. Fuses normally would be seen only on
lower MVA transformers than indicated. See
Section 3.
Transformers of the indicated MVA normally
would have their oil tested for dissolved gasses
(Section 7) as part of routine maintenance.
Largertransformersmayhavecontinuouson-line
monitoringequipment.
Tables III and IV (pages 29-30) provide Basler
Electric relay models and their device numbers.

4
Figure 1 - Protection Example

5
Device Description Comment
24 Volts/Hz For overexcitation detection. Similar to 59 but the pickup is proportionate to
frequency.Typically an inverse time characteristic.
27ll Undervoltage ll = line to line
27ln ln = line to neutral (or line to ground. Note neutral may be isolated from ground in
some systems.)
32 Power Element Used to sense power backfeed through transformer
47 Negative Usually defined by: V2 = (Van+a2
Vbn+aVcn)/3. (a=1∠120).
sequence It is also possible to define V2 in terms of Vll. Some relays define the 47 in terms of
overvoltage a manufacturer-specific voltage unbalance measurement rather than in terms of V2.
49 Thermal Typically top oil temperature RTD.
51P Phase Time A 51 by itself usually refers to phase time overcurrent, but adding the P gives
Overcurrent additional clarity.
51G Ground Time Herein: Ground refers to current on a transformer ground/neutral bushing or
Overcurrent, measured current from a window CT that wraps all three phases (and possibly also wraps
by dedicated window a neutral bus if one exists).
CT, or CT on
transformer neutral
51N Neutral Time Neutral refers to the phase current summation (In = Ia+Ib+Ic), which is the
Overcurrent equation used for 3Io in numeric relays and which is also the summation of the
= 3Io 3 phase CTs ("residual").
= Phase CT
Residual
51Q Negative Defined by: I2
=(Ia
+a2
Ib
+aIc
)/3
Sequence Time
Overcurrent
50P, Instantaneous In some relays, a 50 has the option of being time delayed; hence, it becomes
50G, Overcurrent. a definite time element and may be renamed 50TP, 50TG, 50TN, or 50TQ.
50N, P, G, N, Q have
50Q same meaning as
for the 51.
59ll Overvoltage ll, ln = line-line or line to neutral/ground. N refers to V0 or 3V0 sensing, depending
59ln on the manufacturer. V0
=(Van
+Vbn
+Vcn
)/3.
59N
63 Sudden Pressure There may be separate devices for the tap changer and main tank.
67/50x Directional control, x refers to P, G, N, or Q. The 67 by itself is used inconsistently in the industry.
67/51x directional Herein, for clarity, a 67 is a sensitive forward/reverse polarizing bit that
instantaneous, controls the 50 and 51 element, and the dual term 67/50x or 67/51 is used.
directional time
overcurrent
86 Lockout Auxiliary Most transformer trips are directed to a lockout relay.
87P Phase Differential Comprised of several functional elements. See text for description. Many variations
by relay and manufacturer.
87U Unrestrained Monitors phase differential. Trips when magnitude is much greater than maximum
Differential inrush levels.
87N Ground Differential Sometimes referred to as a "Restricted Earth Fault" sensing element. It is more
commonly applied on systems with impedance in the transformer neutral for the
purpose of limiting ground fault current.
Table II - ANSI Device Numbers (C37.2)
3. Fuses
Fuses are economical, require little mainte-
nance, and do not need an external power
source to clear a fault, which is of great cost and
maintenance benefit. As discussed above, MVA
of a transformer is an imperfect guide to the
appropriate level of transformer protection, but it
may be noted that fuses are probably the
predominant choice for transformers below 10
MVA. Under 3MVA, breakers on the high side
are seen only in special applications (e.g., some
small generation sites may use a high side
breaker).
The use of fuses creates some notable protec-
tion compromises. Fuses are not as precise in
operatingcharacteristics.Characteristicschange
slightly with temperature, pre-fault loading, and
reclose timing. Fuses are subject to gradual
damage from heavy through faults, leading to an

6
eventual fast trip for a low magnitude fault.
Usually only one or two fuses blow, which
introduces single-phasing conditions to down-
stream loads. Single phasing causes very high
negative sequence voltage and current and low
Vln or Vll voltages. The resultant voltage may be
worse than no voltage due to the overheating
that it can cause to certain types of equipment,
such as three phase motors.
Fuses are insensitive and relatively slow except
at very high current levels. Fuses will not sense
low level faults, such as near the neutral of the
transformer, and hence trip only after the fault
has evolved into a high current event. To allow
short overloads, a transformer fuse is typically
selected to carry 150-300% of the transformer
rated current (see NEC article 450, Ref. 5). Most
fuses can carry over 125% of rated current for
very long times, and just begin to reliably trip for
faults in the range of 150-200% of the fuse rating,
and at this level generally take tens of seconds
to trip. The effect is that a fuse might carry
current in the range of 3 to 5 times transformer
rated current for an extended period. At moder-
ate currents, fuses are still relatively slow. For
instance, a fuse must be able to carry inrush
current without damage. A classical measure of
inrush current is 12x full load current for 0.1
second. Due to slow clearing for faults at low
magnitude, when a fuse is used, the transformer
is at higher risk for being irreparable after an
internal fault and at higher risk of failing in some
catastrophic manner, such as a fire.
In some protection schemes, the transformer
high side overcurrent protection scheme is
considered backup protection for faults down-
stream of the secondary protection elements.
The low sensitivity of fuses means they are poor
at backing up secondary overcurrent protection
devices, especially for faults remote from the
transformer secondary and especially for ground
faults on delta/wye transformer banks. See
Section 10.3 for a discussion on the current
ratios seen across delta wye transformer banks.
A fuse has some protection benefits. If fault
current is extremely high, a fuse can be faster
than a breaker and can clear faults within 0.5-2
cycles after inception. The fastest type, called a
“current limiting fuse” (CLF), available at medium
and low voltages, can clear a fault in half a cycle
and can reduce the first half cycle peak current
to a level notably below the available fault
current. The CLF fuse can be damaged by inrush
current if not properly selected. If a fuse is
selected that has fast clearing at peak currents,
it can clear a fault faster than any circuit breaker;
hence, at high fault currents, a fuse downstream
of a circuit breaker tends to coordinate with
the upstream breaker better than two breakers
in series. This application is seen most com-
monly in an industrial low voltage application
where two breakers in series sometimes are
found to both be operating in an instantaneous
trip zone. In a transmission application, when a
high voltage fused transformer is placed as a tap
on a transmission line, coordination is not easily
achieved. The transmission line impedance
relay, operating in its Zone 1 reach, frequently
will be committed to a trip within 1-1.5 cycles
after a fault begins; hence, the transmission line
may trip ahead of or simultaneously with a fuse
on a transformer.
It is recommended that, on fused transformers,
protection should employ a low-side circuit
breaker with phase and ground overcurrent
relays for backup protection of secondary faults.
Ideally, the relay also should have negative
sequenceovercurrent(46),negativesequence
overvoltage (47), and line to ground
undervoltage(27ln),forsensingunbalanced
system operation in the event of a single or
double fuse operation on the transformer
primary. Be aware that, in a radial power flow
application, relays on the transformer secondary
relay will not respond to a transformer fault,
except possibly a 27 or 47 relay can sense
resultant voltage degradation.
4. Differential Relaying (87)
Differential relays sense the unbalance in the flow
of currents in various apparatus or buses. In the
absence of a fault in the protected zone, this
unbalance tends to be small and the flows into
the zone are closely matched to the flows
leaving. Accordingly, such relays can be more
sensitive than phase overcurrent relays and need
not be delayed to coordinate with other relays
during external faults, except for some issues
associated with transient CT saturation, to be
discussedbelow.

7
Figure 2 - Basic Differential Concept
The simplest implementation of differential
protection, as seen in Fig. 2, merely parallels the
CTs on all the connections to the zone, and the
relay monitors the current summation. This
concept is applied on basic bus protection
schemes. The 87 device for this application
feasibly can be simply a 51 device, though a 51
normally would be configured to be neither
extremely sensitive nor fast due to issues with CT
transient performance under the presence of DC
offset in the primary current.
The basic differential concept above will not
work with transformers. When the protected
zone in Fig. 2 is a transformer, there are several
additional components, seen in Fig. 3, that are
required, or at least commonly implemented, for
properperformance:
•CurrentMatchingScheme
•ThroughCurrentRestraintandMinimum
Operate Logic
•Delta/Wye Compensation Logic (may be
implemented in CT connections rather than
in the relay)
•MagnetizingInrushBlockingScheme,
commonlyusingharmonics
•UnrestrainedDifferentialScheme(notin
some basic or early electromechanical
relays, but found in all Basler relays)
Figure3-TransformerDifferentialConcept
4.1 Current Matching Scheme
The relay’s current matching scheme allows
different currents on each input to the relay to be
seen as effectively the same current. In electro-
mechanical relays, the scheme uses tapped
transformers (hence, the source of the term
“tap”), where each tap adjusts the number of
turns used on the input transformer, so that, for
instance, 4.6A on input 1 (at a tap of 4.6) gives
just as much magnetic flux as 3.5A on input 2 (at
a tap of 3.5). In solid state relays, the tap,
typically (as in the BE1-87T) is composed of
switches that change the resistance on the
secondary of the sensing CT circuit. For in-
stance, a two input solid state relay set at taps
4.6 and 3.5 might be designed so that 4.6A and
3.5A on each respective input will both cause a
100MV internal voltage signal. Hence, 4.6A =
3.5A as far as the remainder of the relay circuitry
is concerned. On numeric relays (i.e., micropro-
cessor based relays that convert incoming
currents to digital signals), the current matching
scheme uses mathematics and varying multipli-
cation factors.
A typical concept for setting the taps for a two
winding transformer is to analyze the current
seen at the relay for the peak power rating of the
transformer. For instance, in Fig. 1, assume the
CT ratios as shown (discussed further in Section
10.2) and, the Full Load Amps(FLA) in the lines
at 30MVA is 151A at 115kV, and 1255A at
13.8kV, and the CT secondary current is 1.883A
and 3.138A, respectively. Depending on the
relay design, the delta compensation may
require that the current on the Wye side be
multiplied by sqrt(3) (e.g. the CDS-2x0 relays do
not require the sqrt(3) factor, but the BE1-87T

8
does). For this example, let us assume a BE1-
CDS220 is in use, so we do not need the factor.
The ratio of the taps is the critical factor, not the
specific taps. The minimum tap of the BE1-
CDS220 is 2, so we use taps of 2.00 and 3.33.
The ratio of currents (3.138/1.883) is 1.667, and
the ratio of taps (3.33/2.00) is 1.665, so the error
that the relay sees due to tap ratio not matching
the current ratio will be quite small. In electrome-
chanical relays, the tap selection is much more
limited, and resultant error seen by the relay
tends to be substantially more significant.
Currentmatchinginthreewindingtransformer
applications must, in effect, be analyzed two
windings at a time. One assumes identical power
in windings 1 and 2 (and 0 in winding 3), which
sets taps 1 and 2, and then one assumes
identical power in windings 1 and 3 (and 0 in
winding 2), which sets tap 3. This will force
winding 2 and 3 taps to work correctly for the
case of power flowing from winding 2 to winding
3. In this approach, the constant power that is
assumed can appear to be more than winding 2
or 3 can handle, but since this power level does
not flow in actual application, the high current
flow seen in the calculations does not affect the
validity of the setting.
Figure 4 shows the conceptual implementation
of current matching in the BE1-87T for a single
phase transformer. Relays of other designs will
use parallel concepts appropriate for their
design. When installed per the relay instruction
manual, the current will flow in opposite polarity
on input 1 and input 2. The taps will be set so
that the magnitude of the voltage presented to
the op-amps is the same RMS value. The
voltage at point C, the summation point, will be
at the midpoint between A and B, and in normal
operation, since A and B are opposite in polarity
but equal in magnitude, the voltage at C will be
0 throughout the current sine wave, indicating no
error current. If current at input 1 increases,
voltage at A increases, but not B, and now point
C sees a voltage above 0. Depending on the
relay settings, the relay may trip.
Current matching settings normally are calcu-
lated under the assumption that the load tap
changer is at the nominal / neutral position. The
relay percentage restraint and minimum operate
settings then are configured to accommodate
the unbalance current that the relay will see
when the load tap changer is at the full boost or
buck position, and commonly with some
additional accommodation for the no-load tap
being changed from nominal.
The BE1-87T has a current matching ("tap")
range of 2 to 8.9A. The BE1-CDS relays have
matching ranges from 2 to 20A. The wider range
of the numeric relay can be helpful in installa-
tions where small transformers are placed on
buses with high short circuit duty (the high short
circuit duty forces one to use higher CT ratios
than would be chosen just on load current level
criteria).
4.2 Percentage Restraint and Minimum
Operate
Typically, there is some small difference in the
summation of the sensed currents so that, even
after appropriate tap settings, currents sensed
by the relay do not sum to an effective 0A. The
error (or difference) current becomes the operate
current which, if it rises too high, will cause relay
operation. The operate current arises due to
dissimilar CT performance during a through fault
or large load inrush (especially during transient
CT saturation due to DC offset); differences in
the transformer no-load tap that is in use vs. the
no-load tap assumed when the relay settings
were calculated; the operation of the transformer
Figure 4 - Current Matching in a Solid State Relay

auto tap changer (if there is one); transformer
steady state excitation requirements; and inrush
excitation current that has decreased below the
harmonic excitation blocking levels but has not
actually decayed to a negligible value. Several
of these types of errors create an operate current
that rises proportionately with through current.
Because the error rises with load current, a fixed
level for the operate pickup generally is not
acceptable.
To compensate for operate current rising during
high load currents or through faults, the 87 logic
includes a restraint function. As through current
increases, the restraint function causes an
increase in the level of operate current required
for a trip. Each current input is monitored and
becomes part of the overall restraint of the relay.
The concept is referred to in many documents
as a percentage restraint characteristic because
the logic can be described by: "The operate
current must be ##% of the restraint current
before a differential trip is declared."
The concept as implemented in the BE1-87T is
seen in Fig. 5, where operating (or "differential")
current is plotted against the maximum (i.e.,
largest) restraint current. The scaling is in
"multiples of tap." The slope of the characteristic
can be set from 15 to 60%. The relay becomes
desensitized at the higher restraint currents in
order to remain secure in the presence of the
various sources of error previously mentioned.
Figure 5 - BE1-87T Restraint vs. Operate
(BE1-CDS-2x0similar).
9
Note in Fig. 5 the horizontal axis is in "maximum
restraint current." This allows the best-performing
CT to be used for the restraint decision. The
BE1-CDS2x0 has the option to use average
restraint rather than maximum restraint, though
maximum restraint is the preferred setting.
Relays from other manufacturers use other
restraint algorithms. For instance, some relays
on the market desensitize the relay at a greater
slope during very high through current, using a
"dual slope" approach. The BE1-CDS2x0 relays
use an alternative approach via an algorithm that
desensitizes the relay during high currents, but
only when a CT saturation detection logic
scheme is declared true.
Figure 5 also shows the minimum operate
feature. Independent of the through current
restraint, there must be a certain magnitude of
operate current before a trip is issued. This
feature helps prevent false tripping due to
severalissues,suchaswheninrushexcitation
current has decreased below the harmonic
excitation blocking levels. The minimum operate
current is fixed in the BE1-87T and adjustable in
theBE1-CDS2x0.
The "total mismatch" line in Fig. 6 represents the
sum of the imperfect relay-tap match (i.e., the
discrete increments available for the relay taps
may not be the ideal setting) plus the effect of
the power transformer no-load and load (i.e.,
auto) taps not being at the setting used when
calculating settings. The mismatch line is offset
by the transformer exciting current, which
produces its own unbalance. Fig. 6 shows the
BE1-87T slope characteristics at the two
extemes of slope setting (15 and 60%), as well
as the related safety margins at the critical point
where the slope characteristic meets the mini-
mum operate current. The two lines intersect at
approximately full load current on the restraint
axis, but the meeting point reduces with in-
creased slope setting and increases with mini-
mum operate current. A side issue to be aware
of is that, if high prefault load is flowing through
the transformer, some of the load current
continues to flow during the fault and acts to
increase the restraint level.

10
Figure 6 - Percentage Restraint Margin
4.3Delta/WyeCompensation
A delta wye bank introduces a phase shift in
balanced voltages and currents of typically
±30°, depending on system rotation and
transformerconnections.Inelectromechanical
relays, the common approach to the delta wye
winding compensation is to connect the CTs in
wye-delta, as seen in Fig. 7. Note the CT on the
delta transformer winding is connected in wye,
and the CT on the wye transformer winding is
connected in delta. The same concept could be
used with solid state and numeric relays, but
normally such relays have internal delta wye
compensation logic. The preferred approach is
to connect CTs in wye and allow the relay to
perform the delta wye compensation. The BE1-
87T relay performs the delta wye compensation
electronically via analog circuitry, and the BE1-
CDS2x0 performs the compensation via math-
ematical processes, thereby removing the need
to connect any CTs in delta.
Figure 7 - Delta Wye Compensation with CT
Connections
There are several benefits to connecting the CTs
in wye. With wye CTs, metering and event
reports give actual line currents, rather than the
summation of two phases as seen in delta CTs.
Power metering is more accurate when CTs are
in wye. Wye connections allow monitoring
ground current in the lines and the use of ground
relays. A wye connection also reduces lead
burden for a phase fault. The worst case is for a
3-phase fault with delta CTs. Per Fig. 8, the lead
burden voltage as seen by delta CTs is magni-
fied by a factor of three times relative to the
burden that would be seen with wye CTs (note
current in the leads is higher by 1.732 factor,
and that the CT is across two phase leads,
hence seeing higher voltage compared to being
across phase to neutral). See Appendix A of the
BE1-87T instruction manual for a more in-depth
analysis of this issue.

11
Figure 8 - Increased CT Burden from Delta CT
Connection
It is essential that the delta compensation in
either the CT connection or internal to the relay's
algorithms be of a type that matches the trans-
former connection. For example, if a "DAB" delta
is used, the compensation needs to be DAB, not
DAC. Incorrect selection of DAB vs. DAC com-
pensation is a common source of trouble during
the commissioning stage and, on lightly loaded
transformers, can be a source of misoperations
seen long after initial startup. The difference
between a DAB and DAC delta is seen in Fig. 9.
The figure is drawn to show delta connections
but also to represent the phasor diagrams of
system voltages with balanced ABC phase
rotation (i.e., the secondary wye windings are
drawn in a direction that represents ABC positive
sequence phase rotation phasors, and the
primary delta windings are then drawn in phase
with their corresponding wye windings). Note in
the two cases, the polarity of the A phase
element is either a) connected to the non-
polarity of phase B (referred to as a DAB con-
nection) or b) connected to the non-polarity of
phase C (referred to as a DAC connection). For
balanced voltages as seen in the figure, the
resultant Van phasor on the wye side either a)
lags the delta Van phasor (DAB connection), or
b) leads the delta Van phasor (DAC connection).
Figure 9 - DAB vs. DAC Delta
These two methods to connect a delta have very
large effects on which phases on the delta side
see current during a wye side fault. For example,
examine the wye side faults seen in Fig. 10. Note
that, for the DAB delta, the wye side A phase to
ground fault creates current in the lines A and B
on the delta side, but for a DAC delta, the wye
side A phase to ground fault creates currents in
lines A and C on the delta side.
Figure 10 - DAB vs. DAC during a Wye Side Fault

The difference between DAB and DAC deltas is,
to some extent, the result of how one assigns
phase names to the lines connected to the
transformer. In Fig. 10, the difference between
the DAB and DAC transformers is simply created
by swapping the B and C phases on the wye
side, which also changed phase naming on the
delta side. Hence, though no internal wiring was
changed on the transformer, how one named the
phase leads and windings was changed, and
thisrenamingcauseslong-rangingconse-
quences on transformer phase shift and exten-
sive confusion for the relay and substation
engineer. Delta/wye compensation as seen in
Fig. 9 is the most common type of transformer
phase shift found in the industry, but there are
many other three phase transformer winding
configurations. Section 4.7 discusses the matter
further.
An example using CTs for compensation is seen
in Fig. 11. Note that the delta is connected DAC
and that the CTs on the wye side are also
connected DAC.
Figure 11 - DAC Compensation Using CT
Connections
Figure 11 adds a valuable insight into how a
solid state or numeric differential relay compen-
sates for transformer delta/wye connections. It is
a common misunderstanding that delta/wye
compensation involves a phase shifting algo-
rithm in the relay that simply reverses the 30°
phase shift seen in balanced currents. Another
misconception occasionally seen is that the
relay compensation involves removing zero
sequence current from the wye side currents.
Neither is true.
The numeric and solid state relay is actually
doing, in math or electronics, the same thing
being done with CT connections in Fig. 11. Note
the differential element does not see the actual
current in any one winding of the transformer,
but all three differential comparators monitor
current in two phases of the transformer at a
time. Let us name Ia1, Ib1, and Ic1 as the
currents in the transformer delta (not the leads to
the transformer), and Ia2, Ib2, and Ic2 are the
currents in the wye windings. Note the relay's A,
B, and C comparators monitor these sets of
currents:
• A comparator: Ia1-Ic1 to Ia2-Ic2
• B comparator: Ib1-Ia1 to Ib2-Ia2
• C comparator: Ic1-Ib1 to Ic2-Ib2
The numeric and solid state relay performs the
same compensation by adding and subtracting
currents in a process that mimics the effect of
the delta CT connections seen in Fig. 11. Note
there is no deliberate phase shift or zero se-
quence removal algorithm, though, in effect, this
happens. For a more complete description of
compensation in numeric relays, see Ref. 6.
4.4 Inrush Detection and Trip Blocking
Transformerinrushreferstothetransientexciting
current resulting from a sudden change in the
exciting voltage. This occurs at the instant of
energization, after the clearing of an external
fault (recovery inrush), or during the inrush
period of another nearby transformer (sympa-
thetic inrush) (Ref. 7).
Inrush current appears as operate current to a
differential relay so the relay must either a) have
sufficient time delay and insensitivity to the
distorted wave so as to not see the event (of
course, this is the undesirable answer) or b) take
advantage of the inrush's distinctive waveform to
sense the event and block tripping. The most
common means to sense inrush is via the use of
harmonic content in the operate quantity. The
12

second harmonic predominates in inrush cur-
rents (Ref. 7) and is used in most transformer
differential relays, either alone or in combination
with other nonfundamental components, though
there are other waveshape monitoring schemes
in a couple relays in the market. The harmonic
sensing relays most commonly block operation
if the harmonic(s) in the operate leg exceed(s) a
given percentage of the fundamental compo-
nent, though some relays on the market use the
harmonic to increase the restraint current. Some
relays (e.g. Basler products) use a scheme that
uses total harmonic current in all three phases in
the analysis of every phase, and some use a
cross blocking scheme, where if one phase is
blocked, all phases are blocked. In the BE1-87T,
the percentage settings are fixed, but in the BE1-
CDS2x0, the harmonic restraint level is a user
setting.
4.4.1 Energizing Inrush
Energization inrush is caused by remanence
(residual flux) in the core and the point in the
voltage waveform when a transformer breaker
closes. If the instantaneous voltage at
energization calls for flux of the same polarity as
the remanence, the core is driven into saturation,
creating peak exciting currents that can exceed
ten times rated exciting current. As a compari-
son, normal steady-state excitation current is
about 0.01 to 0.03 times rated.
In Fig. 12 the steady-state flux at the instant of
energization matches the residual flux, so no
transientcurrentflows.
Figure 12 - Energization without Inrush
In contrast, in Fig. 13, the steady-state flux at
energization is at its negative peak. Combined
with a positive remanence, this condition
produces the maximum level of transient current.
The inrush current is actually much larger in
relation to steady-state current than indicated by
Fig. 13, in order to keep the figure to a reason-
able size.
Figure 13 - Energization with Inrush
Figure 14 shows a typical inrush waveform. Note
the dead/flat spot where almost no current is
flowing as the core exits and then re-enters the
saturated region. The alternating flat to high
peak current contains the second harmonic that
a relay uses to recognize the existence of an
inrushcondition.
Figure 14 - Energization Waveform
Note in the first two to three cycles of Fig. 14's
waveform, there is an effective DC component of
the waveform. This DC is causing a flux buildup
in the CT steel and a partial saturation of the
core. After about 3 cycles, the flat/dead spot
rises above the 0 current axis, and the compo-
nent of current above the 0 current axis is
roughly equal to current below the axis, indicat-
ing the CT is no longer producing any DC offset
(even though DC may exist on the primary), but
it is still reproducing at least some of the AC
13

components, though in a possibly distorted
fashion. In extreme cases, the CT can saturate
during the first cycle, so the flat spot in the
current waveform never remains at the 0 current
level for any duration. The decay rate of succes-
sive primary-current peaks depends upon the
amount of resistance in the source and the
nonlinear inductance of the transformer.
In three phase transformer differential relays, the
differential relay has the ability to monitor
harmonic levels in all three phase differential
comparators at the same time; hence, it makes
a decision that an inrush condition exists on a
three phase basis, rather than on a per phase
basis. This feature is found in the Basler BE1-87T
andBE1-CDS2x0.
4.4.2RecoveryInrush
A recovery inrush occurs at the clearing of an
external fault as a result of the sudden increase
in voltage from the depressed and unbalanced
level that exists during the fault. This voltage
transient causes a flux transient, with accompa-
nying abnormally high exciting current. The
current level is less than that seen during trans-
formerenergization.
4.4.3SympatheticInrush
The current Ip in Fig. 15 shows sympathetic
inrush current in transformer T1
, resulting from the
energization of an adjacent transformer T2
. The
decaying DC component of current Ie flowing in
T2
develops a drop in the source impedance Rs
and Xs
, producing pulses of inrush current Ip on
the alternate half cycles. Note the delayed
buildup of Ip
. The severity of the sympathetic
inrush is a function of the level of DC voltage
drop across the source impedance. A common
set of differential relays should not be used to
protect both T1
and T2
transformers in Fig. 15 if
they can be switched separately. The sum of the
two transformer currents, Is
, may not contain
sufficient harmonics to restrain the relays once
transformerT1saturatesseverely.
Figure15-SympatheticInrush
4.4.4Overexcitation
Overexcitation results from excessive voltage or
below-normal frequency or a combination of the
two, such that the volts/Hz exceed rated. Fig. 16
shows three situations where overexcitation can
occur: short line and a long line unloaded
transformer condition and long line with load
connected. In addition, an interconnected
system can experience a dynamic overvoltage
following a protracted fault as a result of genera-
tor fields going to their ceiling or following load
shedding. All of these scenarios involve essen-
tially balanced conditions, so phase to phase
and phase to ground voltages increase. Sub-
stantial phase to ground overvoltages can also
occur on sound phases during a ground fault on
impedance grounded systems. In these cases,
delta windings or wye-ungrounded windings will
not be overexcited, since the line to line volt-
ages will not increase.
Figure 16 - Systems with Transformer Overvoltage
Risk
The increase in transformer exciting current with
increased excitation voltage is illustrated in the
14

dashed curve in Fig. 17. The transformer will
tend to become overheated due to increased
excitation current, hysteresis losses, and eddy
currents. The increased exciting current pro-
duces operating current in the differential relay,
but an operation of this relay is not desirable,
since immediate response is not necessary. The
power system should be allowed time to correct
itself. Also, a differential operation indicates a
transformerfailure,requiringunnecessary
investigation and delayed restoration of the
transformer.Accordingly,wheresustained
overexcitation is a concern, a separate volts/Hz
relay should be applied (24). The volts/Hz
function is available in the BE1-CDS240, BE1-
1051,BE1-951andBE1-IPSrelays.
Figure 17 - Transformer Excitation Current vs.
ExcitationVoltage
The solid curves of Fig. 17 illustrate the variation
in fundamental frequency excitation current (I1
)
as a percentage of total current (Im
), and har-
monic content (I3
to I7
) as a percentage of the
fundamental current (I1
), as voltage rises, for
balanced system voltages. If the transformer is
connected in delta, the third harmonic would
exist but remain within the delta windings as a
"circulating current" that would not be seen by
CTs on the lines to the transformer. When a delta
winding is energized, all triplen frequency
currents (triplen = odd multiples of 3, such as
third, ninth) are blocked from exiting the delta
because they are in phase with each other,
similar to the manner in which a delta cannot
supplyfundamentalfrequencyzerosequence
current. Accordingly, the lowest odd harmonic
available to a relay monitoring the lines outside
the delta is the fifth. The BE1-87T restrains if the
fifth harmonic exceeds 35% of the fundamental,
and the fifth harmonic restraint is a user setting in
the BE1-CDS2x0. In Fig. 17, a setting of 35% will
restrain the relay from operation over the voltage
range of about 104 to 138% of nominal voltage.
With normal system operation, the power
system could be operated continuously at 105%
and dynamically as high as about 115% during a
severe disturbance, so the 5th harmonic block-
ing will prevent tripping for these high system
voltages.
If there is load current in the transformer, e.g.
Fig. 16(c), any mismatch current will increase the
fundamental component of operate current and,
hence, will reduce the percent of fifth harmonic
level in the operating current. The relay may not
be restrained by the fifth harmonic. However, the
transformerloadingincreasesthethrough
current restraint level and tends to bring down
the overexcitation to a level where the operating
current is below pickup.
Should the transformer become faulted during
the high excitation condition, the relay will
operate if the operate current is sufficient to
reduce the fifth harmonic component below the
relay's restraint level. Such a reduction occurs
both because of the reduced excitation level
caused by the fault current and the increased
fundamental-frequencyoperatecurrent.
Voltages in excess of 138% can follow full-load
rejection of hydro units. However, generator
speed will be correspondingly high, so the volts/
Hz value will not significantly exceed normal.
4.5UnrestrainedDifferentialElement
The unrestrained element, commonly referred to
as an 87U, responds to the operating or differen-
tial current but with no restraint functionality. It
acts as a high speed trip during severe internal
faults. Its only means of differentiating an internal
fault from inrush is the magnitude of the current
involved; therefore, it must be set above the
largest expected inrush current. It must also not
operate for dissimilar saturation of CTs caused
by DC offset during high current external faults.
For these reasons, the unrestrained element
setting is set fairly insensitive. For example, the
15

87U typically is set to trip for differential current
that is on the order of 30-100 times the minimum
operate setting of the restrained differential
element.
In some relays, the 87U element responds
strictly to fundamental frequency currents in the
operate leg, yet other relays may respond to the
full spectrum of frequencies in the operate leg,
including the DC component. Since the 87U
needs to be set above inrush, and the inrush is
rich in harmonics and DC, one needs to know
how the relay is designed in order to make an
appropriate setting for the 87U element. For
instance, the current fed to the 87U element in
the BE1-87T relay is not filtered to remove DC or
harmonic content and so it responds to all
currents that come into the relay. However, the
87U in the BE1-CDS2x0 responds only to the
fundamental frequency component in the
operate leg, which means the 87U in the BE1-
CDS2x0 typically is set lower compared to the
87U in the BE1-87T. The appropriate settings for
the 87U element are discussed in the respective
instructionmanuals.
4.6ConnectionExamples
Fig. 18 provides application examples for two-,
three-, and four-restraint cases. The relay derives
restraint signals separately from each set of CT
inputs. In Fig. 18(a) the relay protects a delta-
wye transformer, with the CTs connected in delta
on the wye-winding side. These CTs could be
connected in wye when using a BE1-CDS2x0 or
a 3 phase style BE1-87T when using the internal
delta compensation functions of the relay.
A three-input relay protects the autotransformer
in Fig. 18(b). All CTs should be connected in
delta, or the relay internal compensation should
be set up in delta, since the autotransformer
contains an internal unmonitored delta winding
that couples the phases. Due to the delta
winding, for a phase to ground fault, the delta
can cause current to flow on an unfaulted phase.
For example, assume a 230/115kV transformer
with an A phase to ground 115kV external fault.
Due to the delta winding, current may flow on B
and C at the 115kV level and possibly not at all
at the 230kV level. This will look like a B and C
phase internal fault to the relay if no delta
compensation is applied. Delta compensation
prevents the misoperation.
In autotransformer applications, the tertiary is
primarily used to create a zero sequence source,
but it is used in some applications for station
service power. The tertiary commonly is not
included in the transformer differential CT
summation, because the station service load is
too small to be sensed by the transformer
differential relay and with the intent that fuses will
only operate for a station service transformer
fault. Since the tertiary is ungrounded, a ground
fault on the tertiary leads or load is sensed by
voltage based relays that monitor line to ground
voltages or V0
voltage. Sometimes no tertiary
ground fault detection scheme is applied under
the anticipation that the risk is low, and that the
VTs and relay needed to sense a ground fault
would cause more problems and faults than will
ever occur in the transformer. When a second
ground fault occurs, it is seen as a phase to
phase fault. The resulting phase to phase fault
on the tertiary leads or loads is sensed as an
internal transformer fault and, hence, is cleared
by an 87 or 51 element trip.
Note the high voltage winding in the autotrans-
former in Fig. 18(b) is connected to two break-
ers, and each breaker is brought to its own
restraint, R1 and R2. This is a desirable practice
when large through-faults from one side of the
substation to the next, flowing through the two
breakers, could possibly cause unequal CT
performance.TheunequalCTperformance
would appear as 100% operate current to the
relay. If each CT were on its own restraint, the
relay would see high restraint and not trip.
However, only the summation of the CTs should
be used for restraint when a low MVA transformer
is placed on a bus with relatively heavy through
current in the high voltage breakers seen in Fig.
18(b). For example, if the full load current for the
transformer in Fig. 18(b) was only 20% of the
normal load flowing in each of the high voltage
breakers, going from one side of the figure to
the other, a relay that restrains based on the
individual breaker currents as seen by R1 and
R2 could see 5pu restraint just from load flow
(assuming the load flow was not affected by the
16

fault) and tend toward non-operation except for
very high magnitude transformer faults. To
address this issue, the BE1-CDS240 has the
option to restrain on either the individual cur-
rents seen in R1 and R2 or on the summation of
current in R1 and R2.
The relay Fig. 18(c) protects the combination of
a bus and transformer. In this application, any
radial load feeders may have their CTs con-
nected in parallel and connected to a common
restraint input. Paralleling of CTs on non-source
circuits can be safe, within the thermal limitations
of the relay. In this case there is no loss of
restraint for external faults, since these circuits
contribute no fault current.
Source circuit CTs can also be paralleled, but it
must be done judiciously. Ideally, any breaker
that could be a substantial source to through
fault current should be on a dedicated restraint
winding. Hence, in Fig. 18(c), the bus tie, the
feeder to the generator bus, and the lead to the
high side of the transformer all have their own
restraintwindings.
Figure 18 - Connection Examples
Fig. 19 shows a poor application of CT parallel-
ing, using a two-restraint relay for the bus/
transformer combination. Here three sets of CTs
are paralleled and connected to a common
restraint winding R1, but where line 1 is a source
to the bus (e.g., a bus tie). Suppose CT 1
misperforms and delivers 50A rather than the
ideal 60A. The relay sees 10A of restraint and
10A of operate; hence, it trips (Fig. 19b). If
breaker 1 had been on its own restraint and
circuits 2 and 3 had been in parallel, the relay
would have seen 60A restraint and 10A operate,
and it would not have tripped (Fig. 19c).
17
Note 1: CTs that are shown in delta may be
connected in wye, and then have the relay
performthedeltacompensationinternally.
Note 2:Relays are shown in simplistic
fashionforconceptualpurposes.SeeFig.3
formorecompleteinternaldiagram.

Figure 19 - Improper CT Paralleling
4.7 Alternate Transformer Connections and
AssociatedCompensation
The proper configuration of the CT compensa-
tion connection was the major difficulty in
transformer installations through the mid 1990s.
However, with the advent of solid state and
numeric relays that perform the compensation
internally, the difficulty has moved to inside the
relay. To complicate the matter, modern relays
have been made flexible enough to compensate
for more than the common delta-wye transformer
with ±30° phase shifts. There are many trans-
formers that are connected in alternate methods,
especially outside the U.S., creating any phase
shift in 30° increments. For instance, Zig-Zag
windings, seen mainly in grounding banks in the
U.S. market, are sometimes used as power
transformer windings to introduce phase shifts
that may be unusual to some engineers. Modern
numeric relays need to be able to compensate
for any physically possible phase shift in a
process sometimes referred to as "around the
clock compensation." Reference 6 analyzes the
variousarrangementspossiblefortransformer
connections and reviews the mathematics of
how numeric relays, such as the BE1-CDS240,
compensate for these multiple possible sce-
narios.
4.8 Ground Fault Sensing
Ground faults on delta windings connected to a
solidly grounded power system are relatively
easy to sense, but time delays must be used
when set highly sensitive. Transient CT saturation
due to DC offset during inrush and high through
fault currents tends to create a false residual
current, so a time delay is needed that is
appropriate for the expected inrush duration and
sensitivity setting. Ground faults on ungrounded
systems must be detected via zero sequence
voltage (59N) and phase to ground voltage
relays (27ln and 59ln).
On wye windings, where impedance grounding
limits the ground fault current to levels below the
sensitivity of the phase differential, ground faults
in the transformer need special detection
schemes. The concept is sometimes referred to
as "restricted earth fault" protection. Two classi-
cal approaches for a sensitive ground fault
protection scheme are a) differentially connected
overcurrent relay that monitors ground currents
flowing into the two sides of a wye winding, and
b) a directional ground relay that compares the
phase relationship of ground current on each
side of the winding. Both concepts are available
in numeric relays such as the BE1-CDS2x0
(concept a), and the BE1-951/IPS (concept b).
However, the classical approaches are beneficial
because they give a more immediate under-
standing of the technical concepts.
In the classical approach to a differentially
connected overcurrent relay, the auxiliary CTs
are required to balance the residual ground
current from the phase CTs and the ground
current as seen on the neutral of the transformer.
In Fig. 20, the 20Ω resistor limits ground fault
current to 400A. The 2000:5 CT has 1A in its
residual connection. The aux CT multiplies the
current by 6.7. The 300:5 CT on the transformer
neutral also has 6.7A secondary but is wired to
be opposite in polarity so that the summation of
currents is 0 (approximately, given normal error
levels in the CTs). The relay does not operate. If
the fault had been internal, the summation would
have been above 0.
18

Figure 20 - Ground Differential, Classical
Approach
The 51/87N relay pickup in Fig. 20 possibly is
set for 0.5A based on a neutral current contribu-
tion of 0.67A relay current for a ground fault 10%
from the neutral end of the wye winding. Such a
fault will yield 40A primary and 0.67A secondary.
For this 40A fault, the 2000:5 CTs will only push
0.10A into the 87P relay, well below the allowed
minimum pickup of the 87P relay. Other 13.8 kV
ground sources, where available, increase the
level of relay current for an internal fault. How-
ever, the protection must cover the case with no
added current contribution.
One must keep track of burdens quite closely
when using aux CTs. The secondary burden on
the aux CT in Fig. 20 will be magnified by the
square of the current ratio, or 44 times, as seen
on the primary. However, while the ohmic
burden can be very high, the ground current
level is limited by the grounding impedance. For
example, a 0.5 ohm secondary burden reflects
to a 22 ohm primary burden, but the maximum
current on the 2000:5 CTs is just
400A/400:1 CTR= 1A secondary for an external
line-ground fault, yielding a burden voltage on
the 2000/5 CTs of 1*22= 22V.
An overcurrent relay used for the 87N function
will move toward tripping if the fault is internal,
for which the currents will no longer balance.
Such protection must use a delay (e.g. 30
cycles) to ride through the false residual current
resulting from the dissimilar performance of the
phase CTs during a phase fault, especially
during the DC offset period of a through fault
and during transformer inrush. In an impedance
grounded system, the phase fault current can be
100 times the maximum level of ground fault
current. Thus, it does not take much difference in
the performance of the phase CTs to create a
relatively substantial false residual current. For
example, in Fig. 21, a 1200 Amp phase A to B
external fault generates only 28A rather than the
ideal 30A in one of the CTs. The aux CT multi-
plies this to 13.3A in the 51/87N relay. If the
pickup of the 51 element had been the afore-
mentioned 0.5A, this false residual of 13.3A
would become 27 times pickup, giving a fast but
erroneoustrip.
Figure 21 - Example False Ground Differential
During Phase to Phase Fault
The 51/87N function in Fig. 21 could be re-
placed with a percentage differential function for
increased security against CT misoperations,
which is effectively done in a modern numeric
differential relay, such as the BE1-CDS2x0. A
numeric relay would also negate the need for the
aux CT. Such relays have the capability to
calculate 3I0 current from the phase CTs inputs
and multiply by the appropriate factor to allow
comparison to the current from a CT in the
neutral of the transformer without the use of an
aux CT, and to perform the percentage differen-
tial function. Using a percentage differential
approach gives added security to the trip
decision, but a percentage differential relay for
the ground differential function still can be
insecure during external phase to phase faults
and must be set with some level of time delay,
since the neutral current contributes negligible
restraintduringphasefaults.
An alternate method to sense internal ground
faults is to compare the phase relationship of 3I0
19

in the phase CTs to the ground current on the
transformer neutral. An internal ground fault and
an external ground fault will be marked by a 180°
phase shift of the two currents in relation to one
another. A BE1-67N relay is used to compare
the two currents in Fig. 22. Because the residual
current is highly distorted and the waveform
varies from cycle to cycle, directional operation
is intermittent. Each time the directional element
resets, it resets the time-overcurrent element.
The operate leg, terminals 7 and 8, may have a
pickup and time delay set to keep false opera-
tions from occurring. Classically, a solid state
BE1-67N has been used, but a numeric BE1-951
or BE1-IPS relay also can perform the function.
In Fig. 22, the aux CT is used to bring the
ground current as measured by the phase CTs
to a level that is measurable by a 5A nominal
relay. In a modern numeric relay such as the
BE1-951 and BE1-IPS, the ground/aux CT input
may be selected to be rated at 1A and con-
nected to monitor the phase residual, and the
phase CT inputs can be selected to be rated at
5A and connected to the transformer neutral CT.
In this approach, the aux CT could be eliminated
from the scheme.
Figure 22 - BE1-67N as ground directional relay
5. Turn-to-Turn Faults
Phase differential relays may detect a turn-to-turn
fault because the fault changes the transformer
turns ratio. Ground differential relays do not
respond to such faults. A neutral overcurrent
relay may see fault current if an external ground
source exists and the resultant voltage unbal-
ance creates an excessive load unbalance.
For an impedance grounded system, most of
the fault current probably will be contributed by
the delta-side source. A single turn fault may
produce a total less than rated current (Ref. 8). A
turn-to-turn fault may not be detectable initially.
However, a sudden pressure relay (SPR) may
give early detection. The SPR will detect any
abnormality that generates a sudden increase in
pressure due to gas generation (e.g. arcing due
to a loose connection).
6. Sudden-Pressure Relays (63)
Figure 23(a) shows a SPR that detects an
increase in gas pressure, typically applied on
gas-cushioned transformers of about 5 MVA and
higher. The gas pressure is generated by an arc
under the oil, producing decomposition of the
oil into gas products. The change in pressure
actuates bellows 5 closing microswitch contact
7. Equalizer port 8, much smaller than the main
port 4, prevents bellows movement for slow
changes in gas pressure due to ambient tem-
perature changes and load cycling.
Figure 23(b) shows use of the break contact of
the microswitch (63) in conjunction with auxiliary
relay 63X. This circuit prevents tripping for a
flashover of the make contact of 63.
Figure 23 - Sudden Pressure Relay
20

An alternate design similar to that of Fig. 23(a)
mounts the relay within the oil either in gas-
cushionedorinconservator-typetransformers.
There are separate sudden pressure devices for
use in auto tap changer compartments, but such
devices need to be designed so they do not
operate under the presence of the normal arcing
and mechanical operations that occur within the
tap changer.
The SPR is designed to respond only to arcs
within the oil. While more sensitive than a
differential relay, the SPR is not as fast as the
differential relay for some faults. Since redun-
dancy of protection is the mark of a good
protection scheme, both relays should be
applied.
Some users have experienced misoperations of
suddenpressurerelays.Duringhigh-current
external faults, winding movement generates an
oil pressure wave that can tend to cause relay
operation. Earthquakes have been reported to
cause operation. As a result, some users con-
nect them only to alarm. Their security has
improved by installing them on stiffer sections of
the tank and by adding current supervision logic
(e.g., block 63 trip for any high current level that
may be indicative of an external fault that could
cause high winding movements; if the fault is
internal, differential relays can be assumed to
trip rapidly). Security can also be improved by
performing regular maintenance to ensure they
are calibrated properly and not failing or drifting
in a manner that makes them excessively sensi-
tive. There have been cases where a relay
operation has been a precursor to transformer
failure; e.g., a relay that operates for an external
fault may be an indication of winding stabiliza-
tion blocks that have come loose.
Conservator-type power transformers do not
have a gas cushion within the main tank. In-
stead, the cushion resides in a separate auxiliary
tank. A gas accumulator relay ("Bucholz") can be
installed in the pipe connecting the main and
auxiliary tank to detect the generation of gas.
This relay has two elements, an accumulator
alarm and a trip function. The accumulator,
which stores a portion of the gas, provides an
alarm for slowly developing conditions. A baffle
in the pipe actuates the trip element for relatively
fast gas flow to the auxiliary tank.
7. Monitoring for Incipient Problems
The oil in large transformers is normally
checked, as part of routine maintenance, for the
existence of abnormal chemicals and gasses
that are created as a result of oil contamination,
insulation breakdown, and internal arcing. A
number of on-line devices has been developed
in recent years to detect incipient conditions
thatthreatenseriousconsequences.These
include gas-in-oil analysis, acoustic partial-
discharge detection, moisture sensor, tap-
changer-operationsupervision,andpump/fan
supervision.
Reference 9 reviews gas in oil analysis, and Ref.
10 reviews online monitoring schemes. The
website where these articles are found has an
extensive collection of other good articles on oil
analysisandtransformeroperation.
8.OvercurrentRelays
Fig. 1 shows a number of overcurrent relays.
When an 87P is in place in this application, the
50/51P on the primary side is considered
backup protection. In a radial distribution
application, the overcurrent relays on feeders
and buses are also part of the transformer
protection scheme since they limit the duration
of out of zone faults.
The discussions below include references to
50/51Q elements. The Q refers to elements that
respond to negative sequence current compo-
nents. See Section 8.5.
8.150/51TransformerBackup
The CTR for this application is a compromise
between full load current and limiting current in
the CT during faults near the transformer bush-
ings. See Section 10.2 for a discussion on the
CT selection in this case.
In Fig. 1, the 50P and 50Q elements would be
set at approximately 150% to 200% of the
current that will be sensed at 115kV for a 13.8kV
three phase and phase to phase fault. These
elements must not trip for low side faults. A
setting in this range normally would be above
21

the inrush current of the transformer. For in-
stance, assume an impedance of 7%, implying a
secondary fault will cause, at most, 14.3pu fault
current. A setting of 150% would imply a 50P
setting of 21.4pu. This is higher than the typical
inrushcurrent.Someresourcesmentioninrush
current reaching as high as 25xFLA, but this is
typically only the first half cycle, is very harmoni-
cally rich and contains a high DC offset. Relays
responding strictly to fundamental frequency
currents (e.g., most modern numeric relays) will
see an inrush that is less than a relay that
responds to the harmonic and DC component of
inrush. The relay sensing algorithms need to be
part of the setting thought process. This same
concept was also previously discussed in
reference to the 87U element, Section 4.5.
Since the 115kV winding on T1 is delta, no
ground current should be seen at the transformer
backup, so 50/51N elements in the relay can be
set fairly sensitive and fast. However, there may
be some transient false residual due to CT error
during transformer inrush or 13.8kV faults, so the
50/51N elements need to be time delayed
somewhat, possibly about 4 times the system
L/R time constant. Due to the long duration of
inrush current and its associated DC offset
component, highly sensitive ground current
pickup levels should be delayed substantially
longer. In power system cycles, the L/R time
constant is:
T.C., cycles = (X/R ratio of Zfault)/(2*π)
where Zfault refers to the system source imped-
ance looking back from the high side of the
transformer.
The 51P relay time element must be set to carry
the maximum expected load current. Since a
transformer is capable of carrying considerable
overload for a short period, a high pickup is
normally called for (e.g. twice the forced-cooled
rating). The time unit should coordinate with the
51P Bus Backup relay or the 51P Partial Differen-
tial relay (Section 8.2), depending on which is in
use.
The 50/51P operating time needs to be faster
thanthethrough-fault(externalfault)withstand
capability of the transformer (Fig. 26). See Ref. 1
for additional transformer damage curves. Figure
26 illustrates both "frequent" and "infrequent"
limits and recognizes the cumulative effect of
thesestresses.Transformersonsystemswith
underground feeders likely have fewer faults
relative to systems with long overhead wires.
Higher magnitude faults would be less common,
and lower magnitude faults would be more
common, especially on systems with long
feeders.
Because of its high pickup and slow operation,
the 51P element provides relatively poor protec-
tion (compared to other relaying schemes) for
transformer winding and tap changer faults.
Accordingly, this element is not a good substi-
tute for differential and sudden pressure relays.
The consequences of a slow cleared fault
include the threat of an oil fire due to a ruptured
tank or bushing explosion and the higher internal
damage that will occur. Removal generally is
necessary for even a fast cleared winding fault.
This is not the case for a tap-changer flashover
that is cleared before winding damage.
8.2 Bus Backup, Bus Tie, and Partial
Differential 51 Relays (and 50T)
The overcurrent element at this level does not
directly protect the transformer, but it performs a
service of limiting the duration of faults that will
cause cumulative damage to the transformer, as
discussed in the previous section relative to the
transformerdamagecurve.
Breaker D, Fig. 1, would be considered the Bus
Main breaker, but the 51 relay on this breaker
would be considered the Bus Backup relay
under the understanding that there would be
some dedicated bus protection scheme in
place, such as a bus differential (87B) or the
interlocked 50 elements discussed below.
A 50 element normally is not used in these relays
due to coordination issues with feeder relays.
However, a 50T (a 50 element with a small time
definite time delay) is sometimes used in the bus
backup relay in a bus protection scheme that
interlocks the bus and feeder relays. In the
scheme, the bus backup 50 element is delayed
long enough for a feeder relay to communicate
back to the bus relay, telling it, "Do not trip, I
see a fault on my feeder."
22

The partial differential 51 element in Fig. 1
measures the sum of the transformer and bus-tie
breaker currents. Such a connection is appropri-
ate with a normally-closed bus-tie breaker. When
the bus tie is closed, the partial differential sees
the full current going to a feeder fault; hence, it
more directly coordinates with the feeder relays
than a relay that only monitors the current in a
single transformer. A relay such as the bus
backup or bus tie that sees only a portion of the
current will trip slow relative to the partial
differential relay given the same pickup level.
If the bus main and bus-tie breakers are inter-
locked to prevent the secondary of both
transformers from being tied together, the bus
tie overcurrent relay is coordinated to be slower
than a feeder relay but faster than the bus
backup relay. This adds an extra level of
overcurrent protection in the TCC curve that
effectively makes either the transformer relay
slower or the feeder relay faster, which adds a
compromise to the TCC coordination scheme.
If a dedicated high speed bus protection
scheme is in place (e.g., bus differential or the
aforementioned interlocked bus 50 element
scheme), the coordination compromise associ-
ated with including the bus tie in the TCC is
subject to question.
In substations with multiple operating modes,
such as run with and without the tie closed,
smart relays that can change settings on the fly
can be utilized. This feature is found in modern
numeric overcurrent relays with multiple settings
groups, such as the BE1-951, BE1-IPS, BE1-851,
andBE1-700C.
8.3TransformerNeutralOvercurrent
The transformer neutral overcurrent relay in Fig. 1
primarily backs up the bus and transformer
protection scheme and faults on the lead
between the transformer and the 13.8kV bus. In
the absence of the 87N application, in imped-
ance grounded systems the relay provides the
primary ground fault protection for faults close
to the neutral on the transformer low voltage
winding. The relay also backs up the 87T,
depending upon the sensitivity of the 87T to
ground faults. The relay must coordinate with the
bus backup 51N to allow the latter to clear a bus
fault first without tripping one or both transform-
ers.
If the 13.8kV bus tie can be closed with both
transformersinservice,thetransformerneutral
relays on both transformers will operate for a
13.8kV winding or lead fault, unless a 67N or
87N, Section 4.8, is provided that is configured
for faster clearing.
8.4 67/51 and 32 Relays
The 67/51P relay operates for current (and, to
some extent, power) flowing from the transformer
low side toward the high side, i.e., backfeed.
Such flow could occur if the 115kV tie breaker
opens and both transformers remain energized,
and the 13.8kV bus tie is closed. Reverse flow
also can occur with or without the 115kV tie
breaker closed, with local 13.8kV generation.
Reverse current flow also can occur during a
transformer fault. When backfeed under fault
conditions is the concern, the 67/51P element is
the appropriate element to use (rather than a
32). A 67/50P element also may be appropriate,
but the 50 element should be time delayed
slightly if there is any chance of transient
backfeed during load swings or when paralleling
with the other transformer. Since normal load
flow is toward the low side, the 67/51P can be
set more sensitively than bus backup 50/51 and
may also be faster.
A reverse looking 67/51P has some ability to trip
for forward faults and load flow. The element's
maximum torque angle (MTA) typically is
skewed for fault detection, as seen in Fig. 24. In
Fig. 24, the MTA of the relay has been set very
low, maybe 10°. Note the relay sees ±90° from
the MTA as a forward current flow. In this ex-
ample, the 67/51P element reaches very little
into the opposite plane. If the relay MTA had
been set at the more typical 45° to 80° setting,
the reverse looking element would reach well
into the plane of forward power flow and, hence,
would be at greater risk of tripping for normal
load flow. To prevent this risk, one practice is to
set the 67/51P element pickup high enough so
that it will not pick up for any normal load flow
conditions or use a directional power element,
the 32, to sense actual power flow toward the
transformer.
23

Figure 24 - 67 Element MTA vs. Load Flow
The 32 device is designed to be very selective in
the direction of power flow. Ideally it is nonre-
sponsive to VAR flow, though heavy VAR flow in
some relay designs can swamp the relay sensing
algorithms and reduce sensitivity to true power
flow. In fault conditions, watt flow may be low
and the sensed current may be mostly VARs.
Coupled with the reduced sensitivity of the 32
element to watt flow in some designs under the
presence of high VAR flow, a 32 may respond
poorly to fault conditions, so a 32 element
normally is not used for fault detection. Hence,
the overall solution to sensing current
backfeeding into the transformer is both a
67/51P and a 32 element. Both of the 67/51 and
32 functions may be found in the BE1-IPS and
BE1-951.
Numeric relays such as the BE1-951 and BE1-
IPS relays do not have the limitation mentioned
above relative to sensing watt flow under the
presence of high VAR flow.
Recall also that a 67N, looking for ground
current flowing toward the transformer, as
discussed previously in Section 4.8, is a means
of sensing transformer ground faults. In Section
4.7, the directional element used the phase
angle relationship of 3I0 in the phase leads vs.
Ig in the transformer neutral. However, a 67N
relay more commonly is configured in the
design stage to see only local phase voltages
(l-l or l-n, depending on design or configuration)
and phase currents and may not have access to
transformerneutralcurrent.Thedirectional
decision for current flow is most commonly
either current vs. quadrature voltage (solid state
relays) or V2 vs. I2 (numeric relays). See Refs. 2
and 11 for a technical description of how these
two directional elements are designed, as well
as other supporting information in regard to the
67 element.
8.5NegativeSequenceOvercurrent
The advent of and high use of numeric relays
that readily calculate negative sequence current
(I2) has resulted in an increased use of negative
sequence overcurrent relaying. The device is
named 50/51Q herein, but sometimes referred to
as the 46 device. The problem with the term 46
is that it is unclear as to whether a 51 or 50
device is referred to.
The benefit of a 51Q in delta wye transformer
applications is that the I2 component of fault
current is identical, on a per unit basis, on both
sides of a delta/wye transformer (see Section
10.3 and Fig. 25). The 51Q on the high side of a
delta/wye transformer easily can be coordinated
with 51Q elements on the low side for phase to
phase and phase to ground faults. Unlike phase
overcurrent elements, the 51Q does not need to
be set above load current levels and instead
would have a pickup similar, on a per unit basis,
to a ground overcurrent element on a solidly
groundedsystem.
The drawback of the 50/51Q is that it adds some
level of work to the coordination analysis. If set
below load current, recall a 51Q may need to
coordinatewithdownstreamphaseovercurrent
devices that do not monitor I2, and that will only
trip when current rises above normal load
current, such as fuses. One must compare the
51Q response to the 51N relays as well as these
phaseovercurrentelements.
9. Thermal Protection (49)
Conventional thermal relays measure the oil
temperature and transformer current to estimate
the hot-spot temperature. They provide an
indication and means for controlling pumps and
fans. Typically, these devices provide two tem-
perature sensing levels for control and a third,
higher temperature, sensing level for alarm or
tripping.
Recently developed fiber optic sensors, incorpo-
rated in the transformer winding, provide a direct
method of measuring the hot-spot temperature.
About four of these sensors would provide good
coverage.
24

10AssociatedIssues
10.1 Harmonics During CT Saturation
Saturation of CTs happens in terms of AC and
DC components of current. The AC saturation
effect refers to the ability of a CT to reproduce
symmetrical current conditions. The DC satura-
tion effect refers to the saturation that occurs
when the current contains the decaying DC
component associated with a fault, magnetizing
inrush, motor starting, or generator synchroniz-
ing. A CT that experiences negligible distortion
under symmetrical AC conditions might become
saturated and give a severely distorted output
when a DC component exists (Reference 12 is a
good resource but hard to access. Reference 13
provides a somewhat more accessible but less
detailed discussion of the matter).
While faults generally produce the most current,
other conditions, such as a motor starting,
produce much slower DC decay than occurs for
a fault. A smaller DC current that persists longer
can also produce DC saturation. For these
external disturbances, unequal times to satura-
tion, and saturation level, in various CTs results
in false operating current. Either the harmonic-
restraint or the percentage differential restraint
(fundamentalfrequencycharacteristic)prevents
unwanted tripping for this condition.
Under high symmetrical current conditions that
drive a CT to AC saturation (e.g., an attempt by
a CT to drive current that in turn causes voltage
above the CT's kneepoint), CT distortion gener-
ates odd harmonics but no even harmonics. A
CT experiencing DC saturation during an asym-
metrical fault develops both even and odd
harmonics. If a relay is designed to restrain on
odd harmonics (e.g. 5th), it may fail to operate if
the harmonic content from the saturated CT
exceeds the relay's threshold for restraint. Relays
that restrain on even harmonics may be tempo-
rarily restrained until the DC decays enough to
allow the CT to perform again correctly. High-set
unrestrainedelements(87U)supplementthe
restrained elements, so that high current faults,
where CT saturation can be severe, can be
cleared independent of any harmonic restraint.
For satisfactory protection, harmonic generation
by the CTs should not exceed the restraint level
for a current below the unrestrained element
pickup. Poor CT quality can detract materially
from the reliability of the differential relay.
10.2 CT Ratio and CT Accuracy Selection
Let "Ks" be the ratio of a CT kneepoint to the
burden voltage. The knee point herein refers to
the 45° slope point on the CT excitation curve.
The burden voltage in this ratio is the RMS (AC
only, no DC) voltage in the CT loop during a
maximum level fault, and where the CT internal
resistance is included in the CT loop resistance.
If one does not have CT excitation curves, a
likely satisfactory substitute for the knee point is
to assume that the knee point is equal to the C
class of the CT (though in practice, the knee
point ranges from about 70-150% of the C
class). The higher the Ks value, the better the CT
performance. The purpose is to give a high level
of margin in the CT rating so that it can provide
reasonable ability to reproduce the DC offset in
a fault.
For differential protection, a good objective is
Ks=8 or higher for a current at the unrestrained
pickup level during all out-of-zone faults. For in-
zone faults, a Ks of 2 is desirable. Some re-
sources suggest a Ks of "1+X/R" (where X/R is
found from the system impedance looking back
from the fault location) will give reasonably good
assurance of limited risk of DC offset induced
saturation. Reference 13 contains a derivation of
the 1+X/R factor. High Ks values called for by
the 1+X/R factor can create a demand for CT
performance that is sometimes difficult to meet.
It is difficult to generalize about when it is
important for CTs to be highly rated for an
application and, hence, have a high Ks. It is
always very desirable to have a Ks well above 1,
but obtaining a CT resistant to DC offset effects
may be hard to justify. For instance, in a phase
overcurrent application, transient CT saturation
due to DC offset will cause only a small delay in
tripping speed, so the effect of CT error is
generally small; hence, a low Ks is likely satisfac-
tory. However, if a sensitive ground overcurrent
relay is connected in the residual of the phase
CTs, the effect of CT saturation is to false trip, so
a selected CT must be very well rated and able
to reproduce substantial DC offset without
saturating, hence calling for a high Ks.
25

10.2.1 Selection of CT ratio in Figure 1
IssueswithCTperformanceunderfaultcurrent
commonly arise when a relatively low MVA
transformer is placed on a powerful bus. This
cause is an incongruent demand between
selecting a CT ratio that allows good reproduc-
tion of full fault current without excessive CT
secondary current, and selecting a CT ratio that
allows the secondary current to be some mea-
surable level during normal load conditions.
For instance, in the system seen in Fig. 1, the
load current at 30MVA and 115kV is 151A,
implying a CTR of about 200:5 would work well.
However, suppose the fault duty at 115kV was
15,000A. If we wish to limit CT secondary current
to maybe 100A, a CTR of about 800:5 would be
selected. However, at 800:5, the FLA from the
CT would only be 0.94A, which implies measur-
ing accuracy would suffer at light load condi-
tions. A CT actually can carry quite a bit more
than 100A for a short period, so a compromise
selection of 400:5 was chosen for the example.
The burden and CT voltage rating must be
checked during peak fault currents. Assume a
1200:5 CT is used but tapped at 400:5. Assume
the full winding C class is C400, and the internal
resistance is 0.6 ohms. At a tap of 400:5, the
effective C class is C133, and one can assume
that the kneepoint is roughly 130V for lack of an
appropriate excitation curve. Since we are using
only 1/3 of the CT windings, the internal resis-
tance is 0.2 ohms. Assume external burden is 0.4
ohms. Given our assumed 15,000A fault, the
secondary current, if the CT operates ideally, is
187.5A. This is quite high, but most CTs can
withstand it for short periods. The short time
rating of CTs may not be published in common
literature, and the manufacturer should be
consulted on the short time current rating of the
CT (typically given for maximum primary current
for 1 second with a shorted secondary), espe-
cially if fault current rises above 100A secondary
and delayed tripping will be utilized at this
current level. The voltage drop in the CT resis-
tance and external burden is
187.5*(0.2+0.4) or 112.5V. This gives a Ks of
about 1.15. The CT is at risk of entering into
saturation, especially if any DC exists in the
current. We have a relatively weak CT for the
application. Our choice is to either raise the CT
ratio and limit accuracy during normal loads,
raise the CT C class to C800, or accept the Ks
value with the assumption that the CT is rated
well enough to reproduce the current to allow
the unrestrained differential element to rapidly
clear the fault. This becomes an engineering
judgment decision beyond the scope of this
document.
Next, we need to analyze what occurs in the
same CT during an out of zone secondary fault.
Assume the transformer impedance is 8% on the
unit's 18MVA base. For a three phase fault, the
current at 115kV will be about 1/0.08pu, or
1130A at 115kV. This gives us 14.1A from the CT
at 400:5, and the voltage in the CT loop (at 0.6
ohms) is only 8.5V and, given a knee point of
130V, the Ks is about 15, which should work well
under very high DC offset conditions. This gives
added reason to accept the CT as is.
10.3 Delta-Wye Winding Effects on Primary
Current
For delta/wye transformers, fault current as seen
on the primary for secondary ground faults is
substantially lower than for secondary phase to
phase and three phase faults. As those who have
worked in fault analysis are aware, lines on the
delta side of delta/wye banks see 0.577 per unit
current for secondary ground faults, compared
to the current seen by the line during three
phase faults on the secondary of the same
magnitude. The 0.577 factor for ground faults
occurs because ground faults are single phase
events; hence, the current is multiplied across
the transformer by the transformer turns ratio,
rather than by the line-line voltage ratio (the
turns ratio on a delta wye bank is 1/sqrt(3)
relative to the line to line voltage ratio).
Figure 25 shows the per unit phase and se-
quence component currents seen on the delta
side for various wye side faults, for a DAB delta
transformer. Note that positive and negative
sequence magnitudes are the same on both
sides of the transformer but shifted in opposite
directions.
References
Reference 1 provides a reference and bibliogra-
phy section of almost 100 articles and IEEE
Standards on almost every topic associated with
26

transformer protection and is recommended for
review for one who is making a career in protec-
tive relaying. Web searches will invariably find
even more references. The IEEE Xplore ware-
house of technical articles is another resource for
one who has appropriate subscriptions to IEEE.
One should also review the instruction manuals
for various Basler relays, and the variety of
Application Notes and Technical papers avail-
able at www.basler.com. Besides Basler's
resources,othermanufacturers,engineering
companies, and government agencies provide
articles and application guides associated with
transformerprotection.
Publication dates and/or revision levels of the
texts and standards listed are not supplied, as
these are routinely updated; the latest revision
should generally be sought out.
1. ANSI/IEEE C37.91, IEEE Guide for Protective
Relay Applications to Power Transformers.
2. Lewis Blackburn, "Protective Relaying:
Principles and Applications," Marcel Dekker,
Inc.
3. S. Horowitz and A. Phadke, "Power System
Relaying," John Wiley & Sons, Inc.
4. ANSI/IEEE C37.95 "IEEE Guide for Protective
Relaying of Utility-Consumer Interconnec-
tions."
5. National Electric Code, published by NFPA.
6 "Three Phase Transformer Winding
Configurations and Differential Relay
Compensation", available in the technical
resource library of www.basler.com.
7. W. K. Sonnemann, C.L. Wagner and G.D.
Rockefeller, "Magnetizing Inrush Phenomena
in Transformer Banks," AIEE Transactions,
Vol. 77, pt. III, pp 884-892, Oct. 1958.
8. Klingshorn, H.R. Moore, E.C. Wentz,
"Detection of Faults in Power Transformers,"
AIEE Transactions, Vol. 76, pt. III, Apr. 1957,
pp 87-98.
9. "Dissolved Gas Analysis - Past, Present and
Future," F. Jacob, Weidman-ACTI. Available
at http://www.weidmann-acti.com/
knowledge_center/.
10. "An Overview of Online Oil Monitoring
Technologies." 4th Annual (2005) Weidmann-
ACTI Technical Conference. Available at
same site as Ref. 9.
11. "Directional Overcurrent Relaying Concepts."
Available in the technical resource library of
www.basler.com.
12. IEEE Committee Report, "Transient
Response of Current Transformers," IEEE
Special Publication, 76CH1130-4PWR.
13. "Bus Protective Relaying, Methods and
Application," available in the technical
resource library of www.basler.com.
Figure 25 - Delta Side Currents for Wye Side Faults
27

Figure 26 - Example Transformer Damage Curve
Basler Electric Numeric Relays
The numeric relays contain too many features and capabilities to be well represented in a simple listing of
features. Some of the functions listed are optional and dependent upon the relay style that is purchased.
Refer to the product bulletins and instruction manuals for a more complete listing of capabilities and
ordering options. Some of the relays listed would not normally be used directly with a transformer
protection scheme, but they may exist in the scheme due to other equipment in the area of the trans-
former; hence, are included due to possible general interest to the protection engineer. The relays listed
are available in cases designed for either 19" rack mount or 19" half rack, depending on the relay model,
and many are also available in cases designed for fitting into the cutout of older GE and ABB/WH relays,
a particular feature of Basler Electric products that may notably reduce installation costs.
28

Function
Model (ANSI Device if applicable) Comments
BE1-CDS220 87P, 87U, 87N Targets a three phase, two winding transformer
Internal phase compensation for common delta/wye bank design
50P/Q/N,51P/N/Q 3 independent copies of each 50 and 51 element, assignable to any input.
BF BreakerFailure
OtherFeatures Programmable logic
Metering, detailed event reporting, oscillography, remote monitoring, multiple I/O,
4 setting groups
BE1-CDS240 87P, 87U, 87N Three phase, up to 4 windings
87 elements can monitor summation of any set of CT inputs.
Internal phase compensation for any transformer connection configuration.
Two copies of 87N for monitoring two independent wye windings
50P/Q/N, 51P/N/Q, BF 4 independent copies of each 51P/Q, 5 for 51N, 8 for 50P, 4 for 50Q, 5 for 50N
Current elements can monitor summation of any set of CT inputs.
50BF Advanced Breaker Failure, 4 copies
24, 27, 47, 59, 59N, 81 Voltage monitoring standard in all models. Multiple independent copies of several
elements.
4 setting groups
BE1-851 50P/Q/N,51P/Q/N General purpose overcurrent monitoring. Two copies of each 50 element.
BF BreakerFailure
79 4 shot reclosing
4 setting groups
BE1-951 50P/Q/N,51P/Q/N General purpose overcurrent monitoring. Two copies of each 50 element.
67 control of all elements
BF BreakerFailure
24, 25, 27, 47, 59, 59N, 81 Multiple independent copies of several elements.
32 Forward/reverseoverpower
79 4 shot reclosing
4 setting groups
BE1-IPS100 50P/Q/N,51P/Q/N General purpose overcurrent monitoring, ideal for intertie protection
67 control of all elements Two copies of 51P, 51N
Two copies of each 50 element
BF BreakerFailure
24, 25, 27, 47, 59, 59N, 81 Multiple independent copies of several elements. 81 has rate of change.
32 Two copies 32 Forward/reverse, over/under power
79 4 shot reclosing (PI and RI)
2 setting groups
BE1-700C 50P/Q/N,51P/Q/N General purpose overcurrent monitoring. Two copies of each 50 element.
BF BreakerFailure
79 4 shot reclosing (optional)
2 setting groups
Ethernet/Internet access
BE1-700V 24, 25, 27, 47, 59, 59N, 81 Multiple independent copies of several elements.
79 4 shot reclosing
2 setting groups
Ethernet/Internet access
29
Table III - Basler Numeric Relays

Solid State Relays
The list below is not an exhaustive list of Basler solid state relays, but it shows the relays of most common
interest. Some of these relays are not directly applicable to transformer protection, but they are included
due to possible general interest to the protection engineer.
ANSI Device
Relay Model Function Comments
BE1-25 25 Synchronism check
BE1-27 27 Single phase, connected line to line or line to neutral
BE1-32 32 Wide range of sensitivities available; see options table.
Available as over or over/under
BE1-40Q 40 Generator Loss of Excitation
BE1-46 46/51 Inverse time using an I2
t curve
BE1-47 47 Definite time
BE1-50 50 Single or multiple phase
BE1-50/51B 50/51 CT powered; ideal for AC powered switchgear.
Available as a direct GE IAC and ABB/WH CO retrofit.
BE1-51 50/51 Wide variety of models available; single phase, three phase, three
phase and neutral.
BE1-51/27R/C 51 Similar to BE1-51 but has voltage control and voltage restrained
function; generally used in generator protection.
BE1-59 59 Single phase, connected line to line or line to neutral
BE1-59N 59 Design used for monitoring generator neutral and broken delta VT
BE1-59NC 59 Design for use in capacitor neutral voltage monitoring scheme
BE1-64F 64 Generator field ground.
BE1-67 67/51P Similar in functionality to the BE1-51 but uses quadrature voltage
technique for phase overcurrent directional control
BE1-67N 67/51N Similar in functionality to a single phase BE1-51 but uses Vo and Ig
inputs for directional control
BE1-79M 79 4 shot reclosing relay
BE1-81O/U 81 Main application is load shedding and islanding detection.
BE1-87B 87 High impedance bus differential
BE1-87G 87 Generator differential
BE1-87T 87T Transformer differential. Available in single phase up to 5 input
or three phase up to 3 input.
Table IV - Basler Solid State Relays
30

Transfguide transformer protection application guide

More Related Content

What's hot (18)

Similar to Transfguide transformer protection application guide (20)

Recently uploaded (20)

Transfguide transformer protection application guide