1

Fundamentals of Short-Circuit
Protection for Transformers
Bogdan Kasztenny, Michael Thompson, and Normann Fischer, Schweitzer Engineering Laboratories, Inc.

Abstract—This paper reviews principles of protection against and connections of CTs. This enables an in-depth
internal short circuits in transformers of various constructions. understanding of the known current differential compensation
Transformer fundamentals are reviewed as pertaining to rules for standard transformers and teaches how to develop
protection. In particular, the electromagnetic circuit of a
transformer is reviewed that links the terminal currents, winding
proper protection equations for any arbitrary transformer.
currents, fluxes, and ampere-turns (ATs) in a set of balance
equations for a given transformer. These balance equations are II. TRANSFORMER FUNDAMENTALS
used to explain the sensitivity of protection to various types of It is good to go back to basic principles to refresh ourselves
transformer faults. The paper shows that the classical
transformer differential compensation rules have roots in the
on how a transformer works before discussing protection
first principles—they reflect the AT balance of the protected concepts. To maintain a high level of focus, we will limit the
transformer. The rule of building transformer differential discussion to sinusoidal alternating current.
protection equations following the AT balance is used in this
paper to derive differential equations for autotransformers; A. Refresher on Transformer Theory
power zig-zag, Scott-T, and Le-Blanc transformers; and phase In its most basic form, a transformer is two or more coils of
shifters. The restricted earth fault (REF) and negative-sequence conductor in close proximity to each other, such that the
transformer differential (87TQ) functions are explained as a magnetic fields generated by the current in each coil are
means to detect ground faults near the neutral and turn-to-turn
faults, respectively.
linked. Almost universally, a magnetic core is included to
maximize the coupling of the fields between the coils.
I. INTRODUCTION But let us step back one more level. Faraday’s law tells us
that a voltage (electromotive force) is created when either a
The differential protection principle applied to power conductor is moved through a magnetic field or a magnetic
transformers requires accounting for transformer winding field is varied through the conductor. Equation (1) gives the
connections. With reference to transformers of standard
special case of sinusoidal current:
winding connections, these protection rules are commonly
referred to as ratio matching, vector group compensation, and v ( t ) = −n • ω • Φ • cos (ω • t ) (1)
zero-sequence removal. where:
Many protection engineers face difficulties applying these
n = number of turns in the coil.
principles to transformers of nonstandard or unusual
connections (e.g., phase shifters, Scott-T transformers, power ω = angular velocity of the sinusoidal function.
zig-zag transformers) under unusual locations or connections Φ = magnetic flux.
of current transformers (CTs) or under special configurations, The amount of magnetic flux, Φ, in (1) is given by (2):
such as a tertiary delta winding not accounted for in the n •i
differential scheme. When offered a choice of CT locations in Φ= (2)
ℜ
a nonstandard transformer, an engineer may lack the expertise
to select an optimum location, given the targeted protection where:
sensitivity and the intended protection method. i = current.
In addition, it is typically challenging for some ℜ = reluctance in the magnetic path.
practitioners to analyze the coverage and sensitivity of a given In (2), n • i is the magnetomotive force (mmf) that drives
short-circuit protection scheme. For example: Is the buried the flux and is measured in ATs. Equation (2) is the magnetic
tertiary covered by differential protection? Is the restricted circuit corollary to Ohm’s law for electric circuits.
earth fault (REF) sensitive to interwinding faults? Do the An inductor is a single conductor arranged in a coil, such
sensitivity and coverage of a given differential scheme depend that the magnetic field generated by current in each turn
on the type of transformer core (three leg versus five leg) or results in magnetic flux cutting the other turns in the coil. In
the grounding method for wye-connected windings? an air core inductor, the reluctance is relatively high, so the
This paper explains principles of short-circuit protection magnetic flux is relatively low for a given current level. The
for transformers and autotransformers by deriving proper result is a relatively low induced voltage in the coil that
balance equations for differential protection from the ampere- opposes the current flow [note the minus sign in (1)]. Because
turn (AT) equations of a healthy transformer. A number of this opposing voltage is proportional to the current flowing in
standard and nonstandard winding configurations are the coil, we characterize this as impedance.
considered, assuming different core types, as well as locations
 2

Now, if we introduce a magnetic core for the coil that circuit together. According to Kirchhoff’s voltage law, the
creates a low-reluctance path for the magnetic flux, the analogy in Fig. 2 is that the two voltage sources must sum to
amount of flux per ampere is high, and therefore the voltage zero around the voltage loop. Similarly, ATP and ATS must
opposing the current flow per ampere is high. The coil now sum to zero in the magnetic circuit along the closed loop core
appears as an almost open circuit. The current that flows in path.
this case is the so-called magnetizing current that we often
Φ
neglect in our calculations. +
Next, let us introduce a second coil on the magnetic core. Short Circuit
ATP
This creates our basic two-winding transformer. Because the
two coils are on the same core, the flux cutting them is nearly –
–
the same. Also, because of the concentrating effect of the
ATS
magnetic core, the volts per turn in both coils are nearly the
same. We use this characteristic to create the familiar +
transformer, which provides different voltage levels at the coil Fig. 2. Electrical analogy to Fig. 1.
terminals by the ratio of the number of turns in each coil.
This is a good time to introduce the concepts of mutual flux Now, let us close the electrical circuit on the secondary
(ΦM) and leakage flux (ΦL). The coupling between the coils is terminals in Fig. 1 through a load impedance. Because there is
never perfect. In Fig. 1, the flux shown as ΦPL, generated by a voltage across the secondary terminals, current iS flows
nP • iP, does not couple to the secondary coil. Similarly, the through the load impedance, and iP flows in the primary coil,
flux shown as ΦSL, generated by nS • iS, does not couple with such that the ATs sum to zero in the core leg (neglecting the
the primary coil. This flux is called leakage flux, and it results magnetizing current, which sets up the mutual flux). This
in a voltage being induced in each coil that opposes the flow allows a power transfer between the windings through the
of current, per (1). This is represented as the leakage magnetic core.
impedance in the equivalent circuit of the transformer. C. Three-Phase Systems
ΦPM
Most of the systems that we deal with are three-phase
nP
systems. Similar to electrical systems that can be built as
iP
three-wire delta or four-wire wye (star) systems, the magnetic
vP circuit of a three-phase transformer can be built as a three-
legged core or with a low-reluctance return path. Examples of
this type of core include four-legged, five-legged, shell type,
iS
and three single-phase cores. To help maintain the familiar
vS concepts of three-wire versus four-wire systems, we will refer
to all core construction that provides a low-reluctance return
ΦSM
path as a four-legged core. As discussed in Section II,
nS Subsection B, a return (fourth) core leg acts as a magnetic
short circuit that forces the AT to sum to zero on each core
Fig. 1. Fluxes in a transformer.
leg, neglecting the leakage flux:
B. Magnetic Circuit Parameters AT1 = AT2 = AT3 = 0 (3)
It is also useful to review magnetic circuits before going
on. One way to help electrical engineers understand the where:
magnetic circuit is to equate its basic parameters to its AT = the sum of ampere-turns in all coils on the same
equivalent electrical circuit parameters. Referring to (2): core leg.
• The flux, Φ, in a magnetic circuit is equivalent to the 1, 2, 3 = the three core legs with coils.
current in an electric circuit. A three-legged core requires further analysis. We use
• The AT quantity, n • i , in a magnetic circuit is symmetrical component theory to analyze unbalanced three-
equivalent to the voltage in an electric circuit. phase electrical systems. This technique is useful in
• An unsaturated magnetic core has low reluctance, ℜ , understanding the differences between a three-legged and
so it is like a conductor (reluctance is the magnetic four-legged core. Symmetrical components let us break down
equivalent of resistance in an electric circuit). a set of three-phase currents into the following components:
• Air has high reluctance, ℜ , so it is like an open • Positive sequence
circuit. • Negative sequence
Fig. 2 is an electrical analogy to the magnetic circuit shown • Zero sequence
in Fig. 1. n P • i P = ATP is analogous to a voltage source in The positive- and negative-sequence components of the
current are balanced and sum to zero. The AT and resulting
Fig. 2. Similarly, n S • iS = ATS is also analogous to a voltage flux from these components of the current flowing in the coils
source in Fig. 2. The right core leg in Fig. 1 is a low- of the transformer also sum to zero, so there is no need for a
reluctance path that shorts the top and bottom of the magnetic
 3

return path for flux. Thus, there is no difference between a D. AT Equations for Three-Legged Core Transformers
three-legged and four-legged core for the positive- and It is desirable to be able to ignore the core construction
negative-sequence components of the phase currents. when designing transformer protection. To do this, we can
On the other hand, zero-sequence currents are in phase with sum the ATs around the three magnetic circuit loops created
each other and equal in magnitude. Recall that 1xI0 flows in by the three core legs. Two sets of equations are possible for
each phase per symmetrical component theory. Because the the transformer shown in Fig. 3. Creating the loops around a
zero-sequence currents do not sum to zero, they require a pair of core legs in a clockwise direction yields:
return path to flow. This zero-sequence return path must exist
for the currents in the electrical circuit and for the fluxes in the ( ATP1 + ATS1 ) – ( ATP2 + ATS2 ) = 0 = AT1 – AT2
magnetic circuit, or the zero-sequence current cannot flow. As ( ATP2 + ATS2 ) – ( ATP3 + ATS3 ) = 0 = AT2 – AT3 (4)
we might expect, the difference between a three-legged and
four-legged core lies in what happens during zero-sequence ( ATP3 + ATS3 ) – ( ATP1 + ATS1 ) = 0 = AT3 – AT1
unbalances on the power system. Creating the loops around a pair of core legs in a
If the zero-sequence current in all windings provides counterclockwise direction yields:
balancing ATs (i.e., if there is equal 1xI0 AT in each of the
three primary coils versus each of the three secondary coils), ( ATP1 + ATS1 ) – ( ATP3 + ATS3 ) = 0 = AT1 – AT3
the fluxes sum to zero on each core leg, and there is no need ( ATP2 + ATS2 ) – ( ATP1 + ATS1 ) = 0 = AT2 – AT1 (5)
for the fourth core leg. For example, in many cases, a delta
winding is included. A delta winding appears as an electrical
( ATP3 + ATS3 ) – ( ATP2 + ATS2 ) = 0 = AT3 – AT2
short circuit to zero-sequence voltage on the associated wye- In any case, ATs on any pair of core legs are equal at all
connected winding. Thus, for any zero-sequence current times for three- and four-legged cores, as well as for
flowing in the wye-connected windings, there is an easy path transformers built from single-phase units.
for compensating ATs to flow in the delta winding. In other
words, a circulating current in the delta winding provides the
zero-sequence flux that balances the flux along each core leg
to zero. In this case, the AT on each core leg cancels by
nature, and there is little flux that needs a return path.
But, in some cases, there will be zero-sequence ATs
developed in one three-phase set of coils that are not
completely compensated by zero-sequence current in the other
three-phase set of coils. The flux that is driven by these ATs,
per (2), closes through the high-reluctance path outside the
magnetic core (i.e., the insulating oil and transformer tank).
Going back to (1), a coil with a high-reluctance path, such as
air, results in relatively low levels of flux. Low levels of flux
in the coil result in relatively low electrical impedance for that Fig. 3. ATs on a three-legged core.
coil. This low electrical impedance is only associated with
zero-sequence unbalance. The phenomenon of zero-sequence III. SHORT CIRCUITS IN TRANSFORMERS
current flowing in wye-connected windings of a three-legged Now that we have refreshed our knowledge on transformer
core transformer without a compensating delta winding is theory, let us consider various faults a transformer may
commonly referred to as a phantom delta, phantom tertiary, or experience. As we can see from Fig. 4, transformer faults can
tank delta. be divided into three basic categories—winding-to-ground
To summarize, a three-legged core transformer with wye faults (1, 4, 8), winding-to-winding faults (2, 5), and turn-to-
and delta windings is a low-impedance path for zero-sequence turn faults on the same winding (3, 6, 7). Examining each of
current because of the electrical short circuit of the delta these categories separately, we will discuss the magnitude of
winding. A three-legged core transformer without a delta the fault current, the method for detecting the fault, and the
winding becomes a low-impedance path for zero-sequence sensitivity of each method.
current flow because of the low magnetizing impedance of the
oil/air path outside the transformer core.
This seems counterintuitive, because oil and air have low
relative permeability. However, it is this low permeability of
oil and air that results in a high-reluctance path. A high-
reluctance path yields a low flux value from the zero-sequence
current flow. A low flux value means a low opposing voltage
is induced and therefore a low-inductance value. From a zero-
sequence current point of view, the transformer windings have
a low impedance. Fig. 4. Internal transformer faults.
 4

A. Characteristics of Partial Winding Faults differential current is small, as per (6), but the current in the
With reference to Fig. 5, consider a turn-to-turn fault (S1 faulted loop is large. If not isolated in a timely manner, this
closed, S2 opened) or a winding-to-ground fault close to the current will create considerable damage.
grounded neutral (S1 opened, S2 closed). The H winding has a As seen in Fig. 5, the current in the faulted loop circulates
short circuit on the leg labeled 1; the X winding represents the (returns) via the transformer neutral, and if we measure this
healthy winding(s), k is the fractional number of shorted turns loop current, we could reliably detect this type of fault. One
(few percent), and iF is the fault current. No assumption is method uses the neutral current in conjunction with zero-
made regarding the winding connection or core type. sequence current calculated from phase currents to create a
zero-sequence differential element. This is known as the REF
method and is explained in Section VII.
C. Turn-to-Turn Faults
Turn-to-turn faults also produce a relatively small change
in differential current, while fault current in the loop is
relatively large and can cause significant damage if not
detected and isolated rapidly (see Section III, Subsection A).
These faults are detectable via a negative-sequence
differential function, as described in Section VI.
Fig. 5. A general model of a partial winding fault.
D. Winding-to-Winding Faults
The difference in ATs on the affected core leg before and Winding-to-winding faults produce a large change in the
during the fault amounts to the ATs produced by the fault phase currents of windings that are involved in the fault. This
current (iF) flowing through the number of shortened turns, means protection elements using phase currents as operating
(k • nH). As explained in Section II, the ATs are equal between quantities can readily detect these faults. Typically, phase
all three legs, meaning the fault current produces equal ATs differential elements are used to detect and isolate these types
across the other two legs of the core. of faults.
As explained later in Section IV, transformer differential
protection monitors ATs of an unfaulted transformer. IV. TRANSFORMER DIFFERENTIAL PROTECTION
Therefore, the differential current reflects the unmonitored
This section makes a connection between the well-known
extra ATs produced by a partial winding fault, which is:
current compensation rules for transformer differential
i DIF ~ k • n H • i F (6) protection and AT balance equations of the protected
However, the exact level of the fault current (and thus the transformer. It also explains why a differential function that
level of the differential current) is a complex function follows the practical current compensation rules ensures
depending on the resistance and inductance of the shorted detection of internal transformer short circuits.
portion of the winding, distribution of fluxes due to the Going back to first principles and balancing the ATs allow
winding location with respect to other windings on the for the development of current compensation rules for
magnetic core, and possible local saturation of the core. These nonstandard transformer connections, as illustrated in
factors are beyond the scope of this paper. Section V.
Consider the wye-delta power transformer in Fig. 6.
B. Winding-to-Ground Faults
Magnetic
Winding-to-ground faults can be subdivided into two 4 Circuit
categories—namely faults farther away from the neutral point iHA iX3 iXA
* 1
(faults that have a greater driving voltage) and faults close to *
iX1
the neutral point. iHB iXB
2
Winding-to-ground faults farther from the neutral point iHC
iX2
iXC
3
have lower fault current in the faulted loop but result in
nH nX
greater change in phase currents on the windings that are on
the same core as the faulted winding. The large change in VXC
VHC
phase current is due to the increase in the voltage driving the
11*30°
fault current. The end result is that protection elements using
VHA VHA
phase currents as operating quantities, such as phase
VXB VXA
differential, can readily detect this type of fault and speedily
VHB
isolate the transformer, thereby limiting damage.
Winding-to-ground faults close to the transformer neutral Fig. 6. Sample wye-delta transformer.
(see Section III, Subsection A) are associated with a low
driving voltage. The result of this is that the increase in
 5

Neglecting the magnetizing branch of the core, we can Equations in (11) hold true as long as the transformer
write the following AT equations for the three legs of the core: conforms to the assumed model in Fig. 6. Departures from this
AT1 = n H • i HA + n X • i X1 model, including turn-to-turn faults, interwinding faults,
ground faults, or tap changer operation altering the turn ratio,
AT2 = n H • i HB + n X • i X2 (7)
will cause an unbalance. Therefore, expressions in (11) are
AT3 = n H • i HC + n X • i X3 used as operating signals of a transformer differential
Kirchhoff’s current law allows writing the current balance protection element:
equations for the nodes of the delta winding: VH 1
i XA – i X1 + i X3 = 0
i DIF(1) = i XA + • ( i HA – iHC )
VX 3
i XB – i X2 + i X1 = 0 (8)
VH 1
i XC – i X3 + i X2 = 0 i DIF( 2 ) = i XB + • ( i HB – i HA ) (12)
VX 3
Note that we cannot solve for the delta-winding currents,
VH 1
because any common component in these currents cancels and i DIF( 3) = i XC + • ( iHC – i HB )
cannot be measured at the transformer bushings. Instead, we VX 3
eliminate the delta-winding currents in the equations from (8) where: | | denotes magnitude estimation (filtering).
using the AT equations in (7). This yields the following: The restraining signal matches the differential signal but
n X • i XA + n H • ( i HA – i HC ) – ( AT1 – AT3 ) = 0 responds to the sum of magnitudes rather than the sum of
vectors:
n X • i XB + n H • ( i HB – i HA ) – ( AT2 – AT1 ) = 0 (9)
VH 1
n X • i XC + n H • ( i HC – i HB ) – ( AT3 – AT2 ) = 0 i RST (1) = i XA + • i HA – i HC
VX 3
As explained in Section II, ATs on any leg pair are equal V 1
regardless of the core type. By pairing core legs, we can write i RST ( 2 ) = i XB + H• i HB – i HA (13)
VX 3
AT balance equations around the resulting magnetic circuit
loops, similar to the way we would around an electrical circuit VH 1
i RST (3) = i XC + • i HC – i HB
loop using Kirchhoff’s voltage law. VX 3
Given the equations from (5), the balance equations in (9) The equations in (12) and (13) have been used for decades
for the transformer in Fig. 6 become: for the protection of wye-delta transformers. Protective relays
n X • i XA + n H • ( i HA – i HC ) = 0 are connected to CTs, which can be connected either in wye or
delta. Using phase-to-phase currents (delta-connected CTs) in
n X • i XB + n H • ( i HB – i HA ) = 0 (10)
the wye winding and phase currents (wye-connected CTs) in
n X • i XC + n H • ( i HC – i HB ) = 0 the delta winding is commonly referred to as vector
compensation, a phase shift, or delta compensation.
nH VH
Note that = and therefore: Furthermore, using the phase-to-phase current (delta-
nX 3 • VX connected CTs) in the wye winding also removes the zero-
sequence current component and is therefore commonly
VH 1 referred to as zero-sequence removal. The application of the
i XA + • ( iHA – i HC ) = 0 voltage ratio is referred to as ratio matching or tap
VX 3
V 1 compensation.
i XB + H • ( iHB – iHA ) = 0 (11) Considering the vector diagram in Fig. 6, these rules of
VX 3 transformer current compensation yield the diagram in Fig. 7
VH 1 for applications with electromechanical relays. In applications
i XC + • ( iHC – iHB ) = 0
VX 3 with microprocessor-based relays, these rules lead to selection
of the following compensation matrices:
⎡ 1 0 –1⎤ ⎡1 0 0 ⎤
1
TH = • ⎢ –1 1 0 ⎥ , TX = ⎢⎢0 1 0 ⎥⎥
⎢ ⎥ (14)
3
⎢⎣ 0 –1 1 ⎥⎦ ⎢⎣0 0 1 ⎥⎦
 6

iHA iHB iHC

* *

1 2 3

nS
iXA
R R iXB
iXC
R R
R R
nC

O O O

Fig. 8. An autotransformer connection.

Fig. 7. Application with external CT compensation. Equations from (16) explain the benefits of
Regardless of the relay technology (electromechanical, autotransformers compared with two-winding transformers—
static, microprocessor-based), the rules of phase shifting, zero- the transformation ratio is increased by the common turns
sequence removal, and ratio matching have roots in matching without adding extra turns in the series winding. This reduces
ATs between pairs of the magnetic core legs. As such, the both the copper and iron requirements and yields a more
resulting differential currents signify the physical balance in economical design with reduced losses, size, and weight.
the protected transformer, and when out of balance, they point Balancing the ATs per (5) leads to the following
to an internal short circuit. This creates a solid foundation of differential protection equations:
dependability of transformer differential protection beyond the VH
security-oriented view of balancing the transformer currents i DIF(1) = ( i XA – i XC ) + • ( i HA – i HC )
VX
for different phase shifts and ratios.
Development of the balance equations neglects the VH
magnetizing current. The magnetizing current during inrush i DIF( 2 ) = ( i XB – i XA ) + • ( i HB – i HA ) (17)
VX
and overexcitation demonstrates itself as the differential signal
and requires a separate treatment, typically in the form of VH
i DIF( 3) = ( i XC – i XB ) + • ( i HC – i HB )
harmonic blocking or restraining. VX

V. TRANSFORMER DIFFERENTIAL PROTECTION BEYOND THE VH ⎛ n ⎞

where: = ⎜1 + S ⎟ .
COMMON WYE OR DELTA CONNECTIONS VX ⎝ n C ⎠
This section considers transformer connections beyond the
standard wye or delta windings and progresses from Note that the differential protection equations for
autotransformers, through power zig-zag, Scott-T, and autotransformers follow the same current compensation rules
Le-Blanc transformers, to extended delta phase shifters in as a wye-wye transformer. The appropriate compensation
order to illustrate the derivation of proper transformer matrices in a microprocessor-based relay are:
differential protection equations. ⎡ 1 0 –1⎤
1
A. Autotransformers TH = TX = • ⎢ –1 1 0 ⎥⎥ (18)
3 ⎢
Consider the autotransformer in Fig. 8 with the series (S) ⎣⎢ 0 –1 1 ⎥⎦
and common (C) windings. The ATs on each of the core legs Consider an autotransformer built with three single-phase
equal: banks or on a four-legged core. Such an autotransformer can
AT1 = n S • i HA + n C • ( i HA + i XA ) be protected on a per-phase basis, taking advantage of the
observation that ATs in each single-phase core equal zero, per
AT2 = n S • i HB + n C • ( i HB + i XB ) (15)
(3):
AT3 = n S • i HC + n C • ( i HC + i XC )
VH
or i DIF(1) = i XA + • i HA
VX
AT1 = ( n S + n C ) • i HA + n C • i XA VH
i DIF( 2 ) = i XB + • i HB (19)
AT2 = ( n S + n C ) • i HB + n C • i XB (16) VX
AT3 = ( n S + n C ) • i HC + n C • i XC VH
i DIF( 3) = i XC + • i HC
VX
 7

The appropriate compensation matrices in a The appropriate compensation matrices in a

microprocessor-based relay are: microprocessor-based relay are:
⎡1 0 0⎤ ⎡ 1 0 –1⎤ ⎡1 0 0 ⎤
1
TH = TX = ⎢⎢ 0 1 0⎥⎥ (20) TH = TX = • –1 1 0 ⎥ , TY = ⎢⎢0 1 0 ⎥⎥
⎢ ⎥ (23)
3 ⎢
⎢⎣ 0 0 1 ⎥⎦ ⎢⎣ 0 –1 1 ⎥⎦ ⎢⎣0 0 1 ⎥⎦
In some applications, CTs are available at the neutral side If the tertiary delta winding is not brought out (buried), the
of the low-voltage winding. In such cases, we may apply a equations in (22) simplify to (17) as if the transformer were a
bus-like protection zone between the H, X, and neutral CTs two-winding transformer. The buried delta winding is still
because of the metallic connection between the windings. protected to a degree, because it takes part in the magnetic
These zones monitor the electric circuit only, and therefore circuit of the transformer. The case of a buried delta is
they are blind to turn-to-turn faults. In order to detect turn equivalent to the case of a power delta winding with the
faults, the differential function must effectively monitor both breaker opened. However, the degree of protection is limited
the electric and magnetic circuits of the transformer. by the fact that the power rating of the tertiary delta in
Next, consider an autotransformer with a tertiary delta autotransformers is often much less than the full rating of the
winding, as shown in Fig. 9. transformer. This results in a small fault current when
reflected from the delta winding to the high- and low-voltage
windings of the autotransformer.
Also, ground faults on an ungrounded, buried delta
winding cannot be detected based on current signals.
*

Let us go back to a single-phase or a four-legged core

design in the context of a buried delta winding. Quite often,
the winding currents in the delta winding are measured to
polarize directional line protective relays [iY1, iY2, and iY3 in
(21)]. With the inside delta currents available, the
*

autotransformer can be protected as a three-winding

transformer, balancing the phase currents per (21), (20), and
(19):
VH V
i DIF(1) = i XA + • i HA + 3 • H • i Y1
VX VY
VH V
i DIF( 2 ) = i XB + • i HB + 3 • H • i Y2 (24)
VX VY
*

VH V
i DIF( 3) = i XC + • i HC + 3 • H • i Y3
VX VY

Fig. 9. An autotransformer connection with a tertiary delta winding. B. Power Zig-Zag Transformers
The electromagnetic balance for this transformer is: Consider the wye/zig-zag connection in Fig. 10.
AT1 = ( nS + n C ) • i HA + n C • i XA + n Y • i Y1 nH 0.5nX 0.5nX
iHA iXA
1
AT2 = ( n S + n C ) • i HB + n C • i XB + n Y • i Y2
* * *
(21)
iHB iXB
2
AT3 = ( n S + n C ) • i HC + n C • i XC + n Y • i Y3
iHC iXC
3
Following the example of a wye-delta transformer in
Section IV, we eliminate the winding currents and obtain the
following equations for the differential function: VHC VXC

11*30°
VH V
i DIF(1) = ( i XA – i XC ) + • ( i HA – i HC ) + 3 Y • i YA VHA VHA
VX VX
VXB VXA
V V VHB
i DIF( 2 ) = ( i XB – i XA ) + H • ( i HB – i HA ) + 3 Y • i YB (22)
VX VX Fig. 10. Wye/zig-zag power transformer connection.
VH V
i DIF( 3) = ( i XC – i XB ) + • ( i HC – i HB ) + 3 Y • i YC
VX VX
 8

The ATs on each of the core legs equal: iHA iHB iHC VC
n iXQ
AT1 = n H • i HA + X • ( i XA – i XB ) * *
VA
2 3
nH nX 90°
n
AT2 = n H • i HB + X • ( i XB – i XC )
2
(25) VB VD
2 0.5nH 0.5nH
VQ
nX
AT3 = n H • i HC + • ( i XC – i XA ) * *
2 *
nX
iXD
Comparing the ATs on core leg pair 1 and 3 per (5) yields:
nX
AT1 – AT3 = n H • ( i HA – i HC ) + • ( 2 • i XA – i XB – i XC ) (26) Fig. 11. A Scott-T transformer connection.
2
The transformer is made of two independent cores. We
VH VX write AT balance equations for each core separately:
Considering that n H ~ and n X ~ 2 • , we obtain:
nH
3 3 ( i HA – iHC ) + n X • iXD = 0 (32)
2
1 VX 1
AT1 – AT3 = ( i HA – iHC ) + • • ( 2 • i XA – i XB – i XC ) (27)
3 • nH
3 VH 3 • i HB + n X • i XQ = 0 (33)
2
which yields the following differential signal:
By replacing the turn counts with nominal voltages and
1 VX 1 introducing the differential currents, we obtain:
i DIF(1) = ( iHA – i HC ) + • • ( 2 • i XA – i XB – i XC ) (28)
3 VH 3
VH
i DIF(1) = • ( i HA – i HC ) + i XD (34)
For the other two relay elements, we use (28) and rotate the 2 • VX
subscript indices:
3 • VH
V 1 i DIF( 2 ) = • i HB + i XQ (35)
1
i DIF( 2 ) = ( iHB – i HA ) + X • • ( 2 • i XB – i XC – i XA ) (29) 2 • VX
3 VH 3
where:
1 VX 1 VH = the nominal voltage in the three-phase system
i DIF( 3) = ( iHC – iHB ) + • • ( 2 • i XC – i XA – i XB ) (30)
(phase to phase).
3 VH 3
VX = the nominal winding voltage in the two-phase
The appropriate compensation matrices in a system (phase to ground).
microprocessor-based relay are: The two differential currents can be created in a standard
four-winding transformer relay using wye compensation
⎡ 1 0 –1⎤ ⎡ 2 –1 –1⎤
1 1 ⎢ matrices applied to the following four currents (windings):
TH = • ⎢ –1 1 0 ⎥ , TX = • ⎢ –1 2 –1⎥⎥
⎢ ⎥ (31)
3 3 ⎡i D ⎤ ⎡i A ⎤ ⎡ –iC ⎤ ⎡0⎤
⎢⎣ 0 –1 1 ⎥⎦ ⎢⎣ –1 –1 2 ⎥⎦ ⎢i ⎥ , ⎢ 0 ⎥, ⎢ 0 ⎥, ⎢i ⎥
⎢ Q⎥ ⎢ ⎥ ⎢ ⎥ ⎢ B⎥ (36)
C. Summary of Compensation Matrices for Wye, Delta, and ⎢⎣ 0 ⎥⎦ ⎢⎣ 0 ⎥⎦ ⎢⎣ 0 ⎥⎦ ⎢⎣ 0 ⎥⎦
Zig-Zag Windings
The above currents are created by appropriately wiring the
The TX matrix in (31) is often referred to as a double-delta
five currents of the Scott-T transformer to four current
connection, because it is equivalent to connecting the main
terminals of the relay. Three tap compensation factors are
CTs in delta and their secondary circuits in delta again when
required for the first (Tap 1), fourth (Tap 2), and second and
applying external compensation for electromechanical relays.
third (Tap 3) windings in (36), as per the scaling factors in
The TH matrix in (31) is referred to as a single-delta
(34) and (35).
connection. The TY matrix in (23) is referred to as a wye
Alternatively, we can create the A-phase minus C-phase
connection. In general, transformer delta windings require a
currents by connecting secondaries of the main CTs and using
wye compensation matrix, wye windings require a single-delta
the following three windings in a standard transformer relay:
compensation matrix, and zig-zag windings require a double-
delta compensation matrix. ⎡i D ⎤ ⎡i A – i C ⎤ ⎡0⎤
⎢i ⎥ , ⎢ 0 ⎥, ⎢i ⎥ (37)
D. Scott-T Transformers ⎢ Q⎥ ⎢ ⎥ ⎢ B⎥
The Scott-T transformers shown in Fig. 11 historically ⎣⎢ 0 ⎦⎥ ⎣⎢ 0 ⎦⎥ ⎣⎢ 0 ⎦⎥
have been used to connect two- and three-phase systems. They
are still used in applications with two-phase motors, such as in
mining or railway systems and, recently, in power electronics.
 9

E. Le-Blanc Transformers F. Extended Delta Phase-Shifting Transformers

Le-Blanc transformers also serve the purpose of feeding Consider the extended delta phase-shifting transformer
two-phase loads from three-phase systems but are built using a (PST) in Fig. 13 with the series (S) and exciting (E) windings.
three-legged core, as shown in Fig. 12. The AT equations are:
nS n
AT1 = D • • iSA – D • S • i LA + n E • i E1
2 2
nS nS
AT2 = D • • iSB – D • • i LB + n E • i E2 (43)
*

2 2
nS nS
AT3 = D • • iSC – D • • i LC + n E • i E3
2 2
In (43), the source current works with half of the series
windings multiplied by the tap changer position; the load
*

( )
3 − 1 • nX ( )
3 − 1 • nX current works with half of the series windings multiplied by
the tap position but in the opposite direction; and the
excitation current works with the number of excitation
winding turns. The source, load, and excitation currents are
*

tied together with Kirchhoff’s current equations as follows:

iSA + i LA = i E3 – i E2
iSB + i LB = i E1 – i E3 (44)
Fig. 12. Le-Blanc transformer connection. iSC + i LC = i E2 – i E1
The AT balance equations are:
AT1 = n H • i H1 – ( )
3 –1 • n X • i XQ + n X • i XD
AT2 = n H • i H2 + n X • i XQ (38)
AT3 = n H • i H3 – ( )
3 –1 • n X • i XQ – n X • i XD

Comparing the ATs in pairs of core legs per (5) allows

eliminating the winding currents in the delta winding based on

*
(8) and writing the differential protection equations. We
further use nominal voltages defined for the Scott-T
transformer above to eliminate the turn counts:
*

2 • VX
i DIF(1) = i HA + i XD (39)
3 • VH

i DIF( 2 ) = i HB +
VX
3 • VH
( 3i XQ – i XD ) (40) Fig. 13. Extended delta phase-shifting transformer (where S is source side,
L is load side, D = 0 is neutral control, D = +1 is full advance control, and
D = –1 is full retard control).

i DIF( 3) = i HC –
VX
3 • VH
( 3i XQ + i XD ) (41) Comparing the ATs on each pair of core legs in (5) and
substituting the excitation currents with the source and load
currents per (44) yields:
In order to simplify implementation in a standard
transformer relay, we can consider combining the second and D
third differential terms by subtracting (40) and (41) as follows:
i DIF(1) = n S
2
( ( iSA – iLA ) – ( iSC – iLC ) ) + n E ( iSB + iLB )
D
i DIF( 23) = i HB – i HC +
2 • VX
VH
i XQ (42) i DIF( 2 ) = n S
2
(( iSB – iLB ) – (iSA – iLA ) ) + n E ( iSC + iLC ) (45)

D
A multiwinding standard transformer differential relay with i DIF( 3) = n S
2
( ( iSC – iLC ) – ( iSB – iLB ) ) + n E ( iSA + iLA )
wye compensation matrices can be used to implement (39)
through (41) or (39) and (42), as similarly explained for the As expected, the transformer differential protection
Scott-T transformer above. equations are dependent on the position of the tap changer, D.
Note that the protection method in (45) does not require
measuring the excitation currents.
 10

A standard multiwinding transformer relay can be used to of the standard phase differential element, limiting its ability
implement equations from (45). First, let us rewrite (45) to a to detect these types of faults. As explained in Section IV and
more convenient form: commonly used in microprocessor-based relays, the phase
differential function applies matrix compensation to
D
i DIF(1) = n S
2
( ( iSA – iSC ) + ( iLC – iLA ) ) + n E ( iSB + iLB ) transformer terminal currents:
⎡IA ⎤ ⎡ IA ⎤
D
i DIF( 2 ) = n S
2
(( iSB – iSA ) + (iLA – iLB ) ) + n E ( iSC + iLC ) (46) ⎢ ⎥ ⎢ ⎥
⎢ IB ⎥ = T • ⎢ IB ⎥ (51)
⎢⎣ IC ⎥⎦ ⎢⎣ IC ⎥⎦
D
( ( iSC – iSB ) + ( iLB – iLC ) ) + n E ( iSA + iLA )
C
i DIF( 3) = n S
2 We can use (51) to examine how the sequence components
(zero = 0, positive = 1, negative = Q) in the terminal currents
The six currents can be wired to four relay current are converted into sequence components of the compensated
terminals (windings) as follows: currents.
⎡iSA ⎤ ⎡ –i LA ⎤ ⎡ iSB ⎤ ⎡ i LB ⎤ ⎡ I0 ⎤ ⎡ I0 ⎤ ⎡1 1 1⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢i ⎥ ⎢ ⎥ –1 ⎢ ⎥ 1⎢ ⎥
⎢ iSB ⎥ , ⎢ –i LB ⎥ , ⎢ iSC ⎥ , ⎢ LC ⎥ (47) ⎢ I1 ⎥ = A • T • A • ⎢ I1 ⎥ , A = 3 ⎢1 a a2 ⎥ (52)
⎢⎣ iSC ⎥⎦ ⎢⎣ –i LC ⎥⎦ ⎢⎣iSA ⎥⎦ ⎣⎢i LA ⎦⎥ ⎢ IQ ⎥ ⎢ IQ ⎥ ⎢ 2
a ⎦⎥
⎣ ⎦C ⎣ ⎦ ⎣1 a
The first two windings use the following compensation For example, the transformation matrix TH in (14), when
matrix: expressed for symmetrical components, is:
⎡ 1 0 –1⎤ ⎡0 0 0 ⎤
1
• ⎢ –1 1 0 ⎥⎥ (48) ⎢ ⎥
3 ⎢
– j300
⎢0 e 0 ⎥ (53)
⎢⎣ 0 –1 1 ⎥⎦ ⎢ 0 ⎥

⎣⎢0 0 e j30 ⎦⎥
The latter two windings use the following matrix:
Generally, the compensation matrix for a symmetrical
⎡1 0 0 ⎤ component of any three-phase transformer is:
⎢0 1 0 ⎥ (49)
⎢ ⎥ ⎡0 /1 0 0 ⎤
⎢⎣0 0 1 ⎥⎦ ⎢ jδ ⎥
⎢ 0 e 0 ⎥ (54)
Alternatively, external CT summation can be arranged, and ⎢ 0
⎣ 0 e – jδ ⎥⎦
a two-winding transformer relay can be used with the
following currents: and can be interpreted as follows:
⎡iSA – i LA ⎤ ⎡ iSB + i LB ⎤ • δ is the nameplate phase shift of the transformer.
⎢ ⎥ ⎢ ⎥ • The positive sequence in the terminal current shifted
⎢ iSB – i LB ⎥ , ⎢ iSC + i LC ⎥ (50)
by the vector group angle transforms into the positive
⎢⎣ iSC – i LC ⎥⎦ ⎢⎣iSA + i LA ⎥⎦
sequence in the compensated current ( e jδ ).
and matrices (48) and (49), respectively. • The negative sequence in the terminal current shifted
The challenge is to implement a variable tap coefficient by the minus vector group angle transforms into the
dependent on D for the first two windings in (47) or the first negative sequence in the compensated current ( e – jδ ).
winding in (50). This variability can be accommodated with a
• The zero sequence is normally removed (value 0),
coarse granularity via multiple settings groups. We need to
except when the unity transformation matrix (49) is
remember that D can be negative (retard versus advance
used as applicable to delta windings (value 1).
control).
• The three sequence components are decoupled [the
off-diagonal elements in (54) are zeros].
VI. NEGATIVE-SEQUENCE DIFFERENTIAL
The phase-restraining signal is a sum of the magnitudes of
As described in Section III, when a turn-to-turn fault the compensated phase currents. Considering (54), this means
occurs, the magnitudes of the phase currents of the associated that both the positive- and negative-sequence components in
winding do not change significantly. However, as a result of the terminal currents produce restraint for the phase
the fault, the symmetry of the phase currents on both the differential function.
primary and secondary sides of the transformer (or any other For a balanced system, the differential current is
piece of equipment) is disturbed. The change in symmetry approximately zero, and the restraining signal is proportional
during this type of fault favors using negative-sequence to the positive-sequence current (load). This means that the
currents to detect it. sensitivity of the element is directly impacted by the load
But before we examine the negative-sequence differential current.
element, let us understand how the symmetrical components
in the terminal currents transform into the differential currents
 11

With this in mind, let us return to the turn-to-turn fault. We In summary, a negative-sequence differential element can
know that this type of fault creates a disturbance of the be applied to any three-phase transformer as follows:
symmetry of both primary- and secondary-side phase currents, • The differential current is a vector sum of ratio-
which manifests itself as a negative-sequence current. If the matched, negative-sequence terminal currents.
transformer is lightly loaded (i.e., the restraint current is • Prior to adding to the negative-sequence differential
small), then this element has good sensitivity and may detect current, the terminal negative-sequence currents are
turn-to-turn faults. However, as the load increases, the phase-shifted by the negative vector group angle.
sensitivity of the phase differential element is reduced, and the • The restraining signal is a sum of the ratio-matched
ability to detect turn-to-turn faults is diminished. magnitudes of the terminal negative-sequence currents
If we use pure negative-sequence current to build a (other combinations are also possible, such as the
differential element, we have the advantage of making the “maximum of” approach).
element highly sensitive and relatively independent from load. • In the case of PSTs, angle shift is variable, but the
In this respect, note that the compensated currents are operation of shifting the negative sequence is still a
summed to form the phase differential current. Considering simple multiplication by a complex number (but
(54), this means that the sequence components in the variable depending on the tap changer position).
differential currents are sums of appropriately shifted and The above “recipe” for the transformer negative-sequence
ratio-matched sequence components in the terminal currents. differential element may appear heuristic, but as demonstrated
For example, in the case of the wye-delta transformer in in this section, it is well-founded in the physics of a three-
Section IV, the negative-sequence component in the phase transformer. It is based on exactly the same principle of
differential current is: balancing ATs in the core of the transformer. The gain in
V 0
sensitivity is possible because the matching through-fault
I DIF( Q ) = IX ( Q ) + H • e j30 • IH ( Q ) (55) negative-sequence current is used as a restraint.
V X
Equation (55) can be implemented as a negative-sequence
Note that (55) is nothing more than a component of the transformation of the phase differential currents, a sum of the
phase differential signals in (12). As such, it does not by itself negative-sequence components in the ratio-matched and
bring any significant gain in sensitivity. compensated terminal currents [1], or a ratio-matched and
Sensitivity is enhanced because we eliminate the impact of phase-shifted sum of the negative-sequence components in the
load current (positive sequence) and restrain (55) only with terminal currents [2]. Regardless of the method, the outcome
the negative-sequence current: is mathematically identical as long as the phasors are extracted
V using linear filters.
I RST ( Q ) = IX ( Q ) + H • I H ( Q ) (56) As is often the case, this advantage is also a weakness of
VX the element in that false negative-sequence currents are also
A negative-sequence differential element defined by (55) generated during through faults with CT saturation. However,
and (56) is considerably more sensitive compared with the external fault detection guards this sensitive protection
phase differential element, not because the operating signal function very well, allowing for secure, fast, and sensitive
(55) is higher but because the restraining signal (56) is load- operation.
independent and therefore considerably lower during turn-to-
turn faults. VII. REF PROTECTION
The wye-delta transformer example used above can be A. Common REF Schemes
extended to any three-phase transformer, including
REF protection is intended to detect partial winding faults
autotransformers, multiwinding transformers, and phase
to ground near the neutral terminal of a grounded wye or zig-
shifters. The extension is possible because of the way the
zag connected winding, or of an autotransformer. For these
sequence components in the terminal currents blend into the
faults, the phase currents measured at the terminals of the
sequence components in the phase differential currents from
transformer can be quite low, while the current in the shorted
(54).
turns can be very high, quickly damaging the transformer (see
To demonstrate, consider the PST in Section V with the
Section III, Subsection A). REF schemes are also
differential currents defined by (46). The negative-sequence
recommended for impedance-grounded systems where the
differential current is:
ground fault current is limited to a much lower level than the
I DIF( Q ) = IS( Q ) + e ( ) • I L( Q )
jΘ D
(57) phase fault currents. An REF scheme can provide much higher
sensitivity than the phase differential elements to detect these
where: low-level ground faults.
⎛ D nS ⎞ REF protection schemes take advantage of the fact that the
⎜ j– 3• 2 • n ⎟ current in the fault loop can be measured directly by the
Θ ( D ) = angle ⎜ E ⎟ (58) neutral terminal bushing CT. In a solidly grounded system, the
⎜ D nS ⎟ ground current can be quite high, even when the phase
⎜ j+ 3 • 2 • n ⎟
⎝ E ⎠ currents are small.
 12

Fig. 14 and Fig. 15 show the two main schemes used for include a genuine zero-sequence component (line-to-line and
REF protection, using the application example of an three-phase faults). Sensitive REF schemes must be designed
autotransformer [3][4]: to be secure for this condition. The high-impedance scheme in
• Current-polarized directional ground element. Fig. 15 is inherently immune to CT saturation by nature of its
• High-impedance REF differential element. operation principle.
The current-polarized directional element in Fig. 14
Simple directional element
requires additional logic for security. Security from false zero-
Currents all scaled to primary units
3I0 3I0
H1 H2 H3
sequence current is obtained by requiring that current be
present above a threshold in the neutral of the transformer.
32
H0X0
Because zero-sequence current at the neutral of the
transformer is measured by a single CT, it is not possible to
have a false zero-sequence current in this signal.
The current-polarized directional scheme also requires
additional logic for dependability. In a delta-wye transformer,
X1 X2 X3 the transformer can be energized from the delta winding with
High current for partial Single H0X0 CT eliminates possibility of false
winding fault to ground 3I0 in neutral terminal for external fault
the wye-winding terminal open. In this case, no zero-sequence
current can flow at the boundary of the zone, so a directional
Fig. 14. Current-polarized directional REF scheme.
decision cannot be made. In this case, a bypass logic path
allows tripping when there is zero-sequence current in the
neutral and none present in the terminals of the transformer.

VIII. DISCUSSION
Differential protection of transformers against internal
short circuits is accomplished using differential current
equations that effectively emulate the AT equations of the
transformer and, by doing so, monitor both the electric and
magnetic circuits of the transformers. Changes in the
Fig. 15. High-impedance REF scheme. monitored electromagnetic circuits, such as turn-to-turn faults,
ground faults, winding faults, and some open circuits, are
Both of these schemes work on the principle of Kirchhoff’s
therefore detected by the said protection equations within
current law, similar to a bus differential zone, and sum the
sensitivity limits of a given implementation.
zero-sequence currents in the transformer zone. Current must
For standard transformer connections, including wye, delta,
flow from a winding to ground in order for these schemes to
and zig-zag windings, as well as autotransformers, the said
detect the fault. They do not match ATs in the transformer
transformer protection equations can be created using the
cores and therefore cannot detect turn-to-turn faults. REF also
known protection engineering principles of vector group
cannot detect winding-to-winding (phase-to-phase) faults.
compensation, ratio matching, and zero-sequence removal.
It is common to apply a similar current summation (bus-
Various heuristic rules are available to quickly select the
like) differential zone with some uncommon transformers,
proper compensation rules. When in doubt, the AT equations
such as PSTs, either in addition to or instead of transformer
can be consulted to check or confirm the proper compensation
differential protection. For example, because of the difficulty
rules.
in incorporating factor D from (46) in the transformer
For nonstandard transformer connections, such as PSTs,
differential protection system, the PST in Fig. 13 may be
the AT balance allows developing proper transformer current
protected by six current summation zones around each
differential equations. The resulting equations are always
individual winding. To accomplish this, CTs would be
linear but different compared with what is required for the
required on each end of the delta excitation windings. Because
wye, delta, and zig-zag windings. Microprocessor-based
this scheme does not monitor the magnetic circuit, it will not
relays can be used to implement these equations with some
detect turn-to-turn faults. When using this differential
accuracy and sensitivity.
protection approach, it is necessary to rely on sudden pressure
Traditional transformer differential protection faces
relaying to detect turn-to-turn faults.
sensitivity limits for turn-to-turn faults and ground faults near
B. Security and Dependability Issues With REF Schemes the neutral terminal in a grounded winding. Theoretically,
The zero-sequence current at the terminals of the these faults upset the AT balance, but because of the
transformer is formed by the summation of three-phase autotransformer effect, the differential current unbalance can
currents and therefore can include false zero-sequence current be small if a small percentage of the coil is shorted. At the
when one of the phase CTs saturates. This is particularly same time, a given relay needs to maintain security in light of
relevant in dual-breaker transformer terminations when the possible CT errors or an on-load tap changer, typically by
maximum external fault current is not limited by transformer means of restraining, resulting in less sensitive protection.
impedance and when the external fault current does not
 13

REF provides sensitive protection for ground faults near X. REFERENCES

the neutral. This function monitors the electric circuit of the [1] M. Thompson, H. Miller, and J. Burger, “AEP Experience With
protected winding for currents leaking to ground, and Protection of Three Delta/Hex Phase Angle Regulating Transformers,”
therefore it responds to ground faults only. proceedings of the 33rd Annual Western Protective Relay Conference,
Spokane, WA, October 2006.
Negative-sequence differential protection increases
[2] A. Guzmán, N. Fischer, and C. Labuschagne, “Improvements in
sensitivity to turn-to-turn faults. This method is based on the Transformer Protection and Control,” proceedings of the 62nd Annual
AT balance, but instead of monitoring three balance equations Conference for Protective Relay Engineers, College Station, TX,
between the three-phase currents, it monitors a single March 2009.
composite current, known as the negative sequence. This is [3] H. Miller, J. Burger, and M. Thompson, “Automatic Reconfiguration of
Zones for Three-Phase and Spare Banks,” proceedings of the 36th
derived from the phase differential currents. Using the
Annual Western Protective Relay Conference, Spokane, WA,
composite signal does not increase the operating signal, but it October 2009.
redefines the restraining signal, allowing for more sensitive [4] N. Fischer, D. Haas, and D. Costello, “Analysis of an Autotransformer
operation. Similar to REF, this function is susceptible to CT Restricted Earth Fault Application,” proceedings of the 34th Annual
errors and therefore benefits from external fault detector Western Protective Relay Conference, Spokane, WA, October 2007.
algorithms or similar security measures often made available
in microprocessor-based relays. XI. BIOGRAPHIES
Nonstandard transformers invite applications with bus-like Bogdan Kasztenny is a principal systems engineer in the research and
development division of Schweitzer Engineering Laboratories, Inc. He has
differential zones. Like REF, these zones do not monitor the
20 years of experience in protection and control, including his ten-year
electromagnetic circuit of the transformer and therefore will academic career at Wroclaw University of Technology, Poland, Southern
not detect turn-to-turn faults and other faults that do not divert Illinois University, and Texas A&M University. He also has ten years of
the current outside of the monitored bus-like zones. industrial experience with General Electric, where he developed, promoted,
and supported many protection and control products.
Phase transformer differential protection supplemented
with REF and negative-sequence functions provides good Bogdan is an IEEE Fellow, Senior Fulbright Fellow, Canadian member of
sensitivity to transformer faults based on current CIGRE Study Committee B5, and an Adjunct Professor at the University of
measurements. Sudden pressure protection complements the Western Ontario. He has authored about 200 technical papers and holds
16 patents. He is active in the Power System Relaying Committee of the IEEE
electrical protection when larger transformers are used or and is a registered professional engineer in the province of Ontario.
when the electrical functions cannot provide good sensitivity.
Michael Thompson received his BS, magna cum laude, from Bradley
IX. CONCLUSIONS University in 1981 and an MBA from Eastern Illinois University in 1991. He
has broad experience in the field of power system operations and protection.
Differential protection of a transformer follows the AT Upon graduating, he served nearly 15 years at Central Illinois Public Service
balance of the unfaulted transformer. The AT balance brings (now AMEREN), where he worked in distribution and substation field
engineering before taking over responsibility for system protection
the currents from the galvanically isolated winding networks engineering. Prior to joining Schweitzer Engineering Laboratories, Inc. (SEL)
to a common equivalent circuit, allowing application of in 2001, he was involved in the development of several numerical protective
Kirchhoff’s current law to form a differential scheme. relays while working at Basler Electric. He is presently a principal engineer in
the engineering services division at SEL, a senior member of the IEEE, a
The classical rules of current compensation for transformer main committee member of the IEEE PES Power System Relaying
differential protection (vector group compensation, ratio Committee, and a registered professional engineer. Michael has published
matching, zero-sequence removal) replicate the AT balance of numerous technical papers and holds a number of patents associated with
power system protection and control.
the unfaulted transformer.
Following the rule of AT balance allows us to analyze
Normann Fischer received a Higher Diploma in Technology, with honors,
sensitivity and limitations of a given protection method, as from Witwatersrand Technikon, Johannesburg in 1988, a BSEE, with honors,
well as derive proper differential protection equations for from the University of Cape Town in 1993, and an MSEE from the University
nonstandard transformers (i.e., Scott-T, Le-Blanc, phase of Idaho in 2005. He joined Eskom as a protection technician in 1984 and was
a senior design engineer in the protection design department at Eskom for
shifters). three years. He then joined IST Energy as a senior design engineer in 1996. In
REF protection complements the traditional phase 1999, he joined Schweitzer Engineering Laboratories, Inc. as a power
differential scheme, providing good sensitivity to ground engineer in the research and development division. Normann was a registered
professional engineer in South Africa and a member of the South Africa
faults near the neutral point of grounded wye or zig-zag Institute of Electrical Engineers. He is currently a member of IEEE and
windings. ASEE.
Similarly, the negative-sequence differential element
supplements the phase differential by providing sensitive
protection to turn-to-turn faults, particularly during heavy load
conditions.
A combination of phase differential, REF, and negative-
sequence differential allows electrical (current-based)
protection of transformers without sensitivity gaps as
compared with sudden pressure relays. Previously presented at the 2010 Texas A&M
Conference for Protective Relay Engineers.
© 2010 IEEE – All rights reserved.
20100211 • TP6412-01