Practical Experience in Setting Transformer Differential Inrush Restraint

Rich Hunt Joe Schaefer Bob Bentert

GE Multilin Florida Power & Light Company Florida Power & Light Company

Abstract
The second harmonic inrush restraint function of transformer differential relays maintains
security of the differential protection during transformer inrush events. The typical setpoint for
the second harmonic restraint is the relay manufacturer’s default or recommended setting of 20%
of fundamental current, with some adjustment based on operating experience. However, some
operating situations may result in levels of second harmonic current lower than 20% during
inrush, and levels may be as low as 5%. This lower level of second harmonic current requires a
lower inrush restraint setting that may impact the tripping time of the differential element for
fault conditions. In addition, inrush restraint is typically performed on a per-phase basis, so a loss
of security is possible if inrush restraint performs incorrectly on only one phase of the protected
transformer.
This paper provides several examples of actual events where loss of security occurred due to
incorrect settings of the second harmonic restraint function, or due to mis-application of cross-
phase blocking. Based on the information from these events, the paper directly discusses
considerations and recommendations for setting the second harmonic restraint to maintain
security during transformer inrush including a discussion of traditional and adaptive inrush
restraint techniques. The paper also includes recommendations on when to apply cross-phase
blocking techniques such as 1-out-of-3 blocking, 2-out-of-3 blocking and Averaging blocking
methods.
A key message from this paper is the use of the actual inrush characteristic of the protected
transformer to determine optimum setpoints for the differential relay. Microprocessor relays have
the capability to, and should, capture waveforms every time a transformer is energized. This data
should be analyzed to check the adequacy of the existing second harmonic restraint settings, to
ensure no loss of security occurs.

1. Introduction
Florida Power & Light Company (FPL) is in the process of replacing existing transformer
protection panels at select locations on the FPL transmission system. The existing panels use
electro-mechanical or solid-state differential relays, and the new transformer protection panels
will use microprocessor relays with wye-connected CTs in an FPL standard protection package.
The basic protection for these transformers is differential protection where second harmonic
blocking is used to block the differential element from operating during transformer inrush
events.
As part of this replacement process, FPL used an experimental laboratory procedure along
with actual operating experiences to validate settings before placing the transformer in service.
A relay was configured to some inrush restraint function, level setting, and cross-phase blocking
method. These settings were then tested against simulated and actual fault events, by playing
oscillographic records back through a test set to the relay. These tests were repeated using

978-1-4244-1949-4/08/$25.00 ©2008 IEEE 118

Practical Experience in Setting Transformer Differential Inrush Restraint

different setpoints and different restraint functions until a setting that meets FPLs’ operating
criteria was determined. These settings are then confirmed by capturing transformer in-rush
events and comparing the setpoint levels to actual second harmonic levels to confirm there is
enough margin.
One of the drivers for this process is an effort to eliminate misoperations of transformer
differential relays due to low second harmonics on inrush without sacrificing protection
capabilities. Misoperations can occur during the energization of a transformer due to failure of
the harmonic restraint function. A digital fault recorder oscillographic record that captures this
type of harmonic restraint failure was used for the testing later in this paper. This fault record
shows a failure due to the low levels of harmonic current produced during an energization. An
external fault can also trigger a misoperation during the voltage recovery period, causing the
differential relay to operate immediately after a fault is cleared from the system.

2. Review of differential protection principles

Differential protection is a fast, selective method of protection against short circuits in
transformers, and is the standard protection used by FPL to protect transformers. Differential
protection is a practical application of Kirchoff’s current law. The sum of the currents entering
the transformer should equal the sum of the currents leaving the transformer. Differential
protection adds the measured currents entering and leaving the transformer to create a
differential current.

I1 I1 I1
i1 i1 i1
inrush

IDiff = i1+i2 IDiff = i1+i2 IDiff = i1+i2

I1=I2 I1>>I2
i2< i1 i1>> i2
saturated IDiff > 0 IDiff > 0
I2 I2 I2
i2 i2 i2

Differential principle Differential during CT saturation Differential during transformer inrush

Figure 1: Transformer differential protection principle

With the ideal transformer of Figure 1, and assuming ideal CTs, the differential current is
zero when current is flowing through the transformer. A differential current greater than zero
indicates an internal fault condition. In practice, the differential current for a normally operating
transformer is always greater than zero due to CT measurement error, the position of the load tap
changer, and other factors introducing noise into the measurement signals. Therefore, the
sensitivity of the protection is reduced slightly to account for these errors.

119
Practical Experience in Setting Transformer Differential Inrush Restraint

There are two common situations where differential protection may incorrectly declare an
internal fault condition. One condition is CT saturation for a fault outside of the transformer zone
of protection. The error in the measurement signal of the saturated CT results in a significant
error in the differential current. The erroneous differential current may result in undesired
operation of the differential element for an external fault condition. This type of event is beyond
the scope of this paper.
The second common situation is a transformer inrush event. Some operating situations
instantly change the operating flux of the transformer core, requiring a large supply of current.
This inrush of current typically occurs in only one winding of the transformer. Therefore inrush
currents may produce a differential current that results in the operation of the differential
protection. This type of event is not a fault condition, so the differential protection should
restrain from operating for this condition.

3. Transformer inrush phenomena

To properly set a protection function, it is necessary to have a basic understanding of the
power system events the function is intended to detect. To set the inrush restraint function for
transformer differential protection requires some understanding of transformer inrush events,
including the causes and characteristics of these events. This section of the paper defines a
transformer inrush event. The section continues on to discuss how power system conditions
influence the severity and characteristic of the inrush event, and finishes by describing the
common power system events that cause transformer inrush.

3.1. Definition of magnetizing inrush current

A transformer inrush event is actually magnetizing inrush current. The windings in a
transformer are linked magnetically by the flux in the transformer core. The exciting voltage
drives the flux in the core. An increase in the exciting voltage therefore increases the flux. To
maintain this additional flux, which may be in the saturation range of the core steel of the
transformer, the transformer draws more current, which can be in excess of the full load rating
the transformer windings. This additional current is the inrush current necessary to supply the
magnetizing branch of the transformer. [1]
To show magnetizing inrush current graphically, consider the equivalent circuit of
transformer shown in Figure 2. In an ideal transformer (with a 1:1 turns ratio), the currents I1
and I2 are equal except for the small current flowing through the shunt element of the
magnetizing branch. The increase in flux caused by an increase in the exciting voltage draws
more current through the magnetizing branch. When the transformer is being energized, this
current flows through only one winding. In this example, the current I1 is the inrush current.
During inrush events other than energization, the magnetizing inrush current may appear in both
windings, with the inrush current more prevalent in one winding. Remembering the differential
current is I DIFF I 1  I 2 , then in any inrush event, the magnetizing inrush current results in a
differential current. This differential current can lead to operation of the differential protection.
Figure 3 is an example of magnetizing inrush current and the resulting differential current.

120
Practical Experience in Setting Transformer Differential Inrush Restraint

R1 X1

I1 I2

Magnetizing
Current Rm Xm

Figure 2: Transformer equivalent circuit

Figure 3: Inrush current and resulting differential current

A review of AC excitation of magnetic materials helps understand the actual characteristic of
magnetizing inrush current. The magnetic steel used in transformers has a large number of
regions (known as “domains”) with a specific magnetic moment. An external magnetizing force
causes all the magnetic moments of the steel to align with the applied magnetic field. In the case
of transformers, the excitation voltage provides this applied magnetic field. The alignment of the
magnetic moments causes an increase in flux density greater than that of the external magnetic
field. The steel is fully saturated when all the magnetic moments are aligned with the applied

121
Practical Experience in Setting Transformer Differential Inrush Restraint

field. Once the external field is reduced, the magnetic moments maintain a net magnetization
component along by the direction of the field. This effect results in magnetic hysteresis of the
steel.[2] Transformers use grain-oriented electrical steel, where the domains tend to produce
directions of magnetization with high permeability and low core loss.

M
e M M'' M''

M' M'
iM''
iM' Time iM'' iM' iM
iM

Figure 4: Transformer core excitation phenomena

Figure 4 shows the exciting voltage e, the core flux M, and the exciting current iM, of a
transformer. The figure also shows the flux and exciting current mapped to the corresponding
magnetic hysteresis loop. The excitation voltage drives the flux in the core. The exciting current
is needed to produce the magnetic field. The waveform of the exciting current varies from the
sinusoidal waveform of the flux due to the non-linear magnetic properties of the core.[2]
The waveforms and hysteresis loop shown in Figure 4 are typical for a transformer that is in
service supplying load. The flux requirement is very small, and therefore the exciting current is
very small. Now consider what happens when the excitation voltage increases. This voltage
drives an increase in the flux in the core. The flux characteristic is still sinusoidal in shape. The
flux may be shifted in respect to the 0-axis due to the point on the wave when the excitation
increases, and the amount of remanent flux in the core. This flux may be high enough to cause
saturation of the transformer core. The hysteresis loop becomes negligible for this case, as shown
in Figure 5. The resulting current, the magnetizing inrush current, needed to supply the flux is
very high in magnitude, and may approach the magnitude of fault currents. The magnetizing
current will eventually decay due to losses in the circuit.[3]

122
Practical Experience in Setting Transformer Differential Inrush Restraint

iM
M

e
M
saturation knee
pont
Time iM

2D radians

Figure 5: Flux and exciting current hysteresis during core saturation

When the transformer core is in saturation, the exciting current is part of a sine wave for the
period that the flux exceeds the saturation knee point of the core. The exciting current is
essentially zero for the rest of the power system cycle. This results in the classic waveform
signature of magnetizing inrush current, as shown in Figure 6.

Figure 6: Magnetizing inrush current characteristic

The magnitude and characteristic of the inrush current is dependent on the amount of
saturation of the transformer core. There are several factors that influence the likelihood the
transformer core will go into saturation.

3.2. Point on wave

The key factor in determining the magnitude of the magnetizing inrush current is the point on
the voltage wave when the excitation voltage increases. If the excitation voltage is defined by

123
Practical Experience in Setting Transformer Differential Inrush Restraint

e(t ) E max cos(Zt  T ) , then the flux is defined by

M (t ) M max sin(Zt  T )  M max sin(T ) .

Where e(t) = instantaneous excitation voltage,
Emax = peak exciting voltage,
Z = system frequency,
T = voltage angle T defined in Figure 7,
M(t) = instantaneous flux,
Mmax = peak flux.

Voltage
angle

e M

Time

Start of event
Figure 7: Voltage angle during magnetizing inrush

It is obvious that the flux is offset with respect to the 0-axis based on the voltage angle T.
When the voltage angle T is 90q, the flux is fully offset. There is no offset when the voltage angle
T is 0q. Maximum saturation of the transformer core occurs when the flux is fully offset at the
90q voltage angle. [3]

124
Practical Experience in Setting Transformer Differential Inrush Restraint

Voltage M
angle
90o

Time

Start of event
Figure 8: Exciting voltage and flux at 90q voltage angle

3.3. Remanent flux

When a transformer is de-energized, some level of flux remains in the transformer core. This
level of remanent flux is the flux in the core when the exciting voltage is removed. The actual
value of the flux is based on the alignment of the magnetic moments of the steel, and can be
found from the magnetic hysteresis loop of the transformer core. The remanent flux may
therefore be positive or negative in value, and is typically 30% to 80% of the maximum flux of
the core. When the transformer is energized, this remanent flux is added to the flux driven by the
exciting voltage. The flux equation therefore becomes
M (t ) M max sin(Zt  T )  M max sin(T )  M remanent , where Mremanent is the remanent flux in the core.
The flux characteristic during an inrush event is then offset with respect to the 0-axis. Depending
on the sign of the remanent flux, the transformer core may be more or less likely to go into
saturation. The impact of the remanent flux is removed once the core is fully saturated.

M
e e

Time Time

Start of event Start of event

Positive remanent flux Negative remanent flux

Figure 9: Excitation voltage and flux at 90q voltage angle with remanent flux

125
Practical Experience in Setting Transformer Differential Inrush Restraint

3.4. Transformer design and magnetizing inrush current

The design of the power transformer influences the likelihood that the transformer core will
saturate during inrush events. A transformer core is built from thin strips of high-grade electrical
steel called laminations. The laminations are electrically isolated by a thin coating of insulation,
and then stacked or wound to create the core section. The flux density of the steel, the design of
the core, and the method of connecting the laminations all impact the amount and characteristics
of the magnetizing inrush current.
Over the last few decades, there are some changes in transformer design that impact the
second harmonic ratio during magnetizing inrush. The standard transformer design typically uses
M-6 conventional grain-oriented electrical steel. M-6 steel has a saturated flux density of 1.8
Teslas, the highest of any magnetic material. This very magnetically efficient steel results in
lower exciting currents and therefore lower inrush currents. However, this has been the standard
core material in transformer design for many years. Some transformers are now designed using
high-permeability (High-B) electrical steel. High-B steels provide more consistent grain
orientation, resulting in a more linear magnetic hysteresis loop.
A more important change has been in the construction of the core. Laminations were stacked
on top of each other, resulting in an air gap between each lamination. The air gap increases the
reluctance of the core, thereby reducing the magnetic efficiency of the core. Laminations are now
constructed such that they overlap each other to provide a continuous path for the flux. This
construction reduces the reluctance in the core, and therefore increases the flux density and
reduces the exciting current.
The other significant change in transformer design is based around the economic concerns of
loss evaluation. The trend is to select transformers based on loss evaluation. To limit losses,
transformers are designed with lower maximum flux densities. The flux density is limited by
using a core with a larger cross-sectional area. The relation between maximum flux density and
the exciting current is given by the following equation:
E rms I M ,rms 4.44 fAc l c Bmax H rms

where Erms = excitation voltage (rms)

IM,rms = exciting current (rms)
f = system frequency
Ac = cross-sectional area of the core
lc = length of core path
Bmax = maximum flux density
Hrms = permeability of the core (rms)
If the excitation voltage is constant, then a transformer with a lower flux density has a lower
level of exciting current. Reductions in the required exciting current lead to a reduction in the
magnetizing inrush current.[4] So the combination of efficient transformer core steel, better
construction of the core, and the limiting of the maximum flux density, leads to lower exciting
currents and lower magnetizing inrush currents.

3.5. Power system impedance

The physical installation of the transformer also influences the magnetizing inrush current.
The exciting voltage at the transformer is the system source voltage minus the voltage drop

126
Practical Experience in Setting Transformer Differential Inrush Restraint

across the system impedance. As the source impedance decreases, indicating a stronger source,
the magnitude of the inrush current increases. The resistance of the system is also a major
contributor to the decay of the inrush current over time. The change in flux over time is defined
by
t T
'M ³
t
R u i dt

where 'M = flux change per cycle,

R = total series resistance including transformer winding resistance
T = period of one cycle.
The flux in the transformer due to the inrush event begins to decay immediately by this
amount, and decays until steady-state magnetizing flux is reached. As the flux controls the
magnetizing current, the current also decays to steady-state magnetizing levels.[5]

3.6. The characteristics of transformer inrush current

As previously described, the classic inrush restraint current is similar to that of Figure 10. The
non-linear nature of the magnetizing inrush current results in harmonics being present. It is
possible to estimate the level of these harmonics by using Fourier series analysis on the
magnetizing inrush current. The flux is above the saturation knee point for a total angular span of
2D radians. During this span, the exciting current is a portion of a sine wave. During the rest of
the power system cycle, the exciting current is essentially 0.
Magnitude

D radians 2S 4S Time

Figure 10: Inrush current characteristic

This Fourier analysis shows that the second harmonic is the predominant harmonic during
transformer inrush events and is commonly used as the basis for inrush restraint functions. As the
saturation angle D increases, the exciting current becomes more linear and the ratio of second
harmonics to fundamental decreases.[3] This means, in effect, the more severely the transformer
core is saturated, the more linear the magnetizing inrush current.

127
Practical Experience in Setting Transformer Differential Inrush Restraint

Setting the midpoint of the first peak to time t=0, a cosine Fourier series may be used to
calculate the second harmonic current and fundamental frequency component. Assuming the
exciting current is truly symmetric, then the second harmonic ratio will be as high as 70.5%
when D = S/3 radians, and will be 17.1% when D = 2S /3 radians.[3]

3.7. Summary of transformer inrush phenomena

Transformer inrush occurs whenever the excitation voltage on the transformer increases.
Increasing the excitation voltage increases the flux in the transformer core, and therefore requires
more current from the system to supply the flux. The new level of flux, and the period of the
power system cycle the transformer core is in saturation determine the characteristics of the
inrush current. The flux is offset based on the point on the voltage wave when the change in
excitation occurs, and directly influences the level of saturation of the transformer core.
Remanent flux in the core also initially offsets the flux in the core. The design and location of the
transformer also impact the amount of saturation of the transformer core.
Of more interest for protection purposes are the characteristics of transformer inrush current.
The common techniques for preventing the operation of differential elements for inrush events
use the linearity of the differential current signal. The ratio of second harmonic current to the
fundamental current is often used. The more linear the inrush current, the less second harmonic
current is present. Therefore, as the level of core saturation increases, the ratio of second
harmonic current to fundamental current decreases.

4. Events that result in magnetizing inrush currents

Any event on the power system that causes a significant increase in the magnetizing voltage
of the transformer core results in magnetizing inrush current flowing into the transformer. The
three most common events are:
Energization of the transformer. This is the typical event where magnetizing
inrush currents are a concern. The excitation voltage on one winding is increased
from 0 to full voltage. The transformer core typically saturates, with the amount
of saturation determined by transformer design, system impedance, the remanent
flux in the core, and the point on the voltage wave when the transformer is
energized. The current needed to supply this flux may be as much as 40 times the
full load rating of the transformer, with typical values for power transformers
around 2 to 6 times full load rating.[1] The waveforms of Figure 3 were recorded
during energization of a transformer.
Magnetizing inrush current during fault clearing. An external fault may
significantly reduce the system voltage, and therefore reduce the excitation
voltage of the transformer. When this fault is cleared, the excitation voltage
returns to the normal system voltage level. The return of voltage may force a dc
offset on the flux linkages, resulting in magnetizing inrush current. This
magnetizing inrush current will be less than that of energization, as there is no
remanent flux in the core.[3] The current measured by the differential relay will
be fairly linear due to the presence of load current, and may result in low levels of
second harmonic current.

128
Practical Experience in Setting Transformer Differential Inrush Restraint

Sympathetic inrush. Energizing a transformer on the power system can cause

sympathetic magnetizing inrush currents to flow in an already energized parallel
transformer. Energizing the second transformer causes a voltage drop across the
resistance of the source line feeding the transformers. This voltage drop may
cause a saturation of the already energized transformer in the negative direction.
This saturation causes magnetizing inrush current to supply the flux. The
magnitude of the magnetizing inrush current is generally not as severe as the other
cases.[3][5][6]

Current
Tx 1

Tx 2
Time
R + jX

Tx 2
Time
Tx 1

Current
Figure 11: Sympathetic inrush circuit and waveforms

5. Transformer inrush restraint methods for differential protection

Transformer inrush restraint functions are intended to block the differential element from
operating during such an inrush event and permit the differential element to operate for internal
fault events. The challenge, obviously, is that inrush current and an external fault both present a
large differential current to the differential element. There are many different methods that have
been proposed and implemented for restraining the differential element during a transformer
inrush condition. These methods are discussed in a paper presented at this conference in 2000.[7]
FPL has already decided on a specific model of transformer differential relay for their
standard protection of transformers at the transmission level. Therefore, this paper discusses only
the options for inrush restraint available in this model of relay. This paper also discusses the
choice of inrush restraint mode. The restraint mode determines if inrush restraint is applied on a
per-phase or multi-phase basis. The inrush restraint methods available to FPL in their chosen
relays are the harmonic restraint and adaptive harmonic restraint functions.

5.1. Harmonic restraint

Harmonic restraint is the classical way to restrain tripping. There are many variations on this
method. All of these methods work on the assumption the magnetizing inrush current contains
high levels of second harmonic current. The current for an internal transformer fault typically has
very low levels of second harmonic current. The simplest method of harmonic restraint uses the
magnitude of the second harmonic in the differential current compared to the magnitude of the

129
Practical Experience in Setting Transformer Differential Inrush Restraint

fundamental frequency component in the differential current. Tripping of the differential element
is blocked when this ratio exceeds an adjustable threshold.

Differential
Current

Second
SP H/ harmonic
current
Differential
trip level

H/ SP

Second harmonic ratio

(20% typical)

NOT

AND

TRIP

Figure 12: SAMA diagram for second harmonic restraint

In this paper, the term “second harmonic ratio” is defined as:
I Differential ( 2nd harmonic )
I Differential ( fundamental )
This method originated in electro-mechanical relays, and has been carried through as the
most common method in microprocessor relays. The harmonic restraint is typically calculated on
a per-phase basis. Variations include using the RMS current as opposed to the fundamental
frequency component, and using a cumulative three-phase implementation.
The historical setting for harmonic restraint is a second harmonic ratio of 20%, with an
available setting range of 1% to 40%. Set too high, and the differential element may trip during
transformer energizing. Set too low, and inrush restraint may block tripping during some internal
fault events.

5.2. Adaptive harmonic restraint

Adaptive harmonic restraint is a modified version of traditional harmonic restraint that
considers the magnitude and phase of the second harmonic and fundamental frequency
component in the differential current. Some inrush events initially produce low levels of second
harmonic in the differential current, as in the example of Figure 13. This phenomenon is an
indication the remanent flux in the core initially pushes the core deeper into saturation. This low
level of second harmonic current may allow the differential element to operate.

130
Practical Experience in Setting Transformer Differential Inrush Restraint

Figure 13: Second harmonic current during inrush

The adaptive harmonic restraint method dynamically changes the inrush restraint level to
properly restrain the differential element for these cases. This method uses the discriminating
signal:
o
o I2
I 21 o
I 1 u e jZt
o
where I2 is the second harmonic differential current phasor,
o
I1 is the fundamental differential current phasor, and
Z is the system frequency.
o
The phase angle of the I 21 discriminating signal is always 90q or 270q for an inrush
condition. Consider a typical harmonic restraint threshold of 20%, as plotted on the polar graph
of Figure 14. The adaptive harmonic restraint initially has a lower inrush restraint threshold
along the 90q and 270qaxes. This threshold is dynamically raised to the default setting of 20%
over a period of 5-6 cycles.[8]

131
Practical Experience in Setting Transformer Differential Inrush Restraint

90o

120o 60o

INHIBIT

150o 30o

180o 0.30 pu 0.20 pu OPERATE 0.20 pu 0.30 pu 0o

210o 330o

INHIBIT

240o 300o

270o

Figure 14: Adaptive harmonic restraint characteristic

The adaptive harmonic restraint algorithm successfully restrains tripping when faced with
low levels of second harmonic current during an inrush event. However, this algorithm may slow
tripping of the differential element by a few cycles for an internal fault if some second harmonic
is present in the current.

5.3. Inrush restraint mode

Harmonic restraint and adaptive harmonic restraint are normally calculated individually on each
phase. Typically, the operation of the differential element on any phase operates the circuit
breakers supplying the transformer. If the restraint criterion is not met on any phase, the
transformer may be tripped offline during energization or sympathetic inrush. Depending on the
transformer installation, operating requirements, and operating philosophy, this may be
acceptable performance of the differential element. However, for the standard FPL application, it
is more desirable to increase the security of the differential element against inrush events by
looking at inrush restraint functions in more than one phase. The inrush restraint mode is
therefore the method of implementing the inrush restraint function across the entire transformer.
Per-phase. In per-phase mode the relay performs inrush restraint individually in
each phase.
2-out-of-3. In 2-out-of-3 mode, the relay checks second harmonic level in all
three phases individually. If any two phases establish a blocking condition, the
remaining phase is restrained automatically.
Averaging. In averaging mode, the relay first calculates the average second
harmonic ratio, and then applies the inrush threshold to the calculated average.

132
Practical Experience in Setting Transformer Differential Inrush Restraint

1-out-of-3. In 1-out-of-3 mode, all three phases are restrained when a blocking
condition exists on any one phase. 1-out-of-3 mode typically reverts back to per-
phase mode after a short time delay to allow tripping in case an internal fault
occurs during energization.
These restraint modes may be explicit settings of the transformer differential element. They
may also be implemented in the flexible configuration logic of the transformer protection relay.

Figure 15: Second harmonic during energization

The transformer differential protection operating for an inrush event is a loss of security.
These inrush restraint modes are listed in order from the least secure to the most secure.
Comparing these restraint modes to an actual transformer energization event can provide some
illustration of performance. For this event, the magnetizing inrush currents during energization of
the transformer were high enough to cause operation of the differential element. The second
harmonic current ratios are shown in Figure 15. The B-Phase ratio is greater than 1 in this case,
and is not shown on the graph. This differential element uses traditional harmonic restraint set at
20% for the inrush restraint mode. Tripping on any phase de-energizes the transformer.
Table 1 lists the performance of the various restraint modes for this example. Per-phase mode
will allow the differential to operate, while all the other modes will block the differential. The
correct choice is a matter of application, and a matter of operating philosophy. Per-phase mode
may be the most appropriate solution for a three-phase bank made up from single-phase
transformers, for example.

133
Practical Experience in Setting Transformer Differential Inrush Restraint

Table 1: Restraint mode results

Restraint mode Result

Per-phase Differential trips on A-Phase

Differential restrains: B-Phase and C-Phase
2-out-of-3
blocked
Differential restrains: at t=0, = (0.05 + 6.12 +
Averaging
0.29) / 3 = 2.15
Differential restrains: B-Phase and C-Phase
1-out-of-3
Blocked

6. FPL testing of inrush restraint methods

Ideally, setting the inrush restraint function for the transformer differential element is exactly
like setting any other protection function. The key criteria are sensitivity and selectivity. The
setting must be sensitive enough to recognize magnetizing inrush current even with low levels of
second harmonics. And it must be selective to distinguish between an inrush event and other
events that may produce second harmonic current.
The standard process for setting a protection function is to perform an analysis of the system.
This analysis models the response of the system for various transient events to provide a basis for
setting protection functions. For example, to set a distance element, the first step is to perform a
short circuit analysis. The short circuit analysis is built around the known quantity of the system
impedances. Different predictable scenarios are used during the analysis; basically system
operating contingencies and fault location. Then some educated guesses are used for the
influence of unknown variables, such as fault resistance.
A similar process can be identified for setting the transformer inrush restraint function but the
results are not as predictable as short circuit analysis. The known quantity for inrush restraint
analysis is that of transformer design and transformer location. There isn’t any “predictable”
scenario for modeling the inrush. There is only the ability to make educated guesses that relate to
the type of event that causes inrush, the point on the voltage wave when this event occurs, and
the amount of remanent flux in the core of the transformer. And unlike the influence of fault
resistance on a short circuit analysis, there is no empirical model of a transformer for the
influences of some of these factors.
Therefore there is no empirical method for determining the magnitude of inrush currents and
the second harmonic ratio in the differential current. FPL used a model of the system and
transformer to produce some digital representation of inrush events, and also used oscillography
captured during transformer inrush events to develop settings using an experimental process.
A protective relay that FPL has selected for use in the transformer protection panel provide
the following options:
Inrush restraint function: harmonic restraint
adaptive harmonic restraint
Inrush restraint mode: per-phase
2-out-of-3

134
Practical Experience in Setting Transformer Differential Inrush Restraint

averaging
In addition, the 1-out-of-3 inrush restraint mode can be implemented using the
flexible configuration logic of the relay.
This test procedure uses wye-connected CTs that are typical on new installations using
microprocessor-based differential relays. Delta-connected CTs complicate inrush restraint
settings. The currents measured by the relay are phase-phase currents. The subtractive effect of
the delta connection may actually decrease the second harmonic current seen by the relay and
require a lower setting on the inrush restraint function.[9]
Therefore, the concern for setting the inrush restraint function is the ratio of the second
harmonic current to the fundamental current. Having some guidelines that predict this ratio will
help develop inrush restraint settings that are sensitive and selective.
To determine the most appropriate choice for the inrush restraint method and the inrush
restraint mode, a simple bench test experiment was devised. The
procedure was to simply choose some settings for the relay, connect the
relay to a three-phase test set, and play several oscillography files SET RELAY Function
through the relay. SET RELAY Mode
SET RELAY Level

The oscillography file of Figure 17 shows a misoperation of a

differential relay caused by low levels of second harmonic current
during energization of an autotransformer. This transformer is a bank of PLAY FAULT

3 single-phase 500MVA, 500kV/230kV autotransformers, and was

energized from 500kV.
Before beginning the testing, FPL set the following criteria for RECORD RESULTS
acceptable settings.
x A minimum level setting for harmonic restraint of 15%.
x No operating time delay for the differential element is Figure 16: FP&L Test
introduced by the selected inrush restraint function or Procedure
inrush restraint mode.
x The differential element is blocked from operating
during this actual energization event.
Based on these criteria, the adaptive harmonic restraint restraint function and the 1-out-of-3
restraint modes were not tested. The adaptive harmonic restraint function could possibly slow
tripping of the differential element for internal faults where CT saturation could occur. The 1-
out-of-3 mode must be implemented in the flexible configuration logic of the relay. The concern
is this logic may introduce a time delay when blocking the differential element. Also, FPL would
like to avoid custom logic for protection functions as much as practical.

135
Practical Experience in Setting Transformer Differential Inrush Restraint

Figure 17: Oscillography file used for test procedure

6.1. Test results

The more detailed test procedure is shown in Figure 18. The initial setpoint for the inrush
restraint level was 20%. If the differential element tripped at this level, the level was then
reduced to 15%. If the differential element continued to trip, the level was reduced until restraint
was achieved. The results of the test are tabulated in Table 2.

136
Practical Experience in Setting Transformer Differential Inrush Restraint

SET RELAY Mode

SET RELAY Level

PLAY FAULT

Y
Differential
REDUCE Level
Trip

RECORD Mode, Level

CHANGE Mode

Figure 18: Detailed test procedure

These results match the expected results for the different inrush restraint modes. No level
was recorded for the per-phase option, as a review of the fault data indicated the level would be
10% or less. This is an unacceptably low setting. Using the averaging mode, the differential
element was blocked at 20%. The inrush restraint level was raised until a trip occurred to give
some idea of the margin between blocking and tripping for the differential element.
Table 2: Harmonic restraint test results

Mode Level to achieve blocking

Per-phase No level recorded

2-out-of-3 13% (Trips at 14%)
Averaging 20% (Trips at 21%)

7. Conclusions
The traditional settings for inrush restraint for transformer differential protection are to use
harmonic restraint in a per-phase mode, with a restraint level of 20%. Experience shows that for
most transformer protection applications these settings provide high-speed clearing of
transformer faults and proper restraint for inrush events. For a few applications, lowering the
harmonic restraint setting, employing cross-blocking techniques, or a combination of both may
be necessary to ensure that transformer energizations occur successfully where very low second
harmonic inrush levels are present. Capturing energization records to confirm where these
techniques may need to be employed is essential.
Based on operating experience, FPL has a few transformer protection applications where the
traditional inrush restraint settings are not adequate. For these applications, FPL plans to use

137
Practical Experience in Setting Transformer Differential Inrush Restraint

either 2-out-of-3 or averaging mode, to provide greater security, along with a setting for the
harmonic blocking function in the range of 15 to 20%. This range, along with applying cross-
phase blocking techniques, provides a good margin of security for energization of the
transformer where the second harmonic current may fall to low levels. The belief is these settings
will also successfully restrain the differential element when faced with a fault recovery
magnetizing inrush event.
There may be some concern when setting the inrush restraint level to 15% or lower. CT
saturation during internal faults may result in the protection relay seeing a high second harmonic
current ratio and incorrectly restraining. Setting an unrestrained differential element between 8
per unit and 10 per unit provides confidence the transformer protection will trip for an internal
fault even at a lower setting for the inrush restraint.
FPL has used this process to guide the development of inrush restraint settings. The limited
experience to date with intelligent consideration of the settings has been successful. The
following event records are from two different transformer locations that used these settings. The
transformers in both cases had previously been energized, so remanent flux was present in the
transformer core. Figure 19 is for a 560MVA, 230kV/138kV three-phase autotransformer,
energized at 230kV. Figure 20 is for a 224MVA, 230kV/138kV autotransformer, energized at
230kV.

Figure 19: Energization of one autotransformer

138
Practical Experience in Setting Transformer Differential Inrush Restraint

Figure 20: Energization of a second autotransformer

Setting the inrush restraint function takes some knowledge and experience. The transformer
design and the system impedance have some influence on the magnitude and severity of the
inrush event. Capturing oscillography data for every energization event for a specific transformer
may be used to generalize about the characteristics of inrush currents for a specific transformer.
If a transformer seems to provide low levels of second harmonic current over a number of
events, it may be necessary to lower the inrush restraint function, use adaptive harmonic
restraint, apply cross-phase blocking techniques or use a combination of these methods. FPL
plans to capture and analyze oscillography data on every energization of a specific transformer to
develop the operating history for a specific transformer.

139
Practical Experience in Setting Transformer Differential Inrush Restraint

8. Symbols
The process control industry has developed symbols and diagramming formats to represent
both linear and non-linear processes. The symbols and diagramming format are commonly
known as “SAMA diagrams”, as they were originally a standard developed by the Scientific
Apparatus Makers Association. Though the Scientific Apparatus Makers Association has
declared the original standard obsolete, and no longer permits the direct association of the
organization name with the industry standard, these symbols, and the term “SAMA diagram”, are
still in common use in the process control industry. These symbols will grow increasingly useful
as protection systems migrate from traditional protection and control to automatic process
control.

Symbol Function Symbol Function

Measuring H/ High signal monitor

Setpoint Dividing

Automatic Signal Processing NOT Inverter

Final Controlling AND Logical AND

9. References
[1] L. L. Grigsby, editor, “The Electric Power Engineering Handbook”, CRC Press LLC, Boca
Raton, FL, 2001, p. 3-29.
[2] A. E. Fitzgerald, C. Kingsley, Jr., S. D. Umans, “Electric Machinery, 4th edition”,
McGraw_Hill Book Company, New York, NY, 1983, pp. 4-24.
[3] S. H. Horowitz, A. G. Phadke, “Power System Relaying, 2nd edition”, Research Studies Press
Ltd., Somerset, England, 1995, pp. 210-214.
[4] C. E. Lin, C. L. Cheng, C. L. Huang, J. C. Yeh, “Investigation of Magnetizing Inrush
Current in Transformers Part I – Numerical Simulation”, IEEE Transactions on Power
Delivery, Vol. 8, No. 1, January, 1993, pp. 246-254.
[5] H. S. Bronzeado, P. B. Brogan, R. Yacamini, “Harmonic Analysis of Transient Currents
During Sympathetic Interaction”, IEEE Transactions on Power Systems, Vol. 11, No. 4,
November 1996.
[6] M. M. Saied, “A Study on the Inrush Current Phenomena in Transformer Substation”,
Conference Record of the 36th Annual Meeting of the IAS, 2001. pp. 1180-1187.

140
Practical Experience in Setting Transformer Differential Inrush Restraint

[7] R. W. Patterson, W. P. McCannon, G. L. Kobet, “A Consideration of Inrush Restraint

Methods in Transformer Differential Relays”, 54th Annual Georgia Tech Protective Relaying
Conference, Georgia Tech, Atlanta, GA, May 3-5, 2000.
[8] B. Kasztenny, A. Kulidjian, “An Improved Transformer Inrush Restraint Algorithm
Increases Security While Maintaining Fault Response Performance”, 53rd Annual
Conference for Protective Relay Engineers, Texas A&M, April 11th – 13th, 2000.
[9] G. L. Kobet, R. W. Patterson, “Matlab Analysis of Braytown Transformer Differential Inrush
Misoperation”, Georgia Tech Fault and Disturbance Analysis Conference, Atlanta, GA, May
1-2, 2000.

10. Bibliography
[B1] C. E. Lin, C. L. Cheng, C. L. Huang, J. C. Yeh, “Investigation of Magnetizing Inrush
Current in Transformers Part II – Harmonic Analysis”, IEEE Transactions on Power
Delivery, Vol. 8, No. 1, January, 1993, pp. 255-263.
[B2] GE Publication GEK-113327A, “T60 Transformer Management Relay”, Instruction
Manual, Markham, Ontario, Canada, 2006
[B3] G. Ziegler, “Numerical Differential Protection”, Publicis Corporate Publishing,
Erlangen, Germany, 2005. pp. 146-151.

Authors
Rich Hunt, Application Engineer, GE Multilin
Rich Hunt is an Application Engineer with GE Multilin, responsible for technical sales, technical
marketing, and technical support of GE Multilin products. Rich has 20 years experience in
electric utility systems, including 10 years with Virginia Power, and 10 years experience as an
Application Engineer for relay manufacturers. Rich earned the B.S.E.E. and M.S.E.E at Virginia
Tech, with a master’s thesis on applications of protective relays. He is registered Professional
Engineer in the Commonwealth of Virginia, and is a member of the Main Committee of the
IEEE Power System Relaying Committee.
Joe Schaefer, FPL
Joe Schaefer is responsible for developing and testing protective relay standards related to
generation, transmission, and distribution applications. His most recent designs include relay
protection for auto-transformers for applications up to 500kV. Previously, Joe was employed as
a protection field engineer responsible for relay equipment from 480V to 500kV applications
including nuclear plant relaying. Joe received his BSEE from University of Florida and joined
Florida Power and Light in 1987.
Bob Bentert, FPL
Bob Bentert is a Senior Engineer responsible for operations support of protective relay and
control equipment. Bob evaluates protective relays and performs in-depth analysis of protective
relay misoperations. Prior to FPL, Bob was the Customer Services Manager for GEC Alstom
Relays and worked for ASEA Protection and Control as a project manager.

141