Sunteți pe pagina 1din 11

1704

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

Design and Protection of Captive MotorTransformers


Daniel Doan, Member, IEEE, John Crouse, Member, IEEE, N. Kent Haggerty, Senior Member, IEEE, and
Miroslav Ristic, Member, IEEE

AbstractIn a cost-competitive environment, a large motor installation can make up a major part of a process project. This paper
explores the issues of reliability and cost for the sizing and protection of captive transformers of differing technologies, so that the
overall life-cycle cost of a captive motor-transformer installation
can be minimized.
Index TermsCaptive transformer, large motor, protective relaying, transformer selection.
Fig. 1. Captive motor/transformer single line.

I. INTRODUCTION

APTIVE transformer applications have been the subject


of very little recent publication. This configuration is encountered on occasion for powering large motors, with loads
such as a large chiller or process compressor. The usual voltage
level is from 2.3 to 4.1 kV, and the motor size is normally 1000
hp or larger.
The idea for this paper came from startup work done on a
3000-hp chiller machine, with a dry-type captive transformer.
Some design issues of a system of this type are summarized in
the following questions.
What is the best selection and sizing of the transformer?
Would the size of the transformer be different if liquidfilled or cast-coil transformers were considered?
Would these decisions affect the application of protective
relaying?
What is the proper protection of the motor?
What is the proper protection of the transformer, including
consideration of motor starting currents?
For the purposes of this paper, a captive transformer is defined as a transformer in series with a large motor, where the
transformer is under continuous load, and the combination is
used in an across-the-line starting application. See Fig. 1 for
typical single-line diagrams of this system. Both the transformer and motor can be energized at the same time by a
single breaker in the primary circuit of the transformer. Alternately, the motor can be started by a medium-voltage motor
starter in the transformer secondary. The autotransformer rePaper PID 006, presented at the 1999 IEEE Petroleum and Chemical Industry Technical Conference, San Diego, CA, September 1315, and approved
for publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the
Petroleum and Chemical Industry Committee of the IEEE Industry Applications
Society. Manuscript submitted for review September 15, 1999 and released for
publication May 19, 2000.
D. Doan and N. K. Haggerty are with E. I. duPont de Nemours & Company, Inc., Wilmington, DE 19880 USA (e-mail: doan@ieee.org; N-Kent.Haggerty@usa.dupont.com).
J. Crouse is with General Electric Company, Rome, GA 30165 USA (e-mail:
john.crouse@indsys.ge.com).
M. Ristic is with GE MultiLin, Markham, ON L6E 1B3, Canada (e-mail:
miroslav.ristic@edc.ge.com).
Publisher Item Identifier S 0093-9994(00)09245-8.

duced-voltage starting of a large motor, sometimes referred


to as captive-transformer starting outside North America, was
not considered in this paper.
One reason for considering a captive transformer for
across-the-line starting is to ensure a higher starting torque
for the process. In a remote operation, the transformer can be
installed at the motor, and voltage drop due to a long run of
cable would be minimized. Other reasons for consideration
would be if the required motor voltage was not available on
the site, or to allow the use of an optimum motor voltage
rather than the existing system voltage. Another advantage
is that the voltage drop during motor starting is isolated
from other loads.
II. TRANSFORMER DESIGN ISSUES
A. Motor Starting Impact
One engineering factor that must be considered when
selecting a captive transformer is voltage drop at the terminals
during motor starting. Per-unit motor torque will vary with the
per-unit voltage squared. If the terminal voltage is 0.80 per
unit during starting, the delivered motor torque will be 0.64 per
unit. This may or may not be sufficient torque to accelerate
inertia of the
the load to its full speed depending on the
rotor, coupling, gearbox, and driven load.
The two major transformer parameters that can be varied to
reduce voltage drop are percent impedance and kVA rating.
Increasing the kVA rating of the captive transformer while
keeping the percent impedance constant reduces the voltage
drop through the transformer by approximately the ratio of the
lower kVA rating to the increased kVA rating. The per-unit
voltage available during initial motor starting can be found by
using a voltage divider equation that calculates the ratio of the
motor per-unit impedance to system short-circuit impedance
plus motor per-unit impedance as follows:

00939994/00$10.00 2000 IEEE

(1)

DOAN et al.: DESIGN AND PROTECTION OF CAPTIVE MOTORTRANSFORMERS

1705

where
per-unit starting voltage;
per-unit initial or nominal voltage;
per-unit motor starting impedance;
per-unit system short circuit impedance.
Equation (1) calculates the initial voltage at the instant the
varies as the motor accelerates to
motor is started. Since
full speed, the motor terminal voltage and the resultant torque
also will vary.
Increasing the short-circuit availability (stiffness) of the
system by increasing the size of the captive transformer with
a constant percent impedance reduces the system short-circuit
in the denominator, thus increasing the available
impedance
motor starting voltage. Another method of increasing available motor terminal voltage is to specify a lower impedance
transformer. Most transformer manufacturers have standard
impedance offerings, such as 5.50% which is a common
percent impedance for medium-voltage transformers. Lowering
the impedance will also increase the available short-circuit
availability, thus reducing the voltage drop. Reducing the
impedance below the standard impedance may increase the cost
of the transformer. If the impedance is reduced below 5.00%,
the cost of the transformer may increase substantially because
of design changes required by the increased short-circuit forces
on the transformer. The other criteria that must be verified in
either case is that the momentary and interrupting ratings of the
electrical switchgear are not exceeded.
Another consideration for selection of a captive transformer
is the number of starts or pulses that the motor and transformer
will experience. Each time a motor starts, significant forces
are exerted on the transformer windings. Manufacturers utilize
unique construction and design techniques for each type of
transformer. Three types of transformers that can be used for
captive motor installations include liquid, dry, and cast coil. The
cast-coil transformer is really a special type of dry transformer,
but will be treated as a separate category in this paper.
B. Liquid-Filled Transformers
Liquid-filled transformers have their windings immersed
in mineral oil, a less flammable liquid such as silicone, or an
approved high-fire-point hydrocarbon liquid. This category of
transformer is usually the lowest cost of all three types because
the liquid is a better dielectric and a more efficient heat transfer
medium than either air or epoxy. In the size range of 10005000
kVA, liquid-filled transformers can be constructed with a core
having a rectangular, oval, or round cross section.
The secondary (LV) windings are wound on a winding form
that fits tightly over the core. The secondary windings typically
are sheet wound with one or more layers of paper insulation between each turn. The current distribution in the sheet windings
greatly reduces the magnitude of vertical (axial) forces during a
short circuit. A layer (barrel) winding with conductors of rectangular shape is normally used for the primary (HV) winding. Insulation between layers is provided by one or more thicknesses
of aramid or kraft paper.
The radial short-circuit forces are inward on the LV windings and outward on the HV windings. The LV windings can
be solidly supported from the core so that the inward radial

Fig. 2.

Flux and forces in liquid-filled transformer.

forces are resisted by the combined mechanical strength of the


winding and core. HV windings of rectangular construction can
be braced against the HV windings of adjacent phases or against
end channels to resist outward radial forces.
In order to provide increased short-circuit strength for
windings of rectangular construction, the layer insulation can
be coated with a B-staged epoxy that is cured by heat while the
windings are being pressed. This compression bonding secures
the windings together into a strong, solid mass.
Fig. 2 shows the leakage field and resultant forces in a liquidfilled transformer with layer windings. The forces will be perpendicular to the magnetic flux and will be much greater during
motor starting. The captive liquid-filled transformer must be
properly sized not only for the full load of the motor, but also for
starting and pulsating loads. This is discussed in greater detail
in Section II-F.
Advantages of a liquid-filled transformer include relatively
low cost and losses and a long life expectancy. Because of a
smaller size and better space factor, losses can be reduced with a
lower additional cost per watt of loss reduction than for dry-type
or cast-coil transformers. Also, this class of transformer has a
fairly good short-term overload capability. Liquid-filled transformers are suitable for outdoor and harsh environments.
C. Dry-Type Transformers
Open-ventilated dry transformers, typically referred to as
dry-type transformers, use high-temperature-rated insulation
components and air as the insulating medium. Dry-type transformers are slightly higher in cost than liquid-filled units. A
variety of winding types are used in dry-type windings in the
10005000-kVA range. The construction can be either round
or rectangular. The secondary (LV) windings can be of layer
(barrel) or sheet construction. The primary (HV) windings can
be of layer or disc construction. The layer windings are usually
composed of one or two sections. A two-section layer winding

1706

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

Fig. 3. Dry transformer force model.


Fig. 4. Axial force model.

has a gap in the middle of the winding between the top and
bottom sections. If vertical clamping is used, the windings can
be of upset disc or all upwound disc construction. An upset disc
winding is wound from the inside out, but alternate discs are
upset or restacked. This allows the crossovers to go from the
top of one disc to the top of the other disc, and from the bottom
of one to the bottom of the other. An all upwound disc construction can also be used in which the disc sections are wound into
slots cut in glass polyester sheets. An upwound disc is similar
to the upset winding, but alternate discs are not restacked. The
crossover conductor transitions from the top of one disc to the
bottom of the next disc and the winding process continues. Additional insulation must be used for the crossover conductors
due to higher turn voltages.
Fig. 3 is a cross-sectional view of a dry-type transformer with
a disc HV winding and a barrel LV winding showing the axial
and radial component of forces that are exerted during motor
starting.
Primary windings are typically disc wound with aramid paper
as the conductor insulation. After the coils are wound, they are
placed in metal molds, inserted into a vacuum chamber and
filled with a gas-free epoxy mixture. After filling, the molds
are then removed from the casting chamber and placed in a
timetemperature curing oven. After the epoxy is cured, the
coils are removed from the molds and are ready for assembly on
the core. Since the epoxy flows between the discs and has excellent dielectric properties, close disc-to-disc spacing can be utilized which increases the disc-to-disc capacitance and allows for
more uniform distribution of impulse voltages.The radial component tends to push the secondary winding into the core while
driving the primary winding away from the core. Transformer
manufacturers can compensate for radial forces with a system
utilizing wedging sticks, inherent strength of the conductor material, and the winding geometry. Wedging sticks are placed between the secondary windings and the core legs to prevent the
secondary windings from collapsing into the core. The primary
winding is kept in place due to the tension in the winding and
the strength of the conductor material. Circular coil geometry allows for uniform distribution of radial forces in the windings and
reduces the conductor radial movement during short circuits.
Fig. 4 illustrates a force model that can be used to depict the
is proporaxial forces in disc-type windings. The axial force
tional to the square of the ampere turns and inversely proportional to the distance between the discs. The bending moment

Fig. 5.

Impulse model.

is proportional to the axial force times the disc length, and the
deflection of the disc is proportional to the axial force times the
cube of the disc length. During motor starting, the captive transformer will be subjected to locked-rotor currents and forces.
Since air is the dielectric medium in dry-type transformers, the
only support to resist deflection of the disc is the strength of
the disc winding itself. Any increase in disc length will result in
greater deflection of the winding and higher bending moments.
Advantages of the dry-type transformers for captive load
applications include a relatively low cost and environmental
safety. A disadvantage in applications with frequent and
relatively high current pulses is lack of techniques to improve
the mechanical strength, such as compression bonding for
liquid-filled transformers and epoxy encapsulation for cast-coil
transformers. Dry transformers typically are less efficient than
liquid-filled or cast-coil units. The dry-type transformer can
also be adversely affected by environmental contaminants.
D. Cast Coil
The third type of transformer is the cast coil, which is really
a special type of dry-type transformer. The cast-coil technology
was first developed in Europe during the 1950s and later in the
U.S. in the late 1960s. The major difference between cast-coil
technology and that of liquid or dry is that the cast-coil transformer uses an epoxy resin as the major dielectric component
instead of air, mineral oil, or silicone. Epoxy has an excellent dielectric performance of over 3.5 that of air. While epoxy has
excellent dielectric properties, the epoxy has limited mechanical strength when used with no other material. Typically, fiberglass and aramid fiber are combined with epoxy for high me-

DOAN et al.: DESIGN AND PROTECTION OF CAPTIVE MOTORTRANSFORMERS

1707

Fig. 7.

Cast disc windingsforce model.

Fig. 6. Cross-sectional view of cast-coil winding.

chanical strength. Secondary windings are usually sheet windings insulated with an epoxy-coated aramid sheet. The inner
and outer surfaces of the cooling ducts are protected from the
environment with a glass-reinforced polyester sheet. After the
secondary coils are wound, they are placed in an oven where
the epoxy between the turns is cured, binding adjacent turns together. The final step is to fill the end caps of the windings with
epoxy, sealing the conductor from the environment.
Fig. 5 illustrates a model of a disc-type winding under an impulse condition. This closer disc spacing provides a higher basic
impulse level (BIL) than with conventional dry transformers.
Typical BIL levels with cast-coil transformers are similar to
those of liquid-filled units.
Cast-coil transformers have several distinct advantages over
dry and liquid-filled transformers. The main advantage of a
cast-coil transformer is that the polyester glass and epoxy
resin provides excellent mechanical strength and eliminates
conductor movement. Fig. 6 shows the typical winding arrangement of the cast-coil transformer.
As Fig. 6 illustrates, both the primary and secondary windings
are completely encapsulated in the epoxy resin. The forces on a
cast-coil transformer winding act as shown previously in Fig. 3.
As with other dry transformers, the radial component tends
to push the secondary winding into the core, while driving the
primary winding away from the core. In addition to using the
strength of the conductor material and winding geometry, the
cast-coil design incorporates the concrete-like epoxyconductor
matrix. Wedging sticks are placed between the secondary windings and the core legs to center the coils on the core legs. The
primary winding is completely embedded in a block of epoxy,
which prevents any winding movement.
Fig. 7 depicts a typical model used for axial forces in cast-coil
disc windings. The simple beam model clearly illustrates the
mechanical advantage of the epoxy support system over other
earlier technologies. Fig. 7 shows that the conductor is fully supported throughout the winding, eliminating any moments and
reinforcing the conductor. The epoxyconductor matrix forms a
physical structure which requires little maintenance and ensures
reliable operation during captive motor starting applications.
The epoxy resin essentially makes the cast-coil transformer
impervious to harsh environments and has no potential adverse
affects on the environment. Thus, it can be used virtually in any

TABLE I
TRANSFORMER COST AND LOSSES

application. Another advantage is that the cast-coil transformer


has a higher efficiency rating than liquid or dry transformers.
The major disadvantage of the cast-coil transformer is that
the initial cost can range from 1.5 to 2 the cost of a liquidfilled unit. However, the 80 C rise cast coil has an improved
efficiency rating, as can be seen in Table I.
E. Transformer Efficiency
Table I shows each transformer type with associated cost and
losses for both a standard- and premium-efficiency 1500-kVA
transformer. Since liquid-filled transformers typically have the
lowest initial price, a cost of 1.0 per unit was chosen as a reference base for the standard-efficiency liquid-filled transformer.
The Percent IRR gives the internal rate of return (IRR) for
spending the additional dollars to purchase a high-efficiency
transformer as compared to a standard-efficiency transformer.
The IRR numbers shown were based on a discounted cash-flow
analysis using $0.035/kWh, 8760 h of operation, a 75% transformer load factor, an inflation rate of 2.5%, and an expected
transformer life of 20 years. The analysis compares the added
cost for a high-efficiency transformer to the watts savings over
the life of the transformer. The IRR is simply the discount rate
that results in a zero net present value (discounted cash flow) in
the 20th year of the evaluation.
Purchase of high-efficiency liquid-filled transformers is almost always justified. High-efficiency dry transformers may not
be as easily justified, but an economic evaluation should be
made if dry transformers are being considered for motor applications. Cast-coil transformers have the highest first cost,

1708

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

but their standard efficiency is higher than the standard efficiency of liquid units, and higher than the premium-efficiency
dry-type units. In addition, the 80 C rise cast-coil transformer
has a higher short-term overload capability than either dry or
liquid-filled transformers.
F. Proper Transformer Sizing
The transformer size should be selected to be adequate for
the load of the application, the mechanical duty imposed by the
application, and the voltage drop requirements for starting the
motor. In this section, methods to determine the minimum transformer kVA required to handle the load and mechanical duty
imposed on the transformer by high starting currents and pulsating loads are presented.
In addition to the high inrush currents during motor starting,
some captive transformer applications also have periodic high
current pulses that place unusual thermal and mechanical duty
on the transformer. Some examples of these applications include
chippers, reversing hot mills, crop shears, punch presses, car
shredders and fragmentizers, hammer mills, and reciprocating
compressors or pumps.
The equivalent rms load for a typical load cycle of the motor
should be calculated that includes motor starting, steady load,
and pulsating load, if any. The equation for determining the
equivalent rms load is
(2)
where
motor inrush kVA;
steady kVA;
pulse kVA;
time that the kVA is at ;
time that the kVA is at ;
time that the kVA is at .
The maximum kVA rating of the transformer must be at least
as large as the thermal rating calculated by (2).
The forces produced in a transformer by motor starting currents or pulsating loads are the same as those produced by a
full short circuit except for a lower magnitude. The forces produced in the transformer are proportional to the pulse current
squared. Over a period of time, thousands of lower magnitude
short-circuit-type events can damage a transformer. Fig. 8 was
developed to provide guidance in sizing transformers with pulsating-type loads and has been used for many years in sizing
liquid-filled transformers in the 10005000-kVA range for pulsating or short-time loads [1]. It is based on the forced-air cooled
(FA) rating of the transformer, which is 1.15 the base natural
cooled (OA) rating for liquid-filled transformers with base ratings of 2000 kVA or less and 1.25 the base rating for liquidfilled transformers with base ratings of 210010 000 kVA. Most
liquid-filled transformers in the 10005000-kVA range are sold
with either an FA rating or with provisions for a future FA rating.
When a request for quotation is sent to transformer manufacturers, any pulsating-type load that the transformer will experience should be described so that each manufacturer can determine if a design with a higher mechanical strength than the
standard design is required.

Fig. 8.

Transformer application curve for pulsating loads.

The use of Fig. 8 is illustrated by the following example.


1) If the motor starting inrush kVA is 10 000 and the motor
is started fewer than two times per hour, the maximum
allowable value of from Fig. 8 is 4 and the transformer
FA rating must be greater than or equal to 10 000/4, which
is 2500 kVA.
2) If a short-term pulse (impact) load of 5100 kVA occurs
every minute (60 times per hour), the maximum allowable
from Fig. 8 is 1.7, and the transformer FA
value of
rating must be greater than or equal to 5100/1.7, which is
3000 kVA.
3) If the equivalent thermal loading calculated by (2) is 2830
kVA, then the transformer FA rating must be greater than
or equal to 2830 kVA.
The transformer kVA should be selected to be at least the
largest value of 1), 2), or 3), above. In this example, a transformer of at least 3000 kVA (FA) would be selected for the captive transformer application.
III. PROTECTION ISSUES
A. Protection Requirements
The protection requirements for a large captive motortransformer configuration are a complex combination of the different
needs for protection of the motor, transformer, and process. The
first tradeoff the designer will face is to decide which of the
motor, transformer, or process is considered more critical to
save. Another important determination is the relative importance of the available space for equipment and allowable capital investment, compared with the need for process reliability.
Higher reliability will require implementing a fully developed

DOAN et al.: DESIGN AND PROTECTION OF CAPTIVE MOTORTRANSFORMERS

protection, control, and monitoring system. The final captive


motortransformer configuration and the level of applied protection, monitoring and control will depend on individual answers to these questions.
When a large captive motortransformer configuration is a
considerable portion of total capital costs, a separate protection system for each piece of equipment is usually applied. This
means that the transformer and the motor will each have its
own dedicated circuit breaker and several sets of current transformers (CTs) and potential transformers (PTs). The applied
protection philosophy is a simple superposition of typical motor
and transformer protection with possible addition of a common
differential protection for the overall system, similar to a generatorunit transformer overall differential protection. The operation of this protection should be blocked, using 52a contacts, if
both breakers are not closed. In this way, when the transformer is
energized, only its own differential protection will be employed;
the overall differential will be blocked.
During the motor start, all differential protections would be
enabled. Motor start is always followed by zero- and negativesequence currents. Although the neutral point of the motor stator
winding is usually not grounded, the zero component current
can flow through a closed loop made of the capacitance between
the motor neutral point and ground, and back through the transformer ground. For this reason, the overall differential protection should eliminate the zero-sequence component of current
to eliminate false trips. The second reason for differential false
trips is existence of second harmonics during transformer energization, or fifth harmonics during transformer overexcitation.
The applied differential protection in a captive motortransformer configuration should be capable of blocking differential
trip when these harmonics have been detected.
Since the large captive motortransformer configuration consists of a separate motor and transformer, each of these pieces of
equipment has different failure modes, and has protection needs
based on its own characteristics.
B. Large Motor Protection
There are two types of ac motors: synchronous and induction. In general, synchronous motors occupy the top end, and
are used for larger torque and horsepower ratings for the same
speed. The efficiency of synchronous motors is higher, and for
large motors this is a very important economical factor. Another
advantage of large synchronous motors versus induction is that
they do not require power-factor correction. Induction motors
occupy the bottom end and are generally used for smaller torque
or horsepower ratings for the same speed. There is an overlapping range between these two different types of large ac motors.
For the same rating, a synchronous motor is more expensive, and
is more demanding from the protection point of view, as now the
field (i.e., rotor) has to be controlled, monitored, and protected.
Consequently, a large synchronous motor is protected with additional relaying, which is used for field control, protection, and
monitoring.
It is common for a large motor to have a stator winding in wye
connection. This allows each end of each winding to be brought
out and monitored for differential protection.

1709

TABLE II
MOTOR FAILURES AND PROTECTION

One trend in large motor manufacturing is very noticeable in


the last decade. The physical size per horsepower of large motors is dropping and, consequently, the unit price expressed in
dollars per horsepower is dropping. Motor manufacturers are
improving their designs in order to lower their manufacturing
costs, so they can be more competitive in the market. Some manufacturers may not provide all six stator leads at the motor terminal box if they are not specifically requested to do so. By
doing this, these manufacturers are applying different winding
configurations and saving a few percent of manufacturing costs.
This trend has an impact on motor protection. With the neutral
point buried in the stator, CTs cannot be installed on the neutral
side of the motor stator winding and motor differential protection cannot be applied, even though it is considered the most
reliable and fastest protection for motor internal faults.
From the protection point of view, a large motor can be
affected by the connected load, the environment, the electrical
supply, and internal failures. A list of the common failure
modes and corresponding protection methods for a large motor
is shown in Table II. A graphical presentation of a typical
protective relay system designed for a large motor is shown
in Fig. 9, using the standard method of relay and protection
designations common in the literature, as listed in Table III [2].
The latest microprocessor-based relays incorporate all of
these protective devices, and additional control and monitoring
features, into one package. Many additional features are included to provide alarms or shutdowns, to block some functions
during certain conditions, or to monitor important parameters,
which may indicate damage to the motor. These include thermal
capacity limit, vibration, underpower condition (loss of load),

1710

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

TABLE IV
TRANSFORMER FAILURES AND PROTECTION

Fig. 9.

Motor protectiondiagram.
TABLE III
PROTECTIONDEVICE NUMBERS

Fig. 10.

acceleration limit, torque limit, short or open RTD, starts per


hour limit, bearing overtemperature, power-factor limit, and
speed (revoultions per minute) limit. Some relays also add
logic to provide an alarm on demand level, to supervise a trip
coil, and to provide monitoring for breaker failure.
It is not unusual to find a captive motortransformer system
protected by a single motor protection relay. The big dilemma is
where to install CTs and PTs for this protection. The relay manufacturer normally advises to install them on the motor terminals,
as this protection is dedicated to the motor only and was designed for this purpose. When the relay is applied in this way, it
gives accurate metering for all motor data, but does not give any

Transformer protectiondiagram.

level of protection to the upstream transformer. However, for a


captive motortransformer configuration, it is more beneficial
to install the CTs and PTs on the high-voltage side of the transformer. This option requires more study. The relay is protecting
both the motor and transformer together. They could be considered as a hypothetical equivalent motor with different parameters for the overload and thermal capacity. Once the relay is set
up properly, it would protect both the motor and the transformer.
The metering and monitoring information from the hypothetical
motor that is available from the relay can be very useful.
C. Transformer Protection
The transformer is usually of secondary importance in
large captive motortransformer configurations, because it is
generally cheaper than its dedicated motor. The most common
configuration for a captive transformer is with the primary or
high-voltage winding in delta connection, and the secondary

DOAN et al.: DESIGN AND PROTECTION OF CAPTIVE MOTORTRANSFORMERS

or motor side winding in wye connection. The secondary is


usually grounded either with low or high resistance.
A transformer is also affected by the connected load, the environment, the electrical supply, and internal failures, but the
modes of failure are different than with the motor. A list of the
common failure modes and corresponding protection methods
for transformers is shown in Table IV. A graphical presentation
of a typical protective relay system designed for a transformer
is shown in Fig. 10. The protection device numbers are included
in Table III.
Microprocessor-based relays are available for transformer
protection, incorporating these functions, and more, in one
package. Some of the additional features included in a transformer protection relay are second and fifth harmonic inhibit
during energization, underfrequency and overfrequency, harmonic measurement, negative-sequence overcurrent, total
harmonic distortion (THD), current demand, and restricted
ground fault (87RGF). The 87RGF is a separate differential
protection only for wye transformer windings and is useful in
determining the location of internal transformer faults.
The two most important factors from the protection point of
view are transformer energizing and motor starting. Similar to
motor protection, the most desirable, most reliable, and fastest
transformer protection is differential protection.
It is common to find a transformer used in a scheme where
the downstream load has its own dedicated circuit breaker. In
this case, the transformer is energized first, and the downstream
load would be connected only after transformer energization.
Also, the transformer would be deenergized only after the load
has been removed. In this case, it is very common to find transformers protected with high-voltage side fused switches only.
These transformers have only thermal, or overcurrent protection.
D. Protection of Captive MotorTransformer Configuration
A fully developed protection scheme designed as described
above with separate protection for the transformer and motor is a
complex system. The main issues are cost and space limitation.
In most applications, designs are used that combine or reduce
this protection to eliminate some duplication.
One cannot expect to save on installation costs by eliminating
protection elements from a multifunctional integral relay. Some
savings can be expected if one relay is used instead of two for
the total captive motortransformer configuration. The savings
are again small, as the relay cost is usually in the range of twothirds of the cost for one set of CTs or PTs, and in the range of
one-tenth or less of the cost of one circuit breaker.
The largest savings can be realized if one common circuit
breaker is used for both transformer and motor. This circuit
breaker can be installed upstream, on the high-voltage transformer side. The large captive transformer configuration would
be protected in this case with two independent multifunctional
integral relays, one for the motor and the other for the transformer. This is the most desirable scheme from the protection
point of view, as practically nothing is sacrificed and one circuit
breaker is saved.
Also, by applying two integral relays, selectivity is saved. The
piece of equipment in fault and the cause of the fault will be

1711

Fig. 11.

Suggested protection.

obvious, and only the protection of the faulted equipment will


act. From the protection point of view, the troubleshooting and
downtime are minimized. Even more, this scheme allows the use
of one set of CTs and PTs for both multifunctional relays, as the
burdens of these relays are very low. This configuration has approximately the same reliability as the configuration with two
circuit breakers. It is possible to add ground-fault protection of
the secondary transformer winding if there is an available CT in
the transformer neutral. The restricted ground-fault differential
(87RGF) can also be utilized. These elements are usually available on any integral protection for transformers for negligible or
no extra cost. Fig. 11 presents a suggested solution.
A system with one breaker may be desirable from the voltage
drop point of view. This will be a softer start than in the case
when two circuit breakers are employed. It has been observed
that negative-sequence currents and current unbalances are
higher in the case when the transformer and motor are started
together. It might be necessary to increase some pickup values,
such as motor acceleration time, or block some protection
functions during the motortransformer start to compensate for
some of these considerations.
Another configuration, already mentioned under transformer
protection, has the transformer protected with a primary fused
switch, and the motor fully protected with a dedicated motor
protection relay. The advantage of this scheme is simplicity,
however, it still has two switching devices as compared with
the previously considered scheme with only one circuit breaker.
There are savings of one set of CTs on the high-voltage transformer side, and savings of one transformer protection relay,
but the additional fused switch will offset these savings. The

1712

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

disadvantage of this scheme is that the transformer has only


thermal/overcurrent protection.
The final configuration to consider as mentioned under motor
protection is where the motor protection is used to protect the
whole captive motortransformer configuration. This scheme
requires the CTs and PTs only on the high-voltage side of the
transformer. The advantages of this scheme are the cost of installation, the occupied space, as only one relay is used, and
only one set of CTs and PTs. This configuration provides full
motor protection, and also protects the transformer with instantaneous and time overcurrent elements. Thermal modeling in
this case should include the thermal capacity of both the motor
and the transformer. The relay must be set up to treat the entire
motortransformer configuration as one hypothetical equivalent
motor connected on the high-voltage side, when, in reality, only
the transformer is connected there.
The use of this scheme may raise some tough questions. Because of the transformer inrush and motor starting current combined, the instantaneous current element pickup values must be
set very high. Motor protection usually does not have filters of
second and fifth harmonics, which are characteristic of transformer inrush. These are usually found on transformer protection relays. It is also difficult to properly protect the transformer
with the time-delay overcurrent element, due to the different
relay design, which is primarily intended for motors. Also, there
would be no selectivity of protection, so fault locating and troubleshooting will become more difficult. As there are usually no
removable connections between the motor and transformer, it
will take a longer time to disconnect the motor from the transformer in order to troubleshoot the fault. Once they are disconnected, both have to be fully tested in order to find the fault.
In the future, manufacturers may consider the need for a combined relay that provides the required motor protection, but also
adds harmonic restraint and other functions normally required
for a transformer. Although the user would lose the functionality
of separately troubleshooting the motor and transformer, there
could be additional cost savings.
E. Existing System Example
A recent startup effort centered around a large plant chiller
system. The chiller motor, a 3000-hp 2.4-kV induction motor,
is powered by a dry-type 3750-kVA transformer in a captive
motortransformer configuration. The single-line diagram of
this system is shown in Fig. 12.
Protective relay functions of instantaneous and time-delay
overcurrent, RTD thermal shutdown, undervoltage and
overvoltage, ground-fault definite-time overcurrent, and
transformer secondary ground overcurrent were installed.
The ground-fault definite-time overcurrent function used a
residual connection of the three main CTs at the load side of
the vacuum breaker. The intention was to use a residual CT
connection and save on one open-window ring-type CT. The
lower accuracy of this connection was offset with the fact
that zero component current would be expected only during
transformer energization. The secondary ground overcurrent
was wired to an alarm only, since it was expected that
higher ground currents would be present in the secondary,
especially during motor starts. This system is high-resistance

Fig. 12.

Existing system example.

grounded, and is allowed to continue in operation after the


first ground fault. The operator can start up other equipment
before shutting down the chiller. The secondary ground-current alarm level was set at 3.5 A.
Some issues of protection were encountered during startup
that exemplify the flexibility of these relays in special applications. This example also points out the issues of using a motor
protection relay for a captive system. The relay settings were
adjusted to the required calculated values. The entire system
was inspected, and the transformer taps were set properly for
the expected input voltage. The definite-time settings for the
ground fault function were 0.1 s and 24 A. At the first start,
the ground-fault function of the relay activated. The maximum
value of ground-fault current during the 0.1-s period was 76 A,
as displayed by the relay. The phase currents were all near 700
A during the period.
The startup team tested the cables, transformer, and motor
for any ground faults and found no problems. At the second
start, the ground-fault function activated again, with a maximum
ground current of 93 A. The team felt that the transformer inrush current, rich in harmonics, could cause an unbalanced current for a short time and act like a ground fault to the residually
connected ground-fault measuring circuit. The definite-time setting was changed to 0.3 s, with the current level of 24 A remaining unchanged. The third start was good, and the motor
ran for 1 min, before being turned off remotely for a mechanical issue. The maximum ground current measured during the
good start was 35 A, but since it did not stay above 24 A for the
0.3 s, the relay did not trip.
After the mechanical problem was cleared up, the motor was
started again. The ground-fault function activated again, with a

DOAN et al.: DESIGN AND PROTECTION OF CAPTIVE MOTORTRANSFORMERS

1713

IV. CONCLUSIONS
A. Transformer Selection and Sizing

Fig. 13.

Ground coordination of example system.

maximum measurement of 162 A. Clearly, the variation in inrush current between the phases was high enough that the definite-time relay function would not be adequate.
One suggestion was that a window CT around all three conductors would give a more accurate measure for the ground-fault
current, in place of the residual connection. This is due to the
decaying asymmetrical components (X/R was roughly 50) and
the remmanence in the CT cores. It was believed that the transformer starting currents put the existing three CTs in temporary
dc saturation.
The team concluded that the transformer inrush current was
tripping the ground-fault function. Fig. 13 is a coordination diagram of the definite-time ground-fault relay and the upstream
bus protection curve that it must coordinate with. A possible solution was to use the second ground-fault function of the motor
protection relay. This second ground-fault function is normally
used as an alarm, but can be configured as a trip if needed. The
two definite-time ground-fault functions were set to 0.2 s, 122
A, and 9 s, 60 A. Fig. 13 also shows the final settings. These
values were chosen to allow time for the motor to start with the
compressor load, with an estimated duration of 7 s.
One thing learned from this startup is that the flexibility of the
relay allowed the team to set the ground-fault protection at an
acceptable level. An issue is that a transformer protection relay
would have been better to work with the inrush and harmonic
currents. Full protection of this system would require two relays
with the required motor and transformer protection functions, as
discussed above.

Selecting the proper rating and characteristics for a captive


transformer of a large motor are extremely important to the reliability of a manufacturing process. The captive transformer and
motor may serve a variety of different loads and operate continuously or experience multiple starts or pulses. The type of motor
load, environmental conditions, voltage drop, and desired energy efficiencies are important engineering factors that must be
considered when selecting and sizing a captive transformer.
Liquid-filled transformers typically are the lowest cost and
provide excellent service for outdoor applications and with approved high fire point liquids for indoor applications. Liquidfilled transformers have proven to be suitable for motor starting
and pulse loading when sized and designed properly. Very high
efficiencies can be obtained with liquid-filled transformers. Dry
transformers can be designed with disc windings and clamps to
resist the axial forces during motor starting. However, repetitive motor starts will tend to loosen the clamping structure and
continual preventive maintenance will be required to retighten
clamps for winding integrity. Dry transformers are restricted to
indoor applications and tend to have lower overall energy efficiencies than liquid or cast coil.
Cast-coil transformers have a higher cost than liquid or dry
units but offer significant advantages when used as a captive
transformer for starting large ac motors. Since the windings of a
cast-coil transformer are encapsulated in epoxy, the conductors
are unaffected by the axial and radial forces from inrush currents
during motor starting. The 80 rise cast-coil transformer offers benefits from improved efficiency and energy savings over
liquid or dry transformers. Also, the cast-coil transformer can be
used in harsh environments either inside a building or outdoors.
Regardless of the type of transformer that is used, selecting
the proper kVA rating of the captive transformer is critical
to long transformer life and reliability. The ultimate kVA
transformer rating must be evaluated based on the type of
loads, number of motor starts, and high current pulses that the
transformer will experience. The maximum transformer kVA
value, for liquid-filled or cast-coil units, should be selected
based on the maximum required from the calculated thermal
kVA, starting kVA requirements, and short-term pulse kVA
requirements.
B. Protection
The most cost-effective system from a protection standpoint
has one primary breaker and two relays. Each relay provides
needed functions, as described above. It may be possible to use
the same set of CTs in the primary of the transformer to provide
the signal needed for both relays.
The intention of the authors was also to highlight some important characteristics of different protection configurations and
quantify the savings versus available features in different cases.
This can be a useful tool at the preliminary stage when different
designs and different layouts are analyzed and compared with
the available budget.

1714

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 6, NOVEMBER/DECEMBER 2000

Manufacturers may consider the need for a combined relay


that provides the required motor protection, but also adds harmonic restraint and other functions required for a transformer.
In this way, more cost savings may be achieved.
ACKNOWLEDGMENT
The authors thank G. Miranda (GE Industrial Systems),
M. Martin (Equistar Chemicals), and D. Doughty (E. I. duPont
Company) for their help in reviewing and commenting on this
paper.
REFERENCES
[1] F. J. McCann and R. J. Ristow, Problems of Impact Loading on Unit
Transformers. Rome, GA: GE Industrial Power Systems, June 1970.
[2] Applied Protective Relaying, Westinghouse Electric Corporation, Coral
Springs, FL, 1982.
[3] N. K. Haggerty, T. P. Malone, and J. Crouse, Justifying use of high
efficiency transformers, presented at the IEEE PCIC, Sept. 1996, Paper
PCIC-96-23.
[4] B. W. Kennedy, Energy Efficient Transformers. New York: McGrawHill, 1998.
[5] E. F. Byars and R. D. Snyder, Engineering mechanics of deformable
bodies, West Virginia Univ., Morgantown, WV, Feb. 1969.
[6] D. G. Fink and H. W. Beaty, Standard Handbook For Electrical Engineers. New York: McGraw-Hill, 1993.

Daniel Doan (S81M81) received the B.S.E.E.


and M.S.E.E. degrees from Massachusetts Institute
of Technology, Cambridge.
He is currently a Senior Consulting Engineer in
electrical technology with E. I. duPont de Nemours
& Company, Inc., Wilmington, DE. He has been
engaged in project engineering and consulting on
drives, power, and electrical safety for duPont for 17
years.
Mr. Doan is a Licensed Professional Engineer in
the State of Pennsylvania.

John Crouse (S62M70) received the B.S.E.E.,


M.S.E.E., and Ph.D. degrees from the University of
Tennessee, Knoxville.
He is currently a Senior Engineer in transformer
technology with General Electric Company, Rome,
GA. He has held various design, development,
and managerial positions with Westinghouse and
General Electric. He has been mainly involved
with small, medium, and large power transformers.
He has also been involved with power systems
engineering studies and pulse power systems. He is
the holder of three patents for improved transformer winding designs.

N. Kent Haggerty (M76SM99) received the


B.S.E.E. degree from the University of South
Carolina, Columbia.
He is currently a Consultant in electrical technology with E. I. duPont de Nemours & Company,
Inc., Wilmington, DE. He was with Duke Power
and Dow Badische prior to joining duPont. His
experience of 28 years has included responsibility
for electrical transmission system and distribution
facilities in a variety of positions in supervision,
project design, and consulting.
Mr. Haggerty is a Registered Professional Engineer in the State of South Carolina and is currently a member of the IEEE Working Committee for Transformers.

Miroslav Ristic (M98) received the B.Sc.E.E. degree from the University of Novi Sad, Novi Sad, Yugoslavia.
He is currently an Applications Consultant with
GE MultiLin, Markham, ON, Canada. Prior to this,
he was an Applications Engineer with GE Large
Motors, where he was involved in applications
with large motors and industrial generators. He was
with the Electric Power System in Yugoslavia for
more than ten years. He has served on the Board of
Electric Power Transmission Network of Yugoslavia
and the Board of Power Stations and Electricity Production of Yugoslavia.
Mr. Ristic is a Licensed Professional Engineer in the Province of Ontario,
Canada.

S-ar putea să vă placă și