[James_H._Harlow]_Electric_Power_Transformer_Engin(BookSee.org)

considerably; furthermore, each material interface further attenuates the propagated signal. Therefore,
acoustic signals can only be detected within a limited distance from the source. Consequently, the
sensitivity for PD inside transformer windings, for example, may be quite low. In typical applications,
many acoustic sensors are carefully distributed around the tested equipment (Eleftherion, 1995;
Bengtsson et al., 1997).
Though not disturbed by the electric network, external influences in the form of rain or wind and
non-PD vibration sources, like loose parts and cooling fans, will generate acoustic signals that interfere
with the PD detection. One way to decrease the external influence is to use acoustic waveguides (Harrold,
1983) that detect signals from inside the transformer tank. This solution is typically only considered for
permanent monitoring of important transformers. As an alternative, phase-position analysis can be used
to reject these disturbances (Bengtsson et al., 1997).
A transformer generates disturbing acoustic signals in the form of core noise, which can extend
up to the 50 to 100-kHz region. To diminish this disturbance, acoustic sensors with sensitivity in
the 150-kHz range are usually employed (Eleftherion, 1995). Such sensors may, however, have less
sensitivity to PD signals as well (Bengtsson et al., 1997). The properties of these signals are such
that it is relatively easy to distinguish them from PD signals; thus, their main effect is to limit
sensitivity.
Regarding the electric multisensor systems discussed here, there are a few descriptions of combined
electric and acoustic PD monitoring systems for transformers in the literature (Wang et al., 1997). Rather
elaborate software must be employed to utilize the potential sensitivity of these systems. If both the
acoustic and the electric parts are designed with the considerations above in mind and an effective
software constructed, systems like this will become effective yet costly.
3.13.3.1.4 Oil Temperatures
Overheating or overloading can cause transformer failures. Continuous measurement of the top-oil
temperature is an important factor in maximizing the service life. Top-oil temperature, ambient temperature,
load (current), fan/pump operations, and direct readings of winding temperatures (if available)
can be combined in algorithms to determine hottest-spot temperature and manage the overall temperature
conditions of the transformer.
3.13.3.1.5 Winding Temperatures
There is a direct correlation between winding temperature and normally expected service life of a
transformer. The hottest-spot temperature of the winding is one of a number of limiting factors for the
load capability of transformers. Insulation materials lose their mechanical strength with prolonged
exposure to excessive heat. This can result in tearing and displacement of the paper and dielectric
breakdown, resulting in premature failures. Conventional winding temperature measurements are not
typically direct; the hot spot is indirectly calculated from oil temperature and load current measurements
using a widely recommended and described test method (Domun, 1994; Duval and Lamarre, 1977; Feser
et al., 1993; Fox, 1983; IEEE, 1995). Fiber-optic temperature sensors can be installed in the winding only
when the transformer is manufactured, rebuilt, or refurbished. Two sensor types are available: optical
fibers that measure the temperature at one point, and distributed optical fibers that measure the temperature
along the length of the winding. Since a distributed fiber-optic temperature sensor is capable
of measuring the temperature along the fiber as a function of distance, it can replace a large number of
discrete sensors and allow a real-time measurement of the temperature distribution.
3.13.3.1.6 Load Current and Voltage
Maximum loading of transformers is restricted by the temperature to which the transformer and its
accessories can be exposed without excessive loss of life. Continuous on-line monitoring of current and
voltage coupled with temperature measurements can provide a means to gauge thermal performance.
Load current and voltage monitoring can also automatically track the loading peaks of the transformer,
increase the accuracy of simulated computer load-flow programs, provide individual load profiles to
assist in distribution-system planning, and aid in dynamically loading the transformer.
FIGURE 3.13.2 Power-factor graphical representation.
3.13.3.1.7 Insulation Power Factor
The dielectric loss in any insulation system is the power dissipated by the insulation when an ac voltage
is applied. All electrical insulation has a measurable quantity of dielectric loss, regardless of condition.
Good insulation usually has a very low loss. Normal aging of an insulating material causes the dielectric
loss to increase. Contamination of insulation by moisture or chemical substances can cause losses to be
higher than normal. Physical damage from electrical stress or other outside forces also affects the level
of losses.
When an ac voltage is applied to insulation, the leakage current flowing through the insulation has
two components, one resistive and the other capacitive. This is depicted in Figure 3.13.2. The power
factor is a dimensionless ratio of the resistive current (I R ) to total current (I T ) flowing through the
insulation and is given by the cosine of the angle depicted in Figure 3.13.2. The dissipation factor, also
known as tan delta, is a dimensionless ratio of the resistive current to the reactive current flowing through
the insulation and is the tangent of the angle in Figure 3.13.2. By convention, these factors are usually
expressed in percent. Due to the fact that theta is expected to be large, usually approaching 90 degrees,
and delta is commensurately small, the power factor and dissipation factor are often considered to be
essentially equal.
3.13.3.1.8 Pump/Fan Operation
The most frequent failure mode of the cooling system is the failure of pumps and fans. The objective of
continuous on-line analysis of pumps and fans is to determine if they are on when they are supposed to
be on and are off when they are supposed to be off. This is accomplished by measuring the currents
drawn by pumps and fans and correlating them with the measurement of the temperature that controls
the cooling system. This can also be accomplished by measuring pump/fan current and top-oil temperature.
Mode of operation is verified based on current level. Normal operational modes indicate rotation
of fan blades and correct rotation of pump impeller. Abnormal operational modes are usually the result
of improper control wiring to those devices.
Pump failures due to malfunctioning bearings could be a source of metallic particles, and such particles
could be a potential dielectric hazard. Sensors that detect bearing wear are available. The ultrasonic
sensors are embedded in the pump bearings and measure the bearing length, thus determining whether
metal loss is occurring.
Furthermore, continuous on-line analysis should take into account that:
• The temperature that controls the cooling system can differ from the temperature measured by
the diagnostic system.
• The initial monitoring parameters are set for the cooling stages based on the original transformer
design. Any modifications to the cooling sequences or upgrades must be noted, since this will
change the monitoring system output.
• The sensitivity of the diagnostic system is influenced by the number of motors that are measured
by each current sensor.
© 2004 by CRC Press LLC
© 2004 by CRC Press LLC