Partial discharges or pd’s (known in the past as corona), are spark charges that occur in voids within high-voltage insulation (>5 kV). They occur between the windings and core, or in the end winding region. These are “partial” discharges because there is some remaining insulation. The pd can erode the insulation and therefore contributes to its aging. However, a pd is a symptom of insulation aging caused by thermal or mechanical stresses. The measurement of a pd activity in a stator winding is an indication of the health of the insulation. Partial discharge tests provide the best means for assessing the condition of the insulation without a visual inspection. These tests should be done on stator windings in motors and generators rated higher than 5 kV.
Off-Line Conventional Partial Discharge Test.
The conventional pd test involves energizing the winding to normal line-to-ground AC voltage with an external supply. A pd detector is used to measure the pd activity in the winding. The sparks caused by pd are fast current pulses that travel through the stator windings. These pulses and the accompanying voltage pulses increase with the pd pulse. Figure 1 illustrates a high-voltage capacitor that can block the power frequency voltage and allow the high-frequency pulse signals to reach the pd detector. An oscilloscope is used to display the pulse signals after further filtering.
The pulse magnitudes are calibrated in picocoulombs (pC), even though the actual measurements
are in millivolts (mV). The conventional test is done off-line. A separate voltage supply is used to energize the windings to normal voltage. The interference from high-frequency electrical noise in this test is a minimum.
The conventional test involves isolating the winding from the ground and energizing one phase of the winding by a 60-Hz power supply cable to rated line-to-ground voltage. This test is normally done on each phase separately while the remaining two are grounded. The phases are disconnected from one another at the neutral.
Draining of the water-cooled winding is not required.
The test equipment includes a power separation filter (high-voltage) capacitor and a high-pass filter to block the power frequency and its harmonics—Fig. 1. The oscilloscope displays the pd pulses (Fig. 2). A pulse height analyzer is used to process the pulse data. It gives the pulse counts, pulse magnitudes, and comparisons between positive and negative pulses.
The AC voltage is raised gradually until pd pulses are observed on the oscilloscope. The voltage at which pd starts is called the discharge inception voltage (DIV). When the test voltage reaches the normal voltage, the magnitude of the pulses is read from the screen.
The analysis of the pulse height is normally recorded. As the AC voltage is decreased, the voltage at which the pd pulses disappear is recorded. This is called the discharge extinction voltage (DEV). It is usually lower than the DIV. The actual test takes about 30 min, normally.
However, the setup and disassembly can take up to a day.
There is no general agreement on the acceptable magnitudes of pd, DIV, and DEV. The inductive nature of the windings makes the calibration of the measured pd magnitudes (conversion from millivolts on the screen of the oscilloscope into picocoulombs) difficult. Thus, the measurement of the pulses may not provide an accurate value of the pd activity. These measurements cannot be calibrated from machine to machine or among the different types of commercial detectors.
The most useful method for interpreting the pd test results is by performing the test at regular intervals and monitoring for trends. The recommended interval for air-cooled machines is once or twice per year and every 2 years for hydrogen-cooled generators. As the condition of the insulation worsens, the magnitude of the pd will increase and those of the DIV and DEV will decrease. An increasing trend of pd activity indicates that the insulation is aging. Visual inspection of the winding condition may be required. Partial discharge results should only be compared if the same equipment and procedures are used during testing.
This is due to the calibration problems mentioned earlier. Comparison of results can also be misleading if there are differences between the types of rating of the windings. Comparison of pd results are valid if the windings and test methods are identical.
A pd magnitude of less than 1000 pC indicates that the winding should not fail during the next few years. A visual inspection is recommended if the pd magnitude is more than 10,000 pC, especially if other identical machines have a pd less than 1000, and if the insulation is made of epoxy-mica. The DIV in modern epoxy or polyester windings should be greater than half the operating line-to-ground voltage. The test indicates that slot discharge is occurring if the DIV value is very low in epoxy-mica windings. However, older asphaltic and micafolium windings may not be in danger even if the magnitude of the discharge is high and the DIVs are low. This is in contrast with newer machines that have synthetic insulation, especially Mylar. Their condition deteriorates quickly in the presence of pd. However, older windings should be inspected if there is an increasing trend of pd activity.
There are many disadvantages for off-line conventional pd tests. Since the entire winding, including the neutral end, is fully energized, sites that are not normally analyzed can generate pulses. Large discharges can occur in sites that are not normally subjected to high voltages.
This is misleading because the operator may believe that the winding is deteriorating.
On-Line Conventional pd Test.
This test is similar to the off-line test except that an external power supply is not used to energize the winding. The generator is driven at normal speed by the turbine, and sufficient field excitation is applied. Therefore, the stator is at the normal operating voltage. The test can be performed with the generator synchronized to the grid or not. When the test is done on a motor, the winding is energized by the normal power supply. Extreme caution is required when the test is performed due to the considerable risk to personnel and the machine if the capacitor fails.
The test is more realistic than the off-line one because the voltage distribution in the windings is normal. Also, slot discharges that are caused from bar or coil movement are present.
The equipment used in the off-line test can be used in this test. The blocking capacitors are connected to the phase terminals during an outage. Dangerous events can occur if the capacitor fails during the test. An experienced operator can distinguish true pd from electrical interference from brushes, thyristor excitation systems, and background. If the generator is not synchronized, some generators can handle variations in the field current. In these cases, the DIV and DEV can be measured.
Some utilities leave the test equipment connected during normal operation. The pd activity can then be measured at low and full power. Deterioration in the condition of the insulation is detected by an increasing trend of pd. Since the test is done during normal operation, it gives the most accurate indication of the true condition of the insulation.
External interference (from a power line carrier, radio station, etc.) can be severe during the test, especially in large generators. The interference can be misleading. The operator may believe that high pd activity is occurring while the winding is perfectly good.