Insulation and dielectric
Offline partial discharge testing energizes de-energized cable or a stator winding from a separate source and listens for tiny electrical discharges inside the insulation. It locates weak spots such as voids, cracks and damaged splices before they grow into a failure.
Offline partial discharge testing is an electrical diagnostic that looks for small, localized discharges inside or on the surface of solid insulation without letting the insulation break down. The equipment is taken out of service, disconnected from the system and energized from a separate test source, so the voltage can be raised, held and lowered under control. Sensitive detection circuits then pick up the brief current pulses that each discharge produces. The test is applied to medium- and high-voltage shielded power cable and its splices and terminations, and to the stator windings of large motors and generators, where it is listed as an optional measurement in the acceptance and maintenance standards. The result is a set of measurements: the voltage at which discharges start, the voltage at which they stop, how large and how frequent they are, and where along the cable or in which winding phase they originate. Because it works on the equipment while it is isolated and under controlled voltage, it can be performed in the same outage as the other insulation tests and compared with the same unit from earlier years.
Solid insulation seldom fails suddenly. Small voids left in extruded cable or in a stator groundwall, cracks, delamination from thermal cycling, contamination, a poorly made splice or a damaged termination all concentrate the electric field and produce partial discharges. These discharges slowly erode the insulation around them, and in cable a growing defect eventually forms an electrical tree that ends in a breakdown. Consequences range from a feeder outage and a long search for the fault, to a destroyed motor winding and a lost process. Offline testing finds the defect while the equipment is still operating normally, so the repair can be planned. At acceptance it confirms that factory manufacture, shipping, installation and the making of splices and terminations were done properly. In maintenance it separates equipment that is aging normally from equipment with an active local defect. It also supplements resistance and loss measurements, which describe the insulation as a whole and can miss a small isolated flaw that is nevertheless the one that will fail.
A partial discharge is a breakdown of a small gas-filled void or a surface gap that only partly bridges the insulation between conductors. The gas in a void withstands less electric stress than the surrounding solid, so once the voltage across the void reaches a threshold the gas ionizes and a small current pulse flows. Each event is brief, on the order of a very short fraction of a microsecond, and its size is expressed as an apparent charge. The pulse travels along the cable or winding in both directions and reaches the measuring points attenuated, which is why the reading is called apparent. As the applied voltage is raised, discharges begin at an inception voltage and, on the way back down, cease at a lower extinction voltage. The pattern of when in the alternating cycle the pulses occur reveals the kind of defect, because internal voids, surface discharge and corona each have a distinct phase signature. In cable, the arrival time of a pulse and its reflection at the cable end lets the instrument calculate the distance to the defect. In a machine, the coupling devices at the terminals pick up discharges from the winding as a whole, and the phase pattern, not the position, is the main clue.
The cable or machine is de-energized, isolated and grounded under the owner's lockout and permit procedures, and all connected apparatus such as transformers, surge arresters, switchgear and motor leads are disconnected so that only the item under test is energized. Terminations are cleaned, and clearances and corona sources are controlled because discharges in the air at an end can mimic a defect. A separate test source is connected, such as a very low frequency supply, a damped alternating-voltage system or a resonant set, selected to suit the cable length and capacitance. A coupling capacitor and a measuring impedance, or a sensor on the shield, are connected to a calibrated detector, and the system is calibrated by injecting a known charge so that readings are meaningful. Background noise is measured with the equipment unenergized and, where necessary, filtered or gated. The voltage is then raised in controlled steps while discharges are watched and recorded, held at the level the procedure and the equipment rating call for, and lowered again, noting the inception and extinction levels. On a machine, each phase is tested with the others grounded and the same method is repeated for comparison. After the test, the equipment is discharged and grounded, and the test environment, temperature and humidity are noted because they affect discharge activity.
The record holds the inception and extinction voltages observed, the discharge magnitude and the repetition activity at the test levels, the phase-resolved pattern, and for cable the location of any defect along its length. It notes the type and settings of the voltage source, the calibration that was performed, the background noise level, the phase tested and the connection arrangement, and the temperature and humidity. The as-found condition of terminations and splices, the identity and calibration status of the instrument and any filtering or gating used are recorded as well. These items are what make a later comparison fair, because discharge readings depend strongly on how the test was set up.
No single number decides whether insulation is acceptable, and this page does not give one. Judgment rests on several observations used together. The first is whether discharge begins inside the working voltage range of the equipment, since discharge present at or below normal operating voltage is more significant than discharge appearing only well above it. The second is the magnitude and repetition of the activity at the test level, read against the instrument's calibration and the background noise. The third is the phase pattern, which separates internal voids from surface activity and external corona. The fourth is the comparison between like parts, such as the three phases of a winding or sections of a cable circuit built the same way. The fifth is trending against earlier results on the same equipment, where a falling inception voltage or rising magnitude is the warning. Manufacturer data and the cable or machine specification also come into play. The optional status of the test in ANSI/NETA ATS-2025 and ANSI/NETA MTS-2023 makes the owner's criteria and the testing agency's judgment important. For the governing language, refer to your own licensed copies of the standards named above.
Acceptance and maintenance tests differ in aim. At acceptance the measurement checks newly installed cable and its accessories, or a new or rewound machine, and the result becomes the baseline. In maintenance it is trended against that baseline, and a change matters more than any one result. Cable and machine testing differ in detail. On cable, the test is often run on the whole circuit including splices and terminations, locating defects by pulse timing, and the choice of voltage source affects the discharge behavior that is observed. On machines, the stator winding is tested phase by phase, usually at the terminals, and the main output is the phase-resolved pattern and the level of activity, because the winding is too compact to locate defects by timing alone. A machine may also be tested while running with online sensors, which is a separate technique with different results. The offline method complements insulation resistance, power factor and tip-up testing, surge comparison on windings and loss measurements on cable, and no one of them replaces the others.
The equipment classes and standard sections that call for this test, and the UltraDb forms that carry it.
It detects small electrical discharges within voids, cracks, delaminations or contamination in solid insulation, or at surfaces and terminations, that do not yet bridge the insulation. These are early signs of local degradation, often invisible to resistance tests, and they can grow into a full breakdown if left alone.
Offline testing de-energizes the equipment and uses a separate controlled source, so voltage can be raised and lowered and inception and extinction levels found. Online testing monitors the equipment while it runs at its normal voltage. The two give different information and are often used for different purposes.
A pulse loses strength as it travels from the defect to the measuring point, so the instrument sees less than the charge actually moved at the site. The apparent charge is therefore a calibrated, comparable indicator, and it depends on the setup, which is why calibration is done with a known injected pulse.
When done within the equipment's rating and the procedure's limits it is considered non-destructive, since the voltage is raised under control and the discharges measured are small. Prolonged discharge on a real defect can age it, so the voltage is limited and the exposure time kept short.
UltraDb references ASTM, ANSI, IEEE, NETA, NFPA, NEMA, FAA, TIA, and IEC standards descriptively, to identify the published methods its forms are built from. UltraDb and Verlecta are not affiliated with, endorsed by, certified by, or licensed by NETA, ASTM, ANSI, IEEE, NFPA, NEMA, FAA, TIA, IEC, or Megger. Standard names and section numbers are the property of their respective organizations. Always test to the edition your contract specifies. Where a standard is sold through the ANSI webstore, Verlecta may earn a commission on a purchase made through our link, at no extra cost to you. A commission never decides which standards we cite or how a form is built.