Partial Discharge Testing
Partial Discharges Measurements in Rotating Machines
Equipment Used By A. C. Hargreaves & Co.
LDIC – DOBLE
FULLY PORTABLE PARTIAL DISCHARGE DETECTOR
TYPE: PD SMART
Partial discharges are breakdown phenomena, which do not completely bridge the distance between conductor electrodes. Partial discharges produce high frequency pulses originating at various sections within an insulation system. The three (3) main types of partial discharge activity are:
- Internal discharges (Voids)
- Surface discharges
- Corona
The spectrum of damages due to partial discharge activity includes heating, oxidation, chain scission of polymer molecules, delamination, stress cracking due to UV light, surface erosion and microscopic cracks. External discharges also produce aggressive gases such as Ozone, which can cause the above damages but any possible Partial Discharge activity also reflects the insulation condition of motor and generator windings in operation. Detection of these discharges can identify insulation deterioration long before traditional methods of insulation testing can. It is the only test method that allows locally confined insulation problem to be detected and identified while the unit is in service by using on-line testing techniques.
Most high voltage plants having gas, epoxy or oil-paper based insulation can only tolerate very low levels of discharge activity. Mica based insulation systems used in rotating machines, however, are capable to withstand high levels of discharge activity occurring in the voids of the insulation without apparent damage. In a rotating machine there can be numerous possible defects causing partial discharge activity. Some of the main defects that cause partial discharge activity are:
- Ground wall insulation delamination and/or internal voids often caused by thermal stress.
- Slot discharges between voids at the interface of the surface of the coil and the slot wall, often caused by coil/bar vibrations caused by loose wedges resulting in abrasion of the conductive coating.
- Discharges due to voids close to the copper conductor.
- End winding discharge activity due to contamination or insufficient spacing between coils of different potential.
- Discharges due to deterioration of the semi-conductive coating where the coils exit the iron often caused by contamination or aging.
- Surface discharges caused by contamination.
All stresses act as a “TEAM” on the insulation system
These stresses are:
- Thermal
- Electrical
- Mechanical
- Environmental
Very often “Partial Discharges” indicate the above stresses, they are often not the cause but the result of the problem. These defects often occur in specific locations and there is no overall average ageing.
Partial Discharge patterns represent the best way to detect and identify the type of defects; single weak points in the insulation system are indicated by partial discharges.
Conventional test methods such as Tanδ are sensitive to partial discharge, but they only represent an integral value as the loss created by the partial discharge pulses must be very high.
Partial discharges are plasma processes which take place generally in gas. Whenever the voltage in a gas space exceeds the inception voltage of the gas, partial discharges are created on the surface of an insulator, due to conductive contamination or air inclusions within the insulation as a result of mechanical forces. The gas plasma accelerates the ageing and corrosive substances as ozone and NO2 are created. The mechanical wear or a surface contamination is usually more harmful than the partial discharge action. As the presence of partial discharges cannot be avoided in rotating machines insulation, mica is applied to the coils as a partial discharge resistant medium.
By products of partial discharge are commonly white powders which can be found in many locations of the machines windings.
Typical Partial Discharge Phenomena in Rotating Machines
- Loose Wedges – Causing vibration of the coils and consequently slot discharges.
- Cracked and/or broken insulation.
- End winding surface contamination.
- Damage and/or corrosion of the corona suppression system.
- Discharge on the connection ring due to vibration.
- Inter-Phase discharge
- Insulation degradation.
Partial Discharge Locations:
1) Within the coil (internal) these can be caused by any of the following:
- Delamination of the main insulation
- Delamination of the layers of tape
- Voids with in the insulation system
- Treeing effect
2) Slot discharges – caused by damage and/or abrasion to the corona protection coating on the coils.
3) Discharges within the vent ducts – mainly caused by vibration and/or abrasion of the corona
protection layers.
4) Delamination of insulation on bends of coils.
5) End winding surface discharge – mainly caused by surface contamination.
6) Tracking between coils or bars with insufficient space. (Especially coils or bars with high voltage
Differences – different phases)
7) The connection area between the slot corona protection and end winding corona protection.
If an AC voltage is applied across the terminals of a model a voltage develops across the void capacitance. If the voltage is increased and the electrical breakdown strength of the gas medium inside the void is exceeded, the gap discharges, the potential across the void falls to zero and the discharge extinguishes.
As the applied voltage is still present and increasing, the discharge cycle will repeat itself so that a pattern of recurring discharges can be observed. This pattern is repeated during each half period. This collapse of voltage causes a very short current pulse to flow in the circuit and a subsequent voltage dip across the test sample terminals. It is not practical to detect this very small voltage dip directly. A sensor (blocking capacitor), the detection impedance in series with the capacitor and the detector itself make up the measurement circuit. A discharge in the void of the test object causes a charge transfer from the external blocking capacitor to the test sample. The charge current through the capacitor and the detection impedance is integrated, amplified and fed to the detector. The charge contained in this current pulse is the parameter measured and is expressed in pico coulomb (pC).
The blocking capacitor represents high impedance for the 50 Hz supply voltage and low impedance to the high frequency PD pulses. In some circuit configurations, the capacitor sensor is replaced with a HF current transformer or a Rogowski coil. There are also some more exotic coupler circuits using embedded stator slot couplers or systems which derive PD signals from the RTD’s used to measure the coil temperature. Each type of coupler has its own characteristics.
Analysis of PD Results
Analysis of PD test results is based on the following main indicators:
- Trends
- PD trace signatures as well as PD inception and extinction voltages
- Comparison between phases or readings from similar machines.
- Correlation with off-line and other tests such as DDF tip up as described earlier.
- Change in PD signature and amplitude for different operating conditions such as high and low loads for on-line test.
- Findings from visual inspection that may support PD results such as looses wedges or white powder deposits on the windings.
Calibration Issues
Normally the detector and circuit are calibrated as per AS1026 or IEC270 by injecting pulses with known charge amounts across the test object and setting the pC indicator of the detector to read this amount. Although this method is true for most conventional test circuits, the calibration of a stator is more complicated as the complete winding cannot be treated as a lumped capacitor. A stator winding
is a complex network with distributed inductance, capacitance and resistance. When a PD propagates
through the stator winding, the stator winding attenuates the PD pulse, particularly at high frequencies. In addition pulse ringing resulting in pulse superposition error in the detection circuit can complicate measurements further. This means that discharges measured at the stator terminals via the coupling capacitor may not accurately represent activity occurring within the winding itself.
However, it does provide a scale normalisation and a reference point for subsequent PD measurements and comparisons of similar machines. The calibration also verifies the correct installation and function of all measurement components.
Because of these limitations, different PD detectors with different frequency bandwidths and centre frequencies may produce different apparent pico-Coulomb (pC) values for the same PD event and make comparison without standardising parameters of the detector, calibration routines and couplers, etc., difficult. PD magnitude is expressed either in pC, mV or dB.
As discussed above, Partial Discharge activity can, depending on failure mode, generate pulses ranging from DC to > 100 MHz.
Measurements using the low frequency range (100 – 800 kHz) are able to look further into the winding and discharge activity of the whole winding is detected but they are more affected by external interference and superposition of pulses due to pulse oscillations. Measurements using high frequency range (2 – 20 MHz) can, due to signal attenuation only detect discharges in an area close to the HV terminals but the measurement is more noise immune. We therefore, carry out measurements with the above mentioned two bandwidths. Also comparing the attenuation between the high and low frequency measurement can be helpful in determining the location of insulation deterioration.
Copyright© A.C. Hargreaves & Co. 2008 - 2010
Site managed by Leo Mylonas