PARTIAL DISCHARGE MEASUREMENT Partial Discharges are small electrical sparks which occur predominantly at insulation imperfection. It is the phenomenon which occurs in the insulation on application of high voltage. PD is defined as localized electrical discharges in insulating materials, restricted to only part of dielectric under test. It takes place in cavities and voids inside the insulation.
Partial Discharges: Metallic conductor Partial bridging of the Insulation between the conductors Localized electrical discharges restricted to a part of the dielectric void dielectric
Types of Partial discharges Internal discharges Surface discharges Corona discharges Internal discharges lead to deterioration of insulation and treeing treeing 3
1. ACOUSTIC EMISSION TECHNIQUE A.E. sensing is a non-destructive test method, offers good, real time solution to both -PD detection and PD source location. Because it is not affected by strong electric fields, it is well suited to inspection of on-line transformer. Frequency content of these mechanical pulses are generally located in freq band of 100 to 200 KHZ. Acoustic signals (generated by PD) to be detected by multiple channels which provide enough information for location of discharge source.
Partial Discharges: Associated with degradation of insulation system. Pulse-like in nature and cause high frequency (ultrasonic) pressure pulses to propagate through insulation media. Always produce/cause mechanical stress waves (acoustic waves/emission) which propagate through surrounding oil and can be detected at transformer tank wall. Similar in character to stress waves propagated in solids during crack formation.
Block Diagram of AE System Sensors: Piezo-electric transducers. Converts the mechanical signal emanating from PD source into elect. Signal. Sensors are mounted using a silicone vacuum grease as coupling agent between sensor and Transformer tank wall. A constant force, magnetic hold down is used to press sensors against transformer wall. Pre-amplifier: It is necessary not only to amplify the signal but also to lower its impedance. Filter: The signal can then be filtered to exclude erroneous data before further manipulation takes place. Band pass filter of 100 KHZ to 300 KHZ. Superheterodyne Circuit : Part of the signal is fed to Super heterodyne ckt which drops its freq. into audible range i.e. 20 HZ to 20 KHZ. This enables one to literally listen to Partial Discharges. The remainder of the signal is then further amplified in an adjustable amplifier. Threshold: It passes the signal which has amplitude above the threshold value & cuts all the signals below the threshold value. Counter: The number of pulses exceeding this value is totaled in the COUNTER module. This counter is reset at predetermined intervals so that the output to the strip chart recorder is actually pulse rate.
Case Study (1) 200 KVA Distribution Transformer was monitored. AE signal with PCR (Pulse Count Rate) up to 30,000 per second were detected. Source of signals appeared to be in the general area of TAP CHANGER. Tap Changer was dismantled and it was found that arcing had occurred between rollers and contacts causing minor damage. Following repair and reassembly, no further signals were detected.
Case Study (2) 333 KVA Transformer monitored This emitted pulses at the rate of 2000 per second. Monitoring this for several months showed an increasing level of activity until PCR (Pulse Count Rate) rose to 10,000 pulses/second. At that time, Gas and Oil Samples were taken and transformer was taken out of service for more detailed examination.
The large surplus of carbon dioxide (4 times ambient air) and the reduction of oxygen (0.77 times ambient) together with a trace of hydrogen suggests that Cellulose degradation products were present. The lack of any hydrocarbons suggests that the oil is in good condition despite its age (18 years). It is concluded that PDs had been taking place in the Transformer and were associated with Cellulose materials.
Strategies 1. Amplitude (db) vs Time (sec.) 2. Count vs Time (sec) 3. Energy vs Time (sec) 4. Asl (accumulated signal level) vs Time (sec)
Fig. 2 3 -Phase Generator Transformer 160 MVA, 11.5 / 230 KV TOP VIEW OF TRANSFORMER UNDER TEST. Fig. 1 VIEW OF SENSOR POSITION ON DIFFERENT SIDES OF TRANSFORMER
Fig. 3
Location 13
Location K
LOCATION L
Location M
BETN SIDES A & C ( TOP CORNER OF X MER NEAR TAP CHANGER SIDE)
DISCHARGE PATTERN in Electrical method off line: 0 90 180 270 360 360 0 90 180 270 The Display Of Partial Discharges Can Be Observed On An Oscilloscope, Preferably In The Form Of An Elliptical Display Representing A Sine Wave.
EVALUATION/INTERPRETATION ELECTRODE HV CAVITY + 0 180 DIELECTRIC _ a) Internal discharges in a dielectric bounded cavity i) Discharges occur between Zero transition & peak on both half cycles. ii) Discharges are of same amplitude & number on both sides of ellipse. iii) No variation in magnitude with rise in voltage. iv) Time of voltage application has no effect on discharge pattern. v) Inception voltage is well defined and Ve is slightly below or equal to Vi.
EVALUATION/INTERPRETATION ELECTRODE HV CAVITY + 0 180 ELECTRODE DIELECTRIC - b) Internal discharges in a cavity in the insulator at electrodes. Discharges occur before the voltage peak on both half cycles. Large discharges on one half cycle. If small number of larger pulses are observed in the ve peak and a large number of small discharges at +ve peak, the discharges are on the ground side and if pattern is reversed, discharges are on the hv side. No variation in magnitude with rise in voltage. Time of voltage application has no effect on discharge pattern.
EVALUATION / INTERPRETATION ELECTRODE HV FISSURE + 0 DIELECTRIC _ c) Internal discharges fissures in elastomeric insulation in direction of field. Discharges occur ahead of voltage peaks on both half cycles. Discharges are of similar in amplitude and number on both half cycle. No variation in magnitude with rise in voltage. Time of voltage application effects the discharge pattern. If the maximum voltage is kept for longer time, the discharge level decreases.
EVALUATION/INTERPRETATION HV ELECTRODE DISCHARGES + 0 DIELECTRIC DIELECTRIC ELECTRODES 1 2 3 4 Internal discharges in a number of cavities of different shapes & sizes within a dielectric. (fig.1) and discharges on an external dielectric surfaces at areas of high tangential stress.(fig. 2) and discharges on an external dielectric surfaces between two touching conductors (fig. 3) Discharges occurs ahead of voltage peaks on both half cycles. Discharges are of same amplitude and numbers on both sides of ellipse. The no of pulses increase and become unresolved with rise in voltage. Time of voltage application has no effect on discharge pattern. Ve is slightly below or equal to Vi. -
EVALUATION/INTERPRETATION ELECTRODE HV LAMINAR CAVITY + 0 DIELECTRIC e) Discharge in the laminar cavity. Discharges occur before the voltage peak on both half cycles. Discharges are of the same amplitude and number on both side of ellipse. The no. of pulses increases and becomes unresolved with rise in voltage. Time of voltage application effects the discharge pattern. If max. voltage is kept for longer time, the discharge level decreases. Ve < Vi - 180
EVALUATION/INTERPRETATION ELECTRODE HV CAVITY + CARBON TRACKS 0 180 DIELECTRIC - f) Discharges in a cavity which degenerates into a track. Discharges are of the same amplitude on both half cycles. The no of pulses increases and becomes unresolved with rise in voltage. Magnitude remains the same with rise in voltage. Time of voltage application has no effect on discharge pattern.
ON-LINE PARTIALDISCHARGE MEASUREMENT Partial discharges occur years before failure. Sufficient time to corrective maintenance to avoid In-service failure of the generator or motor. Disadvantages of Off-line Partial Discharge Measurement OR Advantages of On-line Partial Discharge Measurement Although P.D. testing has normally been done with the machine out of service (Off Line) to minimize test cost & inconvenience, plants prefer an ON-LINE. Stator winding problems such as slot discharge are not so apparent during Off-Line PD tests.
The typical 5 years interval between Off-Line tests (i.e. PD tests done during a suitable outage ) is too long a period compared to the time it takes for some problems to originate & lead to failure. ON-LINE PD testing makes it practical to test the motor or generator more frequently. So, Power Utilities and Chemical companies emphasize ON-LINE PD testing, as opposed to tests performed during outage. Find loose, overheated, and contaminated windings in stators of motors and generators well before these problems lead to failure. Assess the need for motor and generator maintenance in stator windings.
Able to determine manufacturing or installation problems, such as poor impregnation with epoxy. Provides an effective way of avoiding interruptions and failures due to slowly progressing electrical deterioration of insulation. Help maintenance engineers identify which stator windings need off-line testing, inspections, and/or repair. The main difficulty in performing ON-LINE PD test is not in detecting PD signals, but rather in distinguishing the PD from electrical noise.
CHALLENGES FOR ON LINE PD MEASUREMENT Different kinds of SENSORS Methods for noise suppression
SENSORS FOR ON LINE PD MEASUREMENT : High voltage capacitive sensor 2. Stator Slot Coupler ( SSC )
High voltage capacitive sensor A bridge like pair of capacitive sensors are permanently mounted on each phase of hydro generator ( or motor). That is key for inherent elimination of noise. Two PD sensors or bus couples are installed per phase, i.e. six per machine (see fig.1). High voltage capacitors connected to the phase terminals are used to detect the PD pulses. Capacitor sizes ranged from 80 to > 1000 pf.
In comparing the time of arrival of pulses at the two couplers in a phase, if a pulse is first detected at couplers nearest to the stator winding (N in Fig.1), then the pulse is likely to be caused by stator PD, and should be counted. However if a pulse is first detected at the coupler closest to the power system, i.e. at the coupler furthest from the stator (F in Fig.1), then the pulse is due to noise and should be ignored.
On- line Partial Discharge tests for Turbine Generators using Stator Slot Coupler ( SSC) An on- line Partial Discharge test using bus couplers was found to be unreliable for high speed turbine generators rated 100 MVA or more. In all large turbine generators there are significant sources of internal noise from brush sparking, poor contacts, and core lamination arcing, and the test results could not be correlated with known turbine generator winding condition. This realization lead to the development of a new type of PD sensor called the Stator Slot Coupler ( SSC ).
Unlike PD sensors using high voltage capacitors, the SSC is not electrically connected to the winding.
The SSC is a low voltage antenna which detects the electromagnetic energy from PD and other signals. The SSC is installed in stator winding slots. SSC is installed underneath the wedges. Each SSC is about 50cm long, 1.7mm thick, and is custom made to be the same width as the stator slot. The substrate of the SSC is an epoxy-glass laminate (NEMA G0 or G11).
The characteristics of PD and noise detected by an SSC are completely different, permitting reliable discrimination between PD and noise. PD pulses within the stator winding results in voltage pulses which last only 1 to 5 ns (see Fig. 5 (a) ). However all pulses not due to stator PD have a pulse width in excess of 20 ns (see Fig 5 (b) )
This clear difference in pulse shape permits distinguishing between all noise (external and internal) and stator PD. The SSC has a relatively flat frequency response in the range 30 MHz to greater than 1GHz. Therefore the SSC can detect the true pulse shape of any high- frequency signal propagating along the stator slot.
Since the SSC is made of the same materials used for conventional wedges and slot packing, the SSC can operate to the same class F (155 C) temperature rating of epoxy windings. Typically six slots in a stator are equipped with SSC s. The main advantage of the SSC is its ability to produce different responses to all types of electrical interference, both from within and without the machine.
SSC test can identify large turbine generators with good and deteriorated windings. A deteriorated winding has a partial discharge activity which is about 30 times higher than a generator of the same make and vintage which in good condition. FINALLY Sensors have been developed which inherently reduce the influence of noise. Experience shows that there is at least a 30:1 difference between a good stator winding and a deteriorated winding which needs maintenance. Thus the new test identifies those motors and generators needing corrective maintenance
Data presentation All data presented were obtained using either 80-pF capacitive couplers or SSC s installed within the stator winding. The pulse magnitude is measured in milli volts (mv). The peak positive and negative PD magnitudes (+Qm and -Qm) represent the highest PD pulses measured in mv. Qm is a reasonable predictor of winding insulation condition. A high Qm measured in a winding compared with a lower Qm in another winding usually implies that the former windings is more deteriorated.
Database Analysis Method A cumulative version of the statistical distribution is shown in Table 1, which also shows the maximum Qm measured as well as the average Qm. Table 1. Distribution of Qm for hydrogenerators with 80 pf sensors with generator windings For a 13.8kV stator, the average Qm from all tests was 312 mv. 25% of tests had a Qm <38mV. 50% of the tests had a Qm < 96mV, 75% were <194 mv. 90% of tests had a Qm <392 mv. Thus if a Qm of 500 mv is obtained on a 13.8 kv hydrogenator, then it is likely that this stator will be deteriorated, as it has PD levels higher that in 90% of similar machines.
Table 2. Distribution of Qm for air cooled stators, 80 pf sensors on terminals of motors and turbogenerators Table 2 illustrates the similar statistical distribution for motors and turbo generators. 80-pF capacitors are installed at the machine terminals. The distributions show that, at the 90% level, turbos seem to have greater PD than hydros.
Table 3. PD magnitude distribution for hydrogen-cooled machines at different pressures equipped with 80 pf couplers. Table 3 also shows that as operating hydrogen pressure increases, the 90% Qm PD activity level decreases significally. Clearly, before interpreting a PD result, one needs to know whether the machine is hydrogen or air cooled, and if hydrogencooled what the pressure was at the time of the measurement.
Table 4. Slot PD magnitude distribution for hydrogen-cooled machines equipped with SSC Table 4 shows the effect of voltage class and hydrogen pressure on the rest results from SSC. Similarly comparing Table 3 & 4, it is obvious that the PD measured with SSC sensors should be interpreted differently from results acquired with capacitive sensors, that is the 90% levels are usually lower with SSC than with capacitive couplers.
Table 5. Effect of winding manufacturing date on PD levels in 13 to 15 KV air cooled stators. Winding Age One surprising result from the statistical analysis of the database was the distribution of Qm as a function of winding age. Table 5 illustrates the PD results in the database from machines that were from 1 yr old to more than 50 yr old.
Table 6. Effect of manufacturer on PD levels for 13 to 15 KV stators. Manufacturer Table 6 shows the results for 13kV to 15kV stator. The cause or the differences between manufacturers unknown, but it may be due to different manufacturing processes, electric stress design levels, and assembly methods.