APPLICATIONS OF ACOUSTIC LOCALIZATION TO ONLINE POWER TRANSFORMERS David Larochelle NDB Technologies
Reasons leading to acoustic testing are different weather it is a manufacturing or field environment
Trigger for field testing DGA IEEE C57.104 Duval Triangle
Trigger for field testing Acoustic localization becomes the next thing to know for decision making
Possible Actions No Actions (non critical) Onsite repair Lower the load Replacement
Field Challenges Mechanical noise o Pumps o Fans o Core vibration o Rain Electrical noise o Corona o Radio Frequencies Physical restrictions o Radiators o Control Cabinets o Unknown internal structure Others o No control on voltage applied
Field Challenges Acoustic waves from PD are often very small in amplitude, even lower that noise baseline Combining electrical signals with acoustic signals enables noise reduction methods
PrPd Phase Resolved Partial Discharge PrPd Gives an overview of electrical or acoustic activity Acoustic pattern Electric pattern
Sensors RFCT : Sensitive to small current peaks created by PD Pros : Convenient to install Cons : Sensitive to electrical noise from the surroundings
Sensors UHF : Sensitive to electromagnetic emissions from PD UHF Antenna is inserted inside the transformer tank Pros : Immunity to environment (Uses tank as faraday cage) Cons : Installation requires fittings that can be different in each transformer.
Phase Angle Windowing Multiple PrPd patterns can be seen Localization efficiency depends on the ability to focus on one pattern at a time
Phase Angle Window Phase angle can be used to focus on one pattern at a time
Phase Angle Window Phase angle can be used to focus on one pattern at a time
Phase Angle Window Phase angle can be used to focus on one pattern at a time
Phase Angle Window Phase angle can be used to focus on one pattern at a time
Averaging Accumulate a number of waveforms Adding all waveforms together (point to point) Reduces noise while PD signal will remain.
Averaging Example Key is synchronization of each waveform. Propagation from PD source to the sensor is repeatable.
Averaging Example Let s take a simulated waveforms And add noise to it
Averaging Example Average by 2 +
Averaging Example Average by 4 +
Averaging Example Original signal Average by 8 Average by 64 Average by 1024
Averaging Example Real life example of averaging efficiency on a single phase power transformer Acoustic background noise was greater than the acoustic PD amplitude Use electrical PD event as the synchronization source for the overlay of acoustic signals.
Averaging = 1 (no averaging) Averaging The noise is high, there is no indication of any acoustic wave.
Averaging = 4 Averaging Noise is still high, but a peak can be distinguished.
Averaging = 32 Averaging The waveform is still noisy, but it is clear that an acoustic wave was captured.
Averaging = 1024 Averaging The noise level is very low. The acoustic waveform is clear, and its time of arrival can be easily found.
Averaging Synchronization can be achieved with one acoustic signals if electrical signal is too noisy Synchronization made on acoustic channel (upper blue waveform) Electrical PD was highlighted from noise
Case Study DGA indicating severe overheating Gases increasing for the past 3 years
RFCT was placed on transformer ground Case Study
Case Study Averaged acoustic signal (1024)
Localization pointed to a lead connection Case Study
Case Study Motive: DGA increase indicating PD.
RFCT clamp was used to acquire electrical signal Case Study
Acoustic peaks show relation between acoustic and network frequency Case Study
Sensors are gradually moved toward the source of PD Case Study Acoustic Waves are acquired and denoised
Acoustic provided localization area Case Study
Case Study At this point, given the relatively slow rate of gassing and confirmation of a source of PD, it was decided to put the transformer back into service while options were weighted as to how to resolve the issue. Before doing so, the inspection cover was opened to see if the PD source could be visually located.
Case Study There were a few small spots in the lead insulation that the technician described as mushy. These spots were reinforced with insulating material. The unit was closed and put back into service.
Case Study a few days after energization, the transformer tripped off causing an unplanned outage. Failure investigation showed a deterioration of the insulation between the core and the X1 conductor, in the same area given by acoustic waves.
Case Study
Case Study In a manufacturing environment, the induced test will confirm if a repair was successful If an onsite repair was done, another acoustic investigation should be done, trying to get the before and after acoustic test.