Transformer Partial Discharge Test Setup

[dpc]

A friend of mine is involved in trying to come up with a power source for an on-site partial discharge test of a very big transformer. One of the requirements is a large shunt reactor on the low side, connected between the test generator (120 Hz) and the transformer LV bushings.

I'd think the transformer would appear as a highly inductive load anyway, so I'm curious about the purpose/need for the shunt reactor?

Anyone familiar with transformer partial discharge test setups?

 

[motorspert]

dpc
We do a lot of PD testing on generators and motors,the principles for generators are much the same.

An offline PD test involves energizing the windings on the transformer. The circuit is from the ohase conductor to earth and so the curent path is through the insulation not through the coils. The neutral end will be at the same potential as the line end.

Thus, the circuit is one big capacitor with the copper windings forming one elctrode and the core acting as the other electrode. the current flow is capacitive and thus a reactor is required to correct the power factor and minimise the size of the test supply.

The capacitance of the windings of course depends on the physical size of the transformer under test.

I am not sure about the reason for 120hz for this test - I would have thought that the supply would be better if it was at the actual supply frequency. Double the frequency would mean double the partial discharges measured.

richard

 

[dpc]

Thanks, Richard. I think I understand what you are saying.

As far as the 120 Hz, my understanding for the this requirement is because they want to test at 1.3 x the nominal voltage on the primary and the core would saturate at 60 Hz. By using a higher frequency, they can reach a higher voltage before saturation occurs.

Cheers,

Dave

 

[electricpete]

I can guess a purpose for the shunt reactor.

I assume they plan to measure partial discharge by installing coupling capacitors on the input to the transformer. If it’s like Iris does for motors, it’s actually an R-C circuit with C acting like isolation for power frequency voltage, but acts like a shot circuit for very high frequency voltage spikes... allows that voltage spike to be seen and measured accross the resistor (in the range of a few hundred millivolts).

Somewhere deep inside the machine, a series insulation capacitance bridged over causing that voltage spike to appear at the termianls. It is clearly filtered by the inductance of the machine itself.

BUT, if we have a perfectly stiff powersupply (ideal voltage source with no series impedance), then there can not under any circumstances be a deviation from sinusoidal voltage at the terminals of that ideal voltage source.

So, maybe the shunt reactor provides a series impedance to enhance the voltage drop created by partial discharg so they can more easily be measured. Although we use no such shunt reactor for our motor partial discharge measurements.

Alternatively, if you are using an electronic power supply which has harmonics and spikes built in, you may want to isolate the transfomer measurement from that external source of voltage transients which may be mistaken for partial discharge.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[waross]

Hi dpc

By using a higher frequency, they can reach a higher voltage before saturation occurs.

Am I missing something here"? I understood that there was no magnetizing current flowing during this test.
respectfully

 

[electricpete]

If you are applying voltage on the low side and you want to to appear ratio'd on the high side, then you have to apply the voltage phase-to-phase, which would create a high magnetizing current. If you tied all three LV phases together and applied a voltage to ground, you'd get nothing on the high side.

Brings to mine another question - which side do they attach the coupling capacitors to? I'm guessing high side, which makes most of my previous comments invalid.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[dpc]

waross,

The transformer is energized from the low side and the testing is done on the HV bushings, so magnetizing current will exist.

 

[waross]

Thanks dpc
Understood
respectfully

 

[dpc]

The test voltage is applied to the LV winding, which is a delta winding. High side is grounded wye. They use bushing taps on the HV bushing for the detector connection.

I'm still not quite seeing the reason for a shunt inductor, since the magnetizing current provided to the LV winding should be at a very low pf, lagging. I would have thought a cap bank would make more sense. But I'm not burdened by a lot of knowledge on this topic, so it's easy to make guesses.

Thanks, all.

 

[motorspert]

We use reactor on large generators because we turn up with a small energising transformer and there is no magnetising current. Because our tranmsformer is small (30kVA) the magnetising current is also small - we are exciting at 200-400V.
In dpc's case things are differeent and I do not I understand the need for the reactors as i would expect the magnetising current to swamp the capacitive currenr.

 

[electricpete]

We use reactor on large generators because we turn up with a small energising transformer and there is no magnetising current. Because our tranmsformer is small (30kVA) the magnetising current is also small - we are exciting at 200-400V.
In dpc's case things are differeent and I do not I understand the need for the reactors as i would expect the magnetising current to swamp the capacitive currenr.

For motorspert - you energize your generators phase-to-ground, not phase-to-phase right? That explains no magnetizing current. I still don't understand why you would need reactor. Is yours shunt or series?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[motorspert]

electricpete,
Yes we energize phase to ground -but with the other two phases grounded
Example:
The winding capacitance of a recent large propuslsion motor was 0.5microF. The capacitive current required (at 11kV and 50Hz) is approx 2 amp. We feed this from our variac at say 240V which gives a LV current of 1 *11000/240 = 90A this will be offset slightly be the transformer magnetising current. But it is too high to connect into a LV socket and our transformer is limited to about 20A on the LV side. So we connect reactors in parallel on the HV side to reduce the LV current. Yes, we could use a bigger transformer but it gets expensive to ship around the world. I can't imagine any of this being a concern for dpc's transformer id they are using the 120Hz supply.

 

[electricpete]

Thanks motorspert. I understand now.

Shunt reactor usefuln when applying voltage to ground against all phases tied together (makes up for capacitive current to ground) with a small capacity test set.

120hz useful to test when applying voltage phase-to-phase (limits magnetizing current) with a larger capacity test set.

Going back to the original question we now see a logical reason to use one ore the other (shunt reactor or 120hz generator) but no logical reason to use both. Unless mabye they want to do multiple tests: one phase-to-phase at 120 hz and one all-phases to ground using a different 60hz test set?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[garyvin]

I hope I can clear all of these up. First usually in Xfrmer testing you use 180hz (used to use 400hz) to lower the magnetizing reactants as was said. You need to use a solid state power supply or a special 180 hz generator. If using the generator you need the shunt reactor for control of inductive impedience. You do not want the generator self exciting with the load. I have been there and it is not a pretty sight.

Usually you run this test at 150% rated three phase voltage. Do not exceed this voltage. Then measure the RIV on the cpacitive taps on the HV bushings. Make sure they are always shorted unless you are reading that bushing and it should be shorted through the test set.

 

[dpc]

SparksAlot,

I'm still not clear on how the shunt reactance controls the inductive impedance - the transformer magnetizing current is already highly inductive, and the shunt reactor just adds another highly inductive load in parallel.

And how would the generator become self excited by an inductive load and how would more inductance help?

I have no involvement in the testing - just trying to (re-)educate myself.

Thanks.

 

[garyvin]

This is kinda difficult to explain. The easiest way to look at it is you don't want the inductance of the generator and the transformer to be exactly the same. There is a slight capacitive effect on a transformer and you don't want a resonance between the generator and Transformer. This can cause a voltage to be induced by the generator with no field power applied. The reactor makes it much more stable by canceling out the capacitive effect.

 

[prc]

Dpc,
What is the rating of the transformer -MVA & HV side voltage.Must be pretty high.In normal transformers, pf is 0.85-0.9 at no-load excitation at power frequency.When you are excitating at 120 Hz and at 1.3 pu pf will improve still further.When HV capacitive current is super imposed over it, the load can be leading and comparable to generator rating, which can cause overexcitation of generator.To avoid this shunt reactor is provided in factory testing of transformers.The same may be the reason at site also.

 

[dpc]

prc,

Thanks - the interwinding capacitance was one thing I thought about - I just didn't see how it could be greater than the reactance of the LV winding, but I could certainly be wrong. I was assuming the normal exciting current would have a much lower power factor.

The HV is either 230 kV or 345 kV, I can't remember. The LV is 18 kV or thereabouts. It's something like 500 MVA. A generator step-up xfmr.

Thanks again.