ungrounded wye-delta 2

[nerb31]

the subject is 3-100kVA, ungrounded wye-delta transformer bank, 2400/4160Y-240/480V. During a spot check-up on this bank, we were surprised with the line-ground voltages 302V/320V/198V that were measured between the secondary terminals and the tank equipment grounding (as ground point). thinking it was due to grounded load or bad connections, the transformers were isolated, and terminals were cleaned. megger, winding resistance and TTR all turns good for these transformers. but after the cleaning, the same voltage magnitudes were measured when the transformer was commissioned. by the way, line-to-line voltages are balanced at around 480V. 4160V side is connected to upstream bus with LRG.

Has anyone experienced the same case or someone perhaps know what's going on with our transformers?

Appreciate your inputs.

 

[jghrist]

What were you expecting to measure to ground on an ungrounded system? The voltages depend on the relative capacitance to ground of the secondary system and the impedance of the meter.

 

[wbd]

Yes, you have no solid ground to reference only capitance to ground as jghrist states.

 

[stevenal]

Not just the capacitance. The admittance of the meter now affects the very thing you are trying to measure. Note that the measured voltages fail not only to meet in the center at 277, they fail to meet at all.

 

[waross]

The readings are a reflection of different leakages to ground of the transformers.
The leakage is so little that it appears that the meter loading is dropping the values slightly so that they don't meet.
I raised the 302 voltage up to 320 and did a quick check with the help of Mr. Pythagoras and the result was below 480 Volts. The error that I introduced should have put the result above 480 Volts.
Conclusion, you have significant meter loading.
Please consider grounding one of the 240 Volt center taps, unless there is an urgent need to run ungrounded.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[odlanor]

waross,
Your suggestion does not define whether the secondary system is grounded three-phase or single-phase.
This is a problem for analysis and calculations.
A more comprehensive solution would be to ground secondary system by grounding transformer with neutral resistor
to limit earth current to 10A.
This current allows you to give alarm at first stage and trip at second stage in a delay of 1 (one) hour!.
This is enough time to restablish the system without loss of service continuity.

 

[waross]

The information given was for a delta connection of 240/480 volt transformer secondaries at 480 Volts. Each of the three 240/480 Volt transformer secondaries will have a center tap available which may be used for an acceptable ground point even if it is not perfectly symmetrical.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[odlanor]

waross,
because is not perfectly symmetrical there is no model representation of transformer in order to calculate short circuits. It will be necessary to made tests in the field.

 

[nerb31]

thanks for all your inputs.

waross,
could you kindly explain what do you mean by "significant meter loading"? We used the same DMM with the adjacent transformer bank (ungrounded wye-delta) and measured balanced voltages (line-ground).

actually, one transformer was used for 240V service (but still not grounded)and that circuit got grounded. we isolated that circuit but still getting the same that voltage unbalance in the secondary.

I am aware that we can't get a well-balanced line-ground voltage on delta side, but would it really be that far between phases?

 

[electricpete]

I think it was referring to the fact that your meter has some non-infinite impedance, which could affect voltage to ground if other impedances to ground were same order of magnitude or higher (although dmm is very high impedance). I don't see any clear indication that meter is affecting these measurements (if they were all below 277, that would be a clear indication). On the other hand, there definitely appears to be an imbalance in impedance (for example stray capacitance) to ground on the low side and everything connected to it.

By the way, what is connected to the low side? Is this transformer feeding some cables and loads? The cause of the unbalance certainly can lie with them. One long cable run closer to ground. Also ground indicating lights on MCC's.... if one burned out you'd probably have 1 high and 2 low voltages (you have 2 lows and a high... maybe 2 burned out). Also of course if you have single phase loads, you may have different ones on each phase and it wouldn't be surprising for associated stray capacitance to be different.

=====================================
(2B)+(2B)' ?

 

[jghrist]

Why do you care what the line-ground voltage is?

 

[waross]

Hi Pete;
Can you calculate what the line to line voltages must be to correspond to phase to ground readings of 302 V, 320 V, and 198 V ?
To me the interesting point is not that the voltages to ground are not equal, it is that the line to ground readings are too low for a 480 Volt system.
This may be meter loading.
Or,
I have seen significant waveform distortion on unloaded wye:wye transformer banks but I have never had the opportunity to put a scope on an unloaded delta.
This also may be a combination of waveform distortion and metering errors.
The tip off to waveform distortion is that averaging meters, (both analog and digital, will not indicate the correct voltage with many distorted waveforms.
If this is a case of distorted waveforms, try your readings again with some load on the transformer. If this is a waveform issue you will probably get close to the same ratio of line to ground voltages but the voltages will be increased so as to be congruent with a 480 Volt delta.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricpete]

Bill - I think that line to line voltages of 467vac (just 2.5% low) separated by 120 degrees, could give the readings 198, 302, 320 to ground.

=====================================
(2B)+(2B)' ?

 

[waross]

Thanks Pete. I agree that may be a little too much error for meter loading with a digital meter.
I suspect a distorted waveform may be at play. The nature of the distortion I have observed, will certainly introduce that much error when averaging meters are used. The shape of the wave form may fool some true RMS algorithms also but not sure.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricpete]

Sorry, not 2.5% low but 2.9% low.

Here is one scenario where balanced line to line voltages 2.5% low (469 instead of 480) can give the measured line to ground voltages.

The sum of the line to ground voltages is not zero=ground, but is some other point, call it n for neutral. We can write the balanced line to neutral voltages as:
Van= 269 Angle 0 degrees
Vbn= 269 Angle 120 degrees
Vcn= 269 Angle -120 degrees
(269 is 2.5% below nominal 277 volt).

Now let us postulate that ground voltage is displaced from the 0 point as follows:
Vgn= 72.4 Angle 9.6 degrees

We can compute Vag = Van – Vgn, Vbg = Vbn-Vgn, Vcg = Vcn-Vgn to give:
Vag= 198.0 Angle -3.5 degrees
Vbg= 302.0 Angle 133.0 degrees
Vcg= 320.0 Angle -130.0 degrees

198, 302, 320 matches the measurements. That doesn't prove anything. For all we know the voltage could be exactly 480 and the meter loading is are influencing the measurement slightly. Or lots of other possibilities.

As you point out, the meter loading can't be the whole story because it affects all phases the same so cannot introduce an unbalance.

jghrist asks why do we care? Good question for the op. I would say in general we should investigate signs of unbalanced impedance to ground on an ungrounded system... it may have a perfectly benign explanation, but it can also be indication of a ground on one phase which should be resolved.

=====================================
(2B)+(2B)' ?

 

[electricpete]

but it can also be indication of a ground on one phase which should be resolved.

Obviously not a hard ground in this case.

=====================================
(2B)+(2B)' ?

 

[electricpete]

And sorry there were typos about the scenario... should've been: 2.9% low: 466 line to line, 269 line to neutral. In spite of those typos, the other numbers were correct.

=====================================
(2B)+(2B)' ?

 

[nerb31]

thanks again for the valuable inputs.

jghrist, its been a normal practice for us to measure line-to-line and line-to-ground voltages during routine check of power distribution. and its been this case where we found unbalanced line-to-ground voltage with no grounded phase at the load side. the curiosity with these weird voltage readings are of course related to whether we need to have the transformers repaired, though I have mentioned that DLRO, Megger and TTR yields acceptable results.

electricpete, the transformer bank feeds some 480 and 220V, but the unbalance line-ground voltages were measured at the transformer secondary, with all the loads disconnected. so our speculation at this point is that the problem is "within" the transformer.

waross, I'll try to clamp a PQ meter to check for the waveform and vector arrangement.

more inputs would be much appreciated.

 

[waross]

The distorted waveform that I have seen on wye:wye banks looks like the positive half of a sine wave, but a little shorter than the expected 8.3 milliseconds. Then there is about 1 millisecond of zero Volts. Then there is a corresponding negative half cycle. That sure messes up the reading of the old d'Arsonval based meters. I suspect that an algorithm that does not properly account for the zero intervals may influence a supposed "True RMS" meter.
Try taking the measurements with a little load on the transformer bank.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[PHovnanian]

Obviously not a hard ground in this case.

nerb31, add a small load (light bulbs?) from one leg to ground and compare the measured voltages to those of the unloaded case. This (with a bit of math) will tell you what the magnitude of the leakage and/or capacitance to ground is that's causing the original unbalance.

odlanor, fault currents should still be calculable (with a bit more work) for the grounded center tap case by modeling the voltage source with the appropriate symmetrical components.