Harmonics and Transformers

[aelectricalman]

I'm looking for some help with equations for determining what percentage of harmonic distortion can be buffered by a transformer. I have used the run of thumb of 1% per transformer bit that is not always acceptable. I want to learn to calculate the percentage Vthd amd Ithd drop through a transformer. Will I need to consider the spectrum analysis including each order harmonic or is there a general formula that can be used. Hope I'm not over thinking this.

Thanks for any help.

 

[junkhound]

You do need to know which harmonics to do a complete analysis. Also the core material and lamination thickness, and of course how the transformer is wound, including leakage inductance and winding resistance.

For instance, the triplett harmonics can be reduced over 90% by a delta-wye transformer connection. If yousketch ut the wveforms, you see that the 3rd, 9th, etc harmonics become short circuited by the primary delta. Hence, if the resistance and reactance of the winding is known, simple linear equations will show the attenuation of the harmonics.

Single phase and non-triples harmonics are tougher, you need to calculate core losses, eddy currents, etc. to get an accurate number. For 3 phase triplets, these are negligible compared to the short circuit effect of the delta-wye type configuration.

Edit PS: do NOT forget to thermally size a delta wye transformer accordingly, as the triplett harmonics do show up as heat!

 

[aelectricalman]

Edit PS: do NOT forget to thermally size a delta wye transformer accordingly, as the triplett harmonics do show up as heat!

We found thermal signatures that were consistent with harmonics ( from our IR study) on the fuses of the 18 pulse rectifier capacitor cabinets. Heat was showing up on the A and B phases.

The case ground on the delta tank was very low Z (1.8%), so there were likely no harmonics created from an arcing ground that were circulating in the delta.

The 18 pulse rectifiers should have only been producing 17th and 19th order harmonics (and multiples there after), while cancelling most of the lower order harmonics.

The utility was providing a Vthd of average 8.4% to the customer. There is one 2000kVA pad mount oil filled XFMR and one dry type drive transformer on the system (3 -6 pulse transformer to make an 18 pulse rectifier) 30 deg. apart, that these harmonics had to penetrate to get to the capacitors. My field experience says that approximately 1% reductions are to be expected through an XFMR although I don't have supporting engineering calcs.

We had 8.4% Vthd on the A and B phase and 6.1% on the C phase averages.

This would place harmonic Vthd levels on the capacitors in the facility at approximately 6.4% on A, 6.4% on B and 4.1 C%. That is still significantly above IEEE 519 thresholds of <5% for less Vthd or 3% for any one harmonic order on a system less than 69kV systems.

Also, we had capacitors on the A and B phase fail along with blown fuses in the A and B phase of the primary side of the 14.4 kV pad mount. The C phase had not blown fuses.

Any thoughts or feelings.

 

[junkhound]

we had capacitors on the A and B phase fail along with blown fuses in the A and B phase

Take a look at if the 17th and 19th are high enough frequency to cause the leakage inductance of the transformer to resonate with the PF correction caps. Have seen this type failure often, as the resonant frequency produces a higher peak voltage across the capacitor than the cap rating, also higher currents than the cap rated currents at the higher frequencies.

Slight phase offsets (e.g. 119 deg vs. 120deg) on an 18 pulse rectifier do produce appreciable 15th harmonics also, which is a triplett.

Edit PS: if you have access to PSpice (or download the free LTSPice from the linear technologies web site) you can model the system to look at your resonances and frequency domain. Multi pole rectifiers are nice for smooth dc, but do produce LOTS of harmonics.

 

[aelectricalman]

Thank you. I have access to E-Tap. I will model it.

This is very interesting conversation.