Reducing the 5th and 7th harmonic 2

[veritas]

I am not a harmonics guru and so need some advice please in planning a network. A client wishes to use 6 pulse inverter VFD's on drives from 250kW to 500kW at 480V. Naturally there will be substantial 5th and 7th harmonics present. Currently, the 13.8kV/480V 1.5MVA transformers, supplying these drives, are delta-star. Star neutral grounded. The client has asked me to consider a Yyn0 and Dd0 trfr instead of the Dyn11 as a mitigation measure.

I am wondering how these will help though as the 5th and 7th are not triplen and therefore not zero sequence harmonics so I do not see much use in a trfr delta winding as a mitigation measure. Besides a Dd0 trfr also means the 480V network is not earthed and that's another problem.

Yyn0 means zero sequence currents from the LV cannot be reflected to the HV (HV neutral not grounded but LV is) as zps currents as there is no delta (3 core trfr) to provide the ampere-turns balance. I do know the tank delta effect comes into play - but that's besides the point as the 5th and 7th are not zps harmonics.

Was wondering if someone could confirm my understanding above or show me the grievious errors of my ways!

Thanks.

 

[Skogsgurra]

The thinking behind the (parallel) use of DY and DD transformers (often combined and with a common primary) is that, if you have a perfect 50/50% load share, you will effectively get the equivalent of a twelve-pulse system, seen from primary.

A twelve pulse system has its first harmonics at 11th and 13th. So, if 5th and 7th are a problem, it helps. But only if both secondaries carry the same load. Otherwise, it MAY help, somewhat. But no guarantee.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Skogsgurra]

Addendum: Do NOT use YY. The triplene harmonics will cause large voltage drops because of inability to "short out" triplenes. You need to have one delta winding, either primary or secondary, or an extra tertiary (not connected anywhere) winding. Using a DD is also OK.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[veritas]

I think though that triplen harmonics will only be a problem if both neutrals of the Yy are grounded. If the LV neutral only is grounded, then the trfr zps impedance is around 5 to 20 times the pps impedance. It would also stop the triplens as only pps and nps currents can pass from LV to HV.

I tend to think the original Dyn11 is probably the best. It will solve the problem of triplen issues (though not expect here with a 6-pulse inverter) and provides a convenient grounding point. My thoughts are towards a single tuned filter to the 4.8th harmonic to get rid of the 5th. The filter will be associated with the capbank used for pf correction.

I've read up that a Dy and Dd in parallel helps but that means that for every one trfr two will now be required which has real estate and cost implications. Also will need to get load shared as close to 50/50 as possible. A possibility nevertheless. Thanks for that!

 

[LionelHutz]

Filters like the MTE Matrix AP, TCI HGP or the Mirus Lineator all work very well to reduce the THD of a drive. I listed them in order of effectiveness from testing I did, but note that the HGP requires an additional 3% line reactor to be that effective. The Matrix AP and HGP were both just over 5% THDi and the Lineator was about 9.5% THDi.

The PFC bank isn't for correcting the VFD loads, right?

 

[jraef]

I agree that a passive harmonic filter in each drive will give you roughly the same, if not better, performance as a 12 pulse arrangement. That is going to be especially true if the aforementioned drives are different sizes and the idea was not to use 12 pulse front ends on each drive, but to attempt to split up the standard 6 pulse drives and loads on each phase-shifted transformer. There is a near zero chance that you will attain perfect load balancing that way and without it, the effectiveness is not going to be there so you may end up spending a lot of money and real estate without achieving the goal. Lastly, a solution like this, even if the loads are balances between one another, loses effectiveness if the incoming line voltages are more than 1% unbalanced, something also difficult to maintain in the real world.

And as Lionel said, I hope the capacitors are there for something else and behind a tuned reactor, otherwise you are asking for trouble.

Another probably better option to consider when you have multiple drives like this is to use an Active Harmonic Filter that is connected at or near the PCC, sized to provide corrective current for the 5th, 7th and 11th harmonics. TCI’s HGA is the best one I have come across, we use them a lot. It has an algorithm that corrects the power factor as well, probably eliminating your need for these extra capacitors too.


" We are all here on earth to help others; what on earth the others are here for I don't know." -- W. H. Auden

 

[LionelHutz]

Yes, for multiple big drives per transformer, it can make sense to use an active filter.

 

[veritas]

Yes, the drives are different sizes. The shunt capbanks are there for pf correction. I added on singe tuned filters to these banks to assist with the harmonics issue. Problem is that it only addresses the 5th. That is the largest but would be nice of the 7th could be eliminated as well.

I will certainly look into TCI's HGA. Thanks for the tip.

 

[veritas]

Just to clarify - the capbanks are for the DOL motor loads.

 

[veritas]

what puzzles me is how the two trfrs in parallel will cancel out the 5th harmonic. Let's say I have two identical VFD drives that generates 1p.u. of harmonic current each. Drive 1 is fed from a Dy11 trfr and drive 2 from a Yyn0 trfr. The two trfrs are paralleled on the HV side. So the harmonic currents will be 30deg out of phase with each other on the HV side.

Now, to me they must be 180deg out of phase to cancel. So how does the 30deg phase shift do it?

Thanks.

 

[davidbeach]

They're only 30 degrees at the fundamental. 5th is a negative sequence, so that gets shifted 150 degrees in the opposite direction, 30 - -150 = 180. 7th is a positive sequence, so it gets shifted 210 degrees in the same direction as the 30 degree shift. 30 - 210 = -180. Both cancel.

 

[LionelHutz]

The shunt capbanks are there for pf correction.

To correct the power factor of what???

 

[Skogsgurra]

DOL, LOL.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[veritas]

David

They're only 30 degrees at the fundamental. 5th is a negative sequence, so that gets shifted 150 degrees in the opposite direction, 30 - -150 = 180.

I've been doing my phasor diagrams and been scratching my head but I still cannot see how the 5th's will cancel. In the attached (How do 5th harmonics cancel.pdf) I have shown two VFD loads. I presume that the I5th from each VFD load is in phase on the 480V side as the motor V & I are supplied from a system which is originally in phase, since the source network is common.

The way I see it, the nominally Dy11 is effectively a yD1 when energised from the LV side. When energised with nps currents then it behaves like a yD11 which means that the I5th (L1) harmonic currents at 13.8kV is still only 30 degrees out of phase with the Ith (L2) so adding them will not yield 0.

What am I missing?

Thanks.

 

[davidbeach]

The phase shift is 30 degrees times the harmonic number. That’s why the 5th is shifted 150 degrees and the 7th is shifted 210 degrees. The sequence determines which direction the shift.

 

[LionelHutz]

Is the problem that you're thinking the 5th on the Yy shifts 150* (0 - -150) and on the Yd shifts 180* (30 - -150)?

 

[veritas]

So in fact it is actually not a perfect cancellation. With reference to my diagram, if the phase shift for the 5th is 150degrees when going from the 480V to the 13.8kV side of the "yD1" trfr (T1) then it will still be 30 degrees out of phase with the 5th from the yY trfr where there is a zero degree phase shift. So adding two waveforms that are 150 degrees out of phase is what we're actually looking at. There will thus be a small residual. Thus if I had 1p.u. 5th at the 480V side of each trfr then the resultant on the HV side is 0.52p.u. (1p.u. + 1p.u.)/0.52p.u. which is about a 75% reduction in the harmonic current. Same goes for the 7th.

Lionel - no I am aware that there is no shift through the Yy0.

I've learnt a great deal with this thread, thanks especially to David.