Remove 3rd harmonics on load-side of delta-star transformer.

[scuba1]

I am looking for a solution to third-harmonic induced problems. The industrial facility in question is fed with a three-phase 400V arrangement, with no neutral, from a 10 kV/400V delta-star transformer (with an earth taken from the star-point secondary).This distribution configuration feeds many large three-phase loads. Single-phase 230V loads are then accomodated by the provision of multiple 50 kVA 400V/230V delta-star transformers distributed throughout the plant, with the secondary star-point providing the neutral. The problem arises due to the fact that the majority of the transformers are designated as lighting transformers, and feed high-frequency fluorescent lighting loads, which produce high levels of harmonic currents (especially third)in the circuits.The transformers being generally approx. 90% loaded are starting to overheat, but more importantly there are very high rates of failure with the fluorescent tubes and electronic ballasts. My questions are
(a) has anyone heard of harmonic producing electronic loads, producing harmonics, being themselves subject to failure from the effects of harmonics especially as in the situation above where the harmonics may be "locked" into the secondary circuit, and
(b)are there any prooven solutions which are not "tuned" to specific fixed load conditions ? The solution must address the lamp failures as much as, if not more than, the transformer overheating.

 

[dpc]

Have you verified that the correct ballasts are being used? By "high frequency" I am thinking you have electronic as opposed to magnetic ballasts. I would not think that third harmonics would be causing electronic ballasts to fail. But we have seen large numbers of early failures in these new electronic ballasts. Seems to be bad batches showing up from time to time.

The failure of both ballasts and lamps makes me wonder if the ballast are being incorrectly applied. There is a wide variety of voltage levels available in ballasts. Also, the lamps used must match the ballasts.

Just curious - have you taken power quality measurements at the 230V level? If so, what levels of harmonic voltages and currents are you seeing?

Transformer and neutral overloading in fluorescent lighting circuits has been an issue for some time. Oversizing neutral conductors and specifying transformers with appropriate K-factor ratings will reduce problems. But this is probably not an attractive option for the existing transformers. But filtering will not be inexpensive, either.

This probably didn't help much, but maybe others have better advice.

 

[peebee]

I agree. I'd first try to verify that harmonics actually are the problem. If they really are, I'd try derated or k-rated transformers.

I can also confirm that the ballast manufacturers have released HUGE lots of defective ballasts. If you're experiencing high ballast failure rates, I'd check there first.

 

[busbar]

Good comments so far. I agree that without some measurements, cures may be a bit speculative. On a retrofit basis, there are carefully-balanced zig-zag autotransformers for triplen neutral-current reduction on existing ∆-Y transformer secondaries, but the only ratings I've found are intended for 60Hz 208Y/120V service.

It might be worthwhile determining if the ballasts meet advertised and standards-based harmonic-current limits. The cost of a device like a Fluke 41 may be quickly offset.

 

[jghrist]

If you find that the voltage harmonic content is very high, you might improve the situation by derating the transformers. The harmonic currents will produce a high harmonic voltage if the impedance is high. Using more or larger transformers (derated), will reduce the impedance to the harmonic currents and reduce the voltage harmonics.

Do the ballasts meet IEC 1000-3 harmonic limits?

 

[electricuwe]

The solution to your problem looks quite simple to me:

Supply the correct voltage to the ballasts:

For a three-phase system the nominal voltage is the phase to phase voltage. If you connect the ballasts between phase and neutral of 400V/230V three-phase transformer you are supplying the ballasts with 127 V.

Check this, connect the ballasts phase to phase and you will not suffer from third harmonics problems and defectve ballasts any longer.

 

[scuba1]

To DPC: Three different types of electronic ballasts from two different manufacturers have been uesd, each typeexperienced regular failures over the last two years. In addition the high frequency tubes have been failing and "pinking" at one or both ends (100 tube failures per week @ $10 per tube!) The selection and matching of ballasts / tubes, connected voltage etc. seems to be correct. Current Harmonics on the 230V secondary (230 V phase to neutral)are in the order of 18%, with 3% Voltage THD, both being predominantly 3rd order harmonics.
To BUSBAR: The ballasts tried are CE rated and meet international standards such as IEC 1000-3
To ELECTRICUWE: The connected voltage is correct, the tubes and ballasts are 230V-rated and the phase-to-neutral voltage on the star-wound secondary is 230 V (Yes its 400V phase-to-phase)

I'm beginning to think that harmonics are the indirect cause of the problem with the fluorescent tubes & ballasts, rather than the direct cause. I have just as a result of some of your bove comments analysed some measurements taken with a Dranetz PP4000 - the 230v voltage waceform out of the transformer shows +350V transients, and sags with the three phase-voltage waveforms becoming co-incidental (one waveform superimposed on the other instead of 120 degrees apart). Is it possible that I have winding-to-winding partial faults/flashovers in the transformers (which are clearly overheating due to current imbalances and 3rd harmonics)and these in turn cause damaging voltage spikes and sags? How can I measure winding to winding inductive reactance to check for short-circuits on the 50 kVA dry-wound transformers ?

 

[dpc]

I don't have any answers, but I will just note that the voltage distortion and current distortion levels you measured are not that bad. I've seen much higher levels on systems with no apparent problems.

Are the transformer secondaries solidly grounded?

Have you tried returning the failed ballast to the manufacturer to get an explanation of the failure mode?

Is this a standard lighting application, or some special application such as UV disinfection or "grow lamps"? How many ballasts do you have on a single circuit?

Not sure what you are seeing on the Dranetz, but a 230V system has peak voltage of 325 volts and peak-to-peak of 650 volts.

 

[jghrist]

The sags with the three phase voltages becoming coincidental makes me think you have a loose primary phase connection on the 400 delta - 400/230 grd wye transformers.

A complete open on one phase would give you 230 volts to ground on one secondary phase and two phases at 115 volts to ground, each at the same phase angle, 180 degrees from the 230 volt phase.

 

[jbartos]

Suggestion: The power supply quality (lighting transformer secondary side) should be verified. According to results, if the voltage Total Harmonic Distortion (THD) is greater than 5%, then passive harmonic filters could be applied. If the existing lighting transformers are deteriorated, then the k-rated transformers should be compared to regular transformers with harmonic filters since the k-rated transformers do not mitigate harmonics, they just withstand them.

 

[busbar]

scuba1— You seem to have received some good advice, and some logical next steps for resolution. IIRC, the expected effect of triplen harmonics in a ∆-Y transformer is to be “dissipated” as heat in the delta winding, as opposed to passing them through the primary terminals. This is considered to be accepted operation, but K-rated transformers are intended have extra headroom to accommodate this function and limit premature insulation failure.

Depending upon the age of the system and current/near-term damage costs, it may be time well spent benching a few of the existing ballasts with your test gear, and maybe compare existing to some currently produced by the same manufacturer, and possibly another brand with similar published ratings. At that point, if the numbers don’t mesh, start asking questions.

With “The transformers being generally approx. 90% loaded are starting to overheat, but more importantly there are very high rates of failure with the fluorescent tubes and electronic ballasts”—again, depending on age and degree of out-of-pocket costs, it may be time to make some phone calls.

You question the situation of “harmonic producing electronic loads, producing harmonics, being themselves subject to failure from the effects of harmonics”—that does not seem ethical of any electrical-component manufacturer or conducive of increased or even continued sales volume of their products. I would be equally concerned in a similar situation for the failing ballasts themselves, possible/likely plant-fire inception and consequent damage to the serving 50kVA drytypes.

It may be wise to secure a copy of IEC 1000 and conducting your own tests—making certain that you equipment is in calibration, and comparing field measurements to those promised by the ballast manufacturer. Your plant would not be first to receive a wholesale changeout of poorly engineered ballast, with every penny paid by the ballast manufacturer. Electronic ballasts are routinely priced at a premium for premium performance, and that goal seems to have been not met.

Repeated “tube pinking” sounds like a clear ballast-design defect, and if so, probably won’t cause the ballast manufacturer to be any easier to work with.

 

[avishai]

Hi, First I would like you to be very very carefull before you try to install any added solution (transformer or anything else) as this might destabilise completely your network and lead to local resonnance with all its implications. If you can forward the grafs you got from your Dranetz it will help to help you. You do not talk here about measuring Cos Fi (Power factor) you should do it but each phase separately and see the vector behaviour of your network and transformer . You can not avoid cos fi problems with high harmonic problem that is why you have to do this measurement. This also will indicate to you what is the nature of your problem or what is "creating" the problem. After that check if you have power factor correction capacitors if you do then tell me and I will send you a hint on checking if the problem is comming from there. By the way the standards at IEC for electricity have recieved an addition of 6 in the beginning so the satndards related are 61000..... When you will read this group of standards you will see a flow-chart (IEC 61000-3-2 page 19) that will "lead " you all the way on how to monitor your problems. Other possibility is IEEE 519 and 1159 for this purpose (requires a little more understanding in network analizing).
At your service if you need more ,
Avishai Rash BScEE&PE
a_rash@bigfoot.com A.Rash BScEE & PE
a_rash@bigfoot.com