Triplen Harmonics

<The problem is that we recorded a large number of triplen harmonics at MV side.>

What is the effect, or the perceived threat, arising from these harmonics?

The problem is the total harmonics distortion exceeds the allowable limit.

Your VSD's are rich in harmonics. Especially the 5, 7, 11th harmonics. Depending on the impedance of your generators some of this may be seen on the MV side.

In addition, you may find there is significant notching and distortion of the waveform on the LV side, and this will also cause a reflected distortion on the MV side.

A model of your system with the reflected impedances would help you to determine what is reflected and at what magnitudes. Be sure to inclide PF correcting caps as they set up there own resonances with the VSDs.

What about the triplen harmonics? Can they be seen on the MV side?

Thanks.

GW has stated: this will also cause a reflected distortion on the MV side, so this question is already answered. Given that the LV harmonics are being presented to a transformer, one might expect some attenuation of the higher-order harmonics on the HV side, though predicting them accurately is going to be extraordinarily difficult. Their magnitude is best determined by testing.

The symptoms described would suggest that a low-quality VSD is in circuit on the LV side, or a number of them perhaps. Isolating, or replacing, the VSD(s) in question might eliminate these harmonics.

Can I say that if I open the neutral connection between the generators, they won't be any triplen harmonics on the MV Side?

Thanks

Adhyartha,

The design of all VSD's has their power input go into a rectifier section. This guarantees that there will be harmonics imposed on the power line feeding the VSD, regardless of brand being used. Depending on the motor's load, which can vary all over the place, the VSD is drawing power from the line for different portions of the AC cycle. Therefore, the distribution of these harmonics cannot be predicted, but can only be measured.

I have put a quality line analyzer on the incoming line, as well as the load side of VSD's, and have measured significant percentages of harmonics as high as 31^st and higher orders. But as others have noted, the lower ones (11^th and below) are usually the most common.

I don't expect you to have any success with interrupting your neutral path. I suggest the following steps:

1. Ensure that for every motor, the motor grounding conductor's supply end is terminated only on the load side equipment grounding terminal in the VSD. Treat it as you would a circuit conductor, and run it with the load conductors without combining it with other grounding conductors. Terminating it even a few inches away from the VSD's load side grounding terminal will significantly increase the number and effects of the harmonics. (I speak from experience here, not just from reading the manufacturer's instructions.)

2. Consult with your VSD manufacturer's engineering staff (not their sales people) to select proper sized line reactors or other line-side filtering equipment, which will attenuate and filter the harmonics to a level which is within the specifications of the generator sets. Typical line reactors are sized for 3% or 5% impedance, but the only difference is the current capacity and inductor rating. It is possible that the engineers will suggest sizes other than the standard ones, as well as other types of equipment. Installing line reactors can create other problems, because they are heavy and can be fairly large.

3. Consider replacing your MV transformers with ones which include better electronic noise shielding between the HV and LV sides (if any are made at your voltage and power levels). Consulting with the transformer manufacturer will probably be the best way to go.

4. Look at ways to improve the cooling of your generators, either with better ventilation of their location/enclosure, or any other method that is appropriate or needed. Certainly you will want to monitor their winding temperatures. Although this will not reduce the harmonics, it may reduce their effect on the generators or allow you to know when other steps need to be taken (such as reducing generator load).

I hope this helps--John M.

AC drives ( I will call VFD) usually have a simple rectifier front end.

However, if it is a DC drive ( I will call VSD)with a phase controlled rectifier then there is additional notching and possibly poor power factor. At 50% speed the pf is approx 0.5.

The triplens are interesting, since the 3rd, 6th, 9th harmonics are missing from the fourier series for 3 phase VFD and VSD......unless they are single phase drives or running with an unbalanced AC line. A 10% unbalance may introduce significant triplens.

GW : I'm told that these triplens, in a 3phase system , add up in the neutral because they always have a 0 degree phase angle amongst the three phase components. What happens when we have a 1 phase system ( i.e. 120/240v/1ph-3w ) and we are also told that each 120v phase is 180 degrees apart from the other 120v phase,and we have two loads with triplen harmonics, each on a 120v phase ? Do the triplens add up in the neutral ?

To be honest I haven't done the math for your example. To have exactly matched loads then I believe the neutral will add and sum to zero, but the triplens will be present in the 2 hot lines.

(But experience says they will not, and always to ones detriment---Murphys law.)

Many 3 phase systems don't have a neutral (delta or ungrounded Y) but the triplens I believe will still be present on unbalanced loads.

"the neutral will add and sum to zero" Greetings from the east side of the continent GW : Ok , if the triplens add up to zero in the neutral of a 120/240v/1ph-3w system,how come the triplens add up to three times the individual phase load amount ( assuming that the three single phase loads are of equal magnitude) in a 120/208v/3ph-4w system . That's the part I don't follow : in a 3phase system, each phase is 120 degrees apart--- but the triplens add up to three times and don't cancel because these triplens are not 120 degrees apart ----- in a 1 phase system, each of the two phase-to-neutrals is 180 degrees apart and you believe that the triplens add up to zero ( or cancel ) in the common neutral ?? The little sheep has lost his way

I think so, because the triplens harmonics have the same phasors as the fundamental (3 x 0 deg and 3 x 180 deg), which result in a cancellation in the common neutral.

"(3 x 0 deg and 3 x 180 deg) " : Very well, but I thought the fundamental , in the 3 phase system was 120 degrees , and so it would follow that in the neutral it would lead to 3 x 120 deg = cancellation , but in reality, they don't cancel because they are of zero sequence and are not 120 deg apart but are 0 deg apart which leads to addition or 3 times the individual amount. --- I believe that in the 1phase system, it should be 2 x 180 deg in your statement ?-------I guess the question is whether the Triplen harmonics, which are defined as "a zero sequence current" ( that don't cancel in a 3 phase system, but add up in the neutral) are still defined as "zero sequence current" when in a 1 phase system, and lead to adding up in the neutral ?

Friends,

I have not done the math either, however in the literature the practical answer is that in a 3-phase 4-wire system the odd-order harmonics can frequently increase the neutral current to 150% of the phase current. The NEC, §220.22, requires that the 4-wire 3-phase neutral be sized for the full current of any phase and has a fine print note warning that this may not be sufficient. Panel board manufacturers in the USA market their equipment with optional neutral bus sizing of 150% and even 200% of the phase bus sizes.

Because of the phase angles involved, I believe that the practical answer for maximum neutral current is never greater than 200% of the phase currents.

--JMM

Hi JMM : I agree with you on the 3 phase system ; what I've been trying to find out is what happens when the system is a 120/240v/1phase-3wire ? Do the Triplens add up or do they cancel ? I'm guessing add up, but am not sure .

Cancel. Since the voltage, phase-to-phase, at base frequency is a sine wave with both lines crossing zero at the same time (one rising to a positive peak and the other to a negative peak), all multiples of this base frequency will cross zero at the same time. The phase shift of current (i.e. the power factor, whether leading or lagging) will be the same number of degrees for both lines. Therefore the current waveforms for both lines also will always cross zero at the same instant for all harmonic orders. Therefore, all harmonic currents will always sum to the difference between the phase currents for each order of harmonic.

JMM and friends : Thanks for the explanations; I finally got it . So , it seems that all these electronic devices ( operating at line-to-neutral voltage) that have 3rd harmonic and other triplen currents are a big problem on a 3phase-4wire system ( and require oversized neutrals and K-rated transfos) but no problem at all on a 1phase -3wire system ( like 120/240v/1ph-3w) and so no supplementary overheating of cables ,panels, etc.--- and if we travel upstream to the transformer(and assume that the primary is from a 3phase system ) , the triplen currents will not overheat the transformer, ( because they won't circulate in it but rather pass through to the primary ? ) and so we don't need to spec a K-rated transfo? If this is how it is , then the best way to mitigate these triplen harmonic current effects caused by electronic loads such as PC, printers,ltg ballasts, is to connect them to a 1phase-3wire supply !?

Ovide Brudo--

I think you have two separate issues tied together here. 1) The neutral current problem. It affects 3-phase 4-wire systems with electronic (non-linear) loads, but not 1-phase 3-wire systems. 2) The higher losses in transformers when they are carrying a significant electronic (non-linear) load. Here, I think that a "K-rating" may be important for any system. Since we are still on the rising curve of increasing use of electronic devices, the problem is still developing.

The electrical codes are based on experience as well as theory. In the future I expect that codes and standards may require engineering evaluation of harmonic current loads, just as they now require the evaluation of available short-circuit fault current.

We may also see the introduction, testing, and listing of equipment to be placed on the load side of transformers, which will filter or otherwise control the magnitude of these harmonic currents.

Probably the best way to mitigate these triplen harmonic current effects caused by electronic loads is to specify and install the circuits and distribution equipment with sufficient over capacity that unanticipated problems are minimal.

--JMM

My background is specifically variable speed drive systems.

About 95% have been 3 phase systems. Very few single phase.

Because I am in my late 50's, most of the experience is with DC phase control. The AC rectifier front end is only a special case of the DC.

Our experience says we size transformers specifically for the drive system. HV (medium voltage) primary preferred. Please do not put other loads on our dedicated AC transformer, even if it is several MVA.

It is the only way we have of controlling K-factor, harmonics, wiring practices, and selection of components. The HV primary gives a significant impedance isolation from other systems.

Non-linear loads from consumer products is another kettle of fish.