Voltage distortion from 3rd harmonics with NGR 4

[camurphy]

Hello,

I am performing a harmonic analysis of a 24 pulse 4160 V VFD. This VFD is replacing existing VFDs of the same size and will be connected to an existing system which has delta-y transformers with neutral grounding resistors. This 24 pulse drive has significant 3rd harmonics as determined by a test by the manufacturer attached to their system which I don't believe has a neutral grounding resistance. My issue is, when I put this drive harmonic spectrum into my harmonics software (Palladin DesignBase) the third harmonic interacts with the transformer grounding impedance creating very large voltage harmonics. Though I haven't heard of this being an issue, it makes sense to me as the 3rd harmonic is zero sequence and will create current in the neutral.

My question is: Would you expect this third order harmonic current to be reduced when the drive is installed in the actual system (with neutral grounding resistance), compared to the measured spectrum I have (without a neutral grounding resistance)?

Thanks

 

[PHovnanian]

Your third harmonic returns to the transformer through the neutral. The grounding resistor connects the transformer neutral point to the grounding system to limit single phase to ground fault currents as they make their way back to this neutral point. Under normal circumstances, no current will flow through this grounding resistor (aside from those due to shunt capacitance). So a solidly grounded or resistance grounded system should make no difference to the third harmonic currents/voltages.

 

[jghrist]

A 24 pulse VFD should not produce significant 3rd harmonic distortion, or any significant harmonic lower than the 23rd.

If there is 3rd harmonic current, as PHovnanian says, it will not flow in the NGR if there is a neutral path to the source transformer. The only thing I can see is if the model assumes the VFD neutral is grounded and there is no path back to the transformer neutral except through ground.

 

[camurphy]

Thanks for the great responses!

jghrist - We too were quite surprised at the level of harmonics below the 23rd that were collected during the manufacturer testing. We went several rounds with the manufacturer and they state it is representative of their drives so it is what we have to work with.

I suspect the software does make the assumption that there is no neutral as there is no way to indicate its presence in the software. In the existing system, the neutral doesn't appear to have been brought back to the switchgear from the transformers according to the record drawings - which may or may not be accurate.

The existing system harmonics with the existing drives were monitored and no troublesome 3rd harmonic voltage magnitudes were found. I would think this would suggest a lower impedance path for the zero sequence currents to get back to the transformer than through the NGR exists.

If there is no neutral provided, as there often isn't to three phase VFDs, how do the zero sequence currents get back to the transformer?

Thanks

 

[jghrist]

If there is no neutral, and no path from the phases to ground, then there will be no zero-sequence currents. How was the VFD connected during the factory test that showed third harmonic currents?

 

[PHovnanian]

If there is no neutral provided, as there often isn't to three phase VFDs, how do the zero sequence currents get back to the transformer?

They don't.

But harmonic currents behave somewhat differently than fundamental zero sequence currents. Think of the nonlinear load as a harmonic current source. The resulting harmonic voltages are a product of this current and the system impedance at each harmonic. So for a low impedance path (solid neutral back to a delta wye transformer), the triple harmonics (the ones that combine in phase and return on the neutral) will be low. If the return path (the neutral) has a higher impedance, then the harmonic voltages at the load will be higher. An open neutral will generate a voltage across the load's own internal impedance. Think of a Norton equivalent circuit for each harmonic.

 

[camurphy]

I have been going back and forth with the software company trying to determine their take on this situation and what they have implemented.

It makes sense to me that if there is no neutral there should be no path back to the transformer for the triplen harmonics. What happens to them isn't as clear to me. Your explanation PHovnanian makes sense. I'm not sure of the details of how the current creates a circuit in the VFD output but I could imagine the triplen harmonics flowing through an internal impedance in the VFD. What confuses me is that if I understand it correctly, this would create a voltage distortion across this internal impedance. VFDs are usually treated a current sources of distortion. How does that work?

In addition, I have data collected at the switchgear between a transformer and a VFD for a system that has no neutral, that indicates the presence of a 3rd harmonic and I don't fully understand how it is getting there.

jghrist, you have a good point. I don't know the system setup of the test facility but I am pretty sure the connection to the VFD only included the phase conductors and no neutral so I expect they tested it that way.

 

[jghrist]

Third harmonics are zero-sequence only if they are balanced. Consider a single-phase load with third harmonic current, like a single-phase computer switched-mode power supply. For the third harmonic current to be purely zero-sequence, the third harmonic current in all three phases has to be equal. Since there is only one phase, this is obviously not the case. The current has to have equal positive-, negative-, and zero-sequence components.

I once measured the distortion in the current on the source side of a delta-delta transformer feeding an arc furnace. It had high third harmonic currents even though there is no path for zero-sequence current in a delta-delta transformer.

 

[RRaghunath]

If there is no path for third harmonics through intercoonected neutrals, then, it has to be through cable capacitance to ground. With triplen harmonics, the phase-to ground cable capacitive reactance is 1/3 that with fundamental component.

 

[camurphy]

Thank you all. These have been very helpful responses.

 

[PHovnanian]

What confuses me is that if I understand it correctly, this would create a voltage distortion across this internal impedance. VFDs are usually treated a current sources of distortion. How does that work?

'Usually' is the key word here. For cases in which no harmonic currents are allowed to flow externally (triplen harmonics back on a neutral, for example) a voltage source model might be more appropriate. However, the phenomena that generate harmonics in a VFD or rectifier are highly nonlinear. So the model becomes dependent on how these voltage waveforms interact with switching devices and whatnot. And since this is an issue of what goes on inside the VFD, we generally leave these concerns up to the VFD designers. If they say it needs no neutral, we say 'fine'. Its your problem then.

Since most of us concern ourselves with what flows back through a neutral, to a transformer or some filter external to the VFD, we tend to concern ourselves with those conditions where neutrals, delta windings and other affected stuff is provided. In these cases (where there is a low impedance path), a current model usually works better. If the path has significant impedance, harmonic voltages will be seen and past some point, things get nonlinear again. And weird.

 

[neelis]

camurphy - Were you able to find out anything more regarding this. I too have data from the Vendor showing triplen harmonics on the input to a 24 pulse drive. This drive is interconnected to an islanded system with neutral resistors. The software tool we are using shows very high harmonic voltage distortion to 'drive' the current through the resistance. I am trying to locate information - 1)should the triplens be ignored when you have a resistance in the neutral or 2) do we really have an issue. Seems like the model uses current source as the choice for modelling the harmonic currents but you would think that the added impedance in the zero sequence network would work to limit current flow.

 

[camurphy]

Neelis, I apologize it took me so long to get back to you and though I suspect you had to solve your issue by now I will post what I did.

The system I was studying had no neutral connection between the VFD and the switchgear. As a result, I do not know how the triplen harmonics were travelling between the two. I did, however, have the advantage of significant amounts of monitoring data from the existing drives. I also know that the software I am using creates the voltage THD by running the current at the triplens through the neutral grounding resistor.

I took the current and voltage harmonic data for the existing drives in the system and calculated an effective impedance that the triplen harmonics see in the existing setup. I then plugged this into the model for the value of the NGR. This got me results that appear reasonable.

I clearly stated this method in the report to the client with the caveat that when the new drives are installed the impedance path from the triplens may be different. I do not fully understand how the tripen harmonics do their thing, however, when there is no neutral so I cannot say that I am 100% confidant in this methodology. When the new drives are installed, harmonic monitoring will be performed and I will be interested to see how well it matches the model.

 

[neelis]

Thanks camurphy