Power quality vs. VFD

[nolabama]

Safety first always. Thanks for the help man, I hope to have this Monday late evening.

 

[wirenut1980]

If the overvoltage fault is because of high DC bus voltage and it is short duration, maybe there is a capacitor switching transient taking the drive out. Any capacitor banks on the high voltage or low voltage side?

 

[Besoeker]

If the overvoltage fault is because of high DC bus voltage and it is short duration, maybe there is a capacitor switching transient taking the drive out. Any capacitor banks on the high voltage or low voltage side?

I'm inclined to think not.
The DC bus generally has a fairly large capacitor bank. A short duration transient should not greatly affect its terminal voltage.

 

[wirenut1980]

I'm inclined to think not.
The DC bus generally has a fairly large capacitor bank. A short duration transient should not greatly affect its terminal voltage.

I have had many many cases where I have tied capacitor banks switching on, to drives tripping on DC bus overvoltage. Sometimes on a normal transient, which typically ranges from 15% above the normal waveform peak up to about 30% above the peak of the waveform. Sometimes I have measured a larger than normal switching transient due to a resonance, or magnification. Attached is a waveform of a larger than normal switching transient from a 12 kV, 1200 KVAR utility capacitor bank just down the street from a building where I was monitoring at the 480 V main disconnect. The peak was around 1120 V.

Many 480 V drives will have their dc overvoltage trip if it gets above about 820 V dc for any amount of time, transients included. A 3% or 5% line reactor is often enough to knock down the transient enough, so that it does not trip out the drive.

 

[ptonsparky]

I have had many many cases where I have tied capacitor banks switching on, to drives tripping on DC bus overvoltage. Sometimes on a normal transient, which typically ranges from 15% above the normal waveform peak up to about 30% above the peak of the waveform. Sometimes I have measured a larger than normal switching transient due to a resonance, or magnification. Attached is a waveform of a larger than normal switching transient from a 12 kV, 1200 KVAR utility capacitor bank just down the street from a building where I was monitoring at the 480 V main disconnect. The peak was around 1120 V.

Many 480 V drives will have their dc overvoltage trip if it gets above about 820 V dc for any amount of time, transients included. A 3% or 5% line reactor is often enough to knock down the transient enough, so that it does not trip out the drive.

Man, I gotta get me some better toys!! but then I would have to go back to school again. Maybe I will just settle for a fishing pole.

 

[Besoeker]

Attached is a waveform of a larger than normal switching transient from a 12 kV, 1200 KVAR utility capacitor bank just down the street from a building where I was monitoring at the 480 V main disconnect. The peak was around 1120 V.

Nice catch.
The transient doesn't surprise me. But it's all done and dusted with less than a cycle.
The bucket capacitors in a variable frequency inverter are there to provide a low impedance source for the inverter stage and to decouple it from supply frequency.
A sub cycle transient ought not to affect it to any appreciable extent.

Not quite related, but your three phases don't look quite right. The three positive half cycles appear to be at 60deg intervals.
Maybe one is reversed?

Should look like this:

 

[nolabama]

Sorry but no o scope - everyone that had one was on vacation last week. Anyway we installed a powerstat/variac and this seems to have helped. We turned the voltage down to around 236vac phase to ground. The drives did not fault at this voltage. The drives were faulting at around 495vac phase to phase. We cannot fully test the system as the bridge suffered unrelated mechanical damages. The bull gears are nearly destroyed and one of the transmissions is busted up pretty good. We have tied a tug boat to it to open and close it for the mean time.

What we did find was this however. The 9 mile HV transmission line is increasing in voltage to the tune of 1400 volts. The Poco is delivering advertised voltage. It was told to me by HV contractor that we may not have a concentric neutral transmission wire, it is possible a drain? jacket? wire. I am unsure of what exactly he called it. If someone can explain how this line can increase in voltage I would thank you. HV guy also said that possibly a loose/broken ground in one of the junction boxes. We have not had enough track time to explore the junction boxes. We may not get that time either.

It has been explained to me that the drives are prolly getting dirty power. If they are getting a harmonic the "window" of allowable voltages is lowered. ie If the power is dirty they will never be able to see the 528vac because they will fault before this. This explains why they fault at around 500vac. I know, I know we need that waveform. I am trying to get it.

Anyways - we have thrown a band aid on it and the powerstat seems to be the hot ticket.

 

[ptonsparky]

By turning the variac down to 236 phase to ground, does that put your L-L at about 408v?

That's right your 1ph input. never mind

 

[wirenut1980]

Nice catch.
The transient doesn't surprise me. But it's all done and dusted with less than a cycle.
The bucket capacitors in a variable frequency inverter are there to provide a low impedance source for the inverter stage and to decouple it from supply frequency.
A sub cycle transient ought not to affect it to any appreciable extent.

Not quite related, but your three phases don't look quite right. The three positive half cycles appear to be at 60deg intervals.
Maybe one is reversed?

Should look like this:

Nice catch yourself!
The monitor I was using was a Megger PA-9+. When recording voltage in delta, I am required to hook up channel 1 A-C, channel 2 B-C only. I hook up channel 3 B-A. It should probably be hooked up A-B, but I never really paid much attention to it since only the first 2 voltage channels (and line currents) are required for power calculations.

Regarding the duration of the transient, I agree with you that there should not really be an effect on damaging the drive. However, the way I understand it, the drive just looks at the voltage instantaneously and if the voltage gets above X, it trips. It does not take into affect the duration. I vaguely recall some drives had a time delay parameter for the overvoltage trip, but more modern ones do not seem to have that (I generally see smaller drives from 1 hp up to about 50 hp).

I attended a course this year put on by ABB just on their drives, and their technical guys flat out told me that adding any time delay for over-voltage would damage the drive, so who knows...

 

[Besoeker]

Nice catch yourself!

Thank you. I'm just used to looking at this kind of thing.

Regarding the duration of the transient, I agree with you that there should not really be an effect on damaging the drive. However, the way I understand it, the drive just looks at the voltage instantaneously and if the voltage gets above X, it trips. It does not take into affect the duration.

I agree. A threshold. Not time dependent.
But on which voltage? Mostly, the inverters I have dealt with use have take control power from a SMPS driven from the DC link. We have some that on a common DC source so ther eis no AC to monitor for over voltage. The bits of the system most likely to be damaged by over voltage are the DC link bucket capacitors and the IGBT inverter devices. Logically, that's where I would expect the voltage monitoring to be.

 

[wirenut1980]

Thank you. I'm just used to looking at this kind of thing.



I agree. A threshold. Not time dependent.
But on which voltage? Mostly, the inverters I have dealt with use have take control power from a SMPS driven from the DC link. We have some that on a common DC source so ther eis no AC to monitor for over voltage. The bits of the system most likely to be damaged by over voltage are the DC link bucket capacitors and the IGBT inverter devices. Logically, that's where I would expect the voltage monitoring to be.

But if the transient occurs on the AC side near the peak of the waveform, why wouldn't it get passed through to the DC side? So even if monitoring occurred on the DC side, it should still see the transient.

 

[Besoeker]

But if the transient occurs on the AC side near the peak of the waveform, why wouldn't it get passed through to the DC side? So even if monitoring occurred on the DC side, it should still see the transient.

The DC side has a substantial capacitor bank. A bucket. It provides a low impedance source for the inverter stage. It decouples the output and input frequencies. A sub-cycle transient ought to be swallowed by it and not produce any significant perturbation of the DC link voltage.

As it happens, we have used this circuit arrangement as a means of suppressing transients.