VFD issues on generator 2

[IBRCAN]

Hello, I am a generator technician working for a company that sells engine driven generators. I have very little knowledge or experience when it comes to VFDs.

A customer has two sites where the loads consist mostly of VFD driven 3 phase 480 volt motors, driving rock cutting machines. One site is powered by utility, the other by a 100 kw diesel generator.

The customer says that at the site with the generator, they have experienced numerous VFD failures. These failures are often several months apart, but there have been many more occurrences than the site on utility. They have experienced these issues with four different generators in place, at different times.

Customer says when a VFD "fails", there are no fault codes on the drive--the symptom is that the motor output is "jerky". The drive is replaced and the problem goes away. I haven't spent any time looking at how the drives are programmed and there was never any failure analysis done on the "failed" drives. The rock cutting machines utilize one TECO N3 drive and one Mitsubishi FR-D740. Apparently both have had this problem.

I performed a power analysis of the entire load over the course of two days. Looking over the data, voltage is maintained within about 5% and frequency within about 2%, the only exception being when one large load is started, the voltage and frequency each drop close to 10%. Imbalance between phases is minimal. As expected, the sinewave is horribly distorted (see attachment showing voltage and current sine waves). For whatever reason, phase A current is much more distorted than B and C. This phase A distortion was present even before any load was put on the generator, therefore I wonder if this is just a problem with my instrument. THDI is about 25% on phase A and 18% on phases B & C. The 5th harmonic is greatest, around 11% of fundamental. Peak load is around 60 kW/65 kVA. I understand that this percentage of THD may shorten generator life, but will this in itself cause the VFD issues I described? Most of the time, the plant electrical loads function normally. I did not see any harmonic filters present.

One issue I see is that the generator main circuit breaker is not equipped with a shunt trip to open the breaker on a generator shut down fault. If the generator engine overheats for example, the load will experience an under voltage condition for a few seconds as the generator winds down. A shunt trip coil is going to be installed on the breaker.

Any thoughts would be appreciated.

 

[IBRCAN]

Current sine wave

 

[itsmoked]

The front end of a VFD is a three phase bridge rectifier. The rectifier charges a capacitor bank to the rectified peak voltage. The drive then takes this stored DC source of energy and simply (not!) re-synthesizes new 3 phase AC to supply the driven motor. However, since the drive is doing this synthesis it can alter the frequency giving the main point of any VFD's use.


The very large downside of this simplicity is that with rectifier front-ends the capacitor bank is charged up to nearly the peak sinewave voltage. Once fully charged, you can see, nothing will be coming thru the rectifiers. As energy is drawn to drive the motor the average voltage on the capacitor bank begins to drop. Wherever this voltage drops to is where on the sinewave the rectifiers begin to conduct. This means the area of the sinewave that actually is 'hooked up' and supplying current to the capacitors backs down the sinewave farther and farther utilizing more of the sinewave with greater loading.

Unfortunately this means all the power drawn from the source comes only from the sine peaks. This is bad news for the power source and especially bad news for generators.

The best way to reduce this undesirable situation is to proceed the VFD with line reactors. These serious inductors prevent the large current gulps that happen at the sine peaks. Since the current is restricted to some lower level the capacitor bank's average voltage drops some and more importantly the conduction angle greatly increases as the current is drawn for a longer time thru the rectifiers. This reduces the sine distortion considerably.

As an example the last VFD I installed listed that the wiring system must be designed to supply 41A out of the box. However, if a line reactor is installed the current needs dropped to 23A!

Part of the problem with your installation is the horrid current pulses demanded by the VFDs if they lack line reactors. Part is possibly how the current gulps "ring" the system essentially finding any resonances and driving them but only under specific loadings. This adds to the mystery of random time periods between VFD deaths since it depends on specific loading that might be avoided for periods then come in other periods of crusher use.

I'd install line reactors even though they are not cheap and not small as they will mitigate a world of VFD supply issues. They'll reduce the thrashing of generator current carrying components too.


Keith Cress
kcress -

http://www.flaminsystems.com

 

[jraef]

I agree with everything Keith said, but I also see some curious "hash" on your current sine wave that is extra concerning. By any chance is your generator an inverter type? That seems to be popular now on smaller generators and this is causing issues with things like VFDs. An inverter based generator has a PMDC generator that feeds DC into a solid state inverter, which is doing the same thing as the VFD in terms of creating a PWM output (other than varying the frequency). The problem is, when using that to power a VFD you have an PWM inverter feeding a rectifier, feeding an PWM inverter! The "gulping" of current when the diodes of the VFD rectifier become forward biased (what Keith described as the peaks of the sine waves) can cause the generator inverter to current limit to protect itself, which then starves the rectifier of voltage and if one of the diodes fails to conduct because the line side voltage didn't reach the forward conduction threshold, then the next one in sequence that does conduct has to "make up" for the depletion of current pulled out of the capacitors by the transistors feeding the motor. That next pulse then attempts to pull current too fast and it can damage the diodes, the capacitors or everything in the drive. When you have an inverter feeding an inverter, it's usually a race to the bottom to see which one dies first...

This same basic thing can also happen on a synchronous generator too, if the generator is sized too closely to the load. You said the generator is only 100kW, but you don't say the size / quantity of VFDs and other loads. If the total load is more than about 75kW, I would be concerned about the VFDs.


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

 

[IBRCAN]

Thank you for your feedback Keith and jraef. I understand now exactly what causes the distortion.
The generator is synchronous, and the load doesn’t exceed 60kw (or didn’t during my power study).
This load is spread among 10 drives. I didn’t document the exact motor sizes but am sure the load is compromised of at least 90% VFD driven motors.

I’m not sure if the other site on utility has line reactors on the VFDs (I will ask the customer). Assuming there aren’t any, would the stability of the grid and possibly lower impedance of the supply transformer helping to minimize sine wave distortion?


In any case, I will advise the customer to look into line reactors

 

[davidbeach]

Is the system neutral grounded the same way in both cases? Many drives don't like anything other than a solidly grounded wye connection.

 

[itsmoked]

Hi IBRCAN; 60kW spread over x number of drives is almost no different than a single 60kW drive because every one of the smaller drives all pull their current in-step with each other the same exact way.

motor output is "jerky"

Really needs to be better defined. I really can't think of how a drive could fail in a way to provide 'jerky' results. They typically fail utterly on the output side or die dead-as-a-door-nail.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ScottyUK]

In some ways having multiple small drives is worse because the smaller drives have looser requirements on input distortion than their larger cousins.

 

[waross]

One issue that stands out.-
An unexplained jitter on the red phase current.

IBRCAN[/b]]For whatever reason, phase A current is much more distorted than B and C. This phase A distortion was present even before any load was put on the generator, therefore I wonder if this is just a problem with my instrument.

The very first thing is to "roll" the inputs to your instrument to rule out instrument error.
No corresponding jitter on the other phases as per Kirkoff's law.
Unexplained drive failures.
It may be productive to investigate:
Is there a line to neutral load causing the jitter?
Is the jitter current returning on the neutral or on the ground?
The transients on A phase are particularly troubling.
Often there is a voltage spike accompanying a current spike, but in the opposite direction.
With that in mind the downward spikes may be more troubling than the upward spikes.
The current is actually going 40 Amps reverse. The accompanying reverse voltage spikes may be an issue, but where are they?
I suspect that the ground or the neutral return path has a higher impedance that the supply conductors and the return current is showing up at the wye point. I suspect that the accompanying voltage spikes are showing up across a high impedance connection in the return path.
Given that there have been issues with several generators and that the current readings had to have been made with the generator under load, I would try to locate the source of the current jitter.
Given that the current is shown to be reversing, there may be capacitors involved. Maybe not.



Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[IBRCAN]

Thank you Bill. After going over the data files, I definitely intend to connect phase A to a different channel on my instrument when I return to site. I will also connect CTs onto the neutral and ground. The neutral is solidly grounded and brought out from the generator into the building, however I didn't see any phase to neutral loads. I wonder if there is a high resistance ground fault on phase A. Time for some megger testing perhaps.

Sorry that the failure mode isn't better defined. I haven't been able to obtain the most useful information from the customer.

See the attachment. The majority of the voltage distortion on all phases seems to correspond with the abrupt changes in phase A current.

Will keep you posted.

 

[ScottyUK]

Looks like a fairly normal current waveform, albeit with a lot of HF switching noise superimposed on it. The characteristic 'double hump' of a 3-phase rectifier load is clearly visible. The commutation notches are quite steep as each phase comes on load and drops out later in the cycle as current transfers from one phase to the next.

My first question is 'are those large transient pulses real?'. What are you measuring them with? Some Hall-effect probes are sensitive to voltage changes on the conductor being sensed, with high dv/dt events on the conductor on which the probe is clipped being coupled into the probe output signal giving the illusion of high di/dt events. You have a relatively high dv/dt event in the form of the commutation notches. If you can get to it, try sensing current at the neutral end of the stator winding where the voltage relative to earth doesn't move, or put some copper foil connected to earth over the phase conductor at the position where you're measuring current so it forms an electrostatic shield between your probe and the phase conductor, and then see how the waveform looks.