OT- How will inverters affect the power factor/phase relationship of a 3 phase supply

OT as not machne tool related.

I work on a boat which has a recently installed Air-con set up with 4 identical compressors which cut in and out as heat load fluctuates fed from a 400V 3phase supply via 4 separate inverters.

Other items of equipment (watertight sliding doors) fed from the same supply each have a phase rotation sensor relay which energises an an exceptionally annoying alarm

It seems that when 3 compressors are running, the supply is altered sufficiently for the alarms to be energised.

I remember (vaguely) that motors on their own draw an inductive load causing the power factor to lag, do inverters create a different characeristic (ie reactive or capacitive) which might be creating the alteration to the supply. I have tried measuring voltage drop but can see none.

The sensor in question is a schneider RM4TG which unfortunately has no adjustment to it's settings and it seems it is a bit too sensitive. If I am armed with what causes the problem, I can start looking for alternatives or perhaps a smoothing circuit or some form of time delay.

Thanks for any light you can shed.

Hi Art!

An inverter works on a pretty simple scheme- incoming AC is converted to DC by a bridge rectifier and capacitor bank. From there, DC power is switched through thyristors to generate a waveform of either single or polyphase output.

Because of it's design, the load the inverter places on line SHOULD be symmetrical, and should vary essentially in proportion to the load being demanded by the device connected to the output... as the motor requires more current, the inverter will DRAW more current. The inductive properties of the motor, however will NOT be reflected back THROUGH the inverter's input in such a way that you'd get an odd power factor/phase relationship beyond what you'd get on any other linear supply. There's a certain amount of capacitor charging that occurs on startup, but aside from that, it's fairly stable.

Not having any firsthand experience with the sensor you referenced, I'm only guessing... but I'd wager that the sensor isn't so much responding to the actual line conditions, but rather, it is probably somehow reacting to RF noise being generated by the inverters' switching cycles. Some RF noise WILL find it's way backwards through the capacitor bank and rectifiers in a way that other electronic devices will go bonkers. A stack of ferrite beads on the sensory leads coming into PLR would be my first attempt.

If you have a broadcast AM/shortwave radio receiver at your disposal, turn it on, and walk around the machinery, listen for noise coming over the radio... associate that noise with what your sensor may be responding to... it will probably steer you in the right direction.

Inverters are well known to introduce 3rd and 5th harmonic content onto the 3-phase supply. A set of Line Reactors will help immensely - alternately an isolation transformer can be used.

My hunch is a line reactor between each drive and the 3-phase supply will clean things up quite a bit. Good web resources to check:

TCI, LLC - Trans-Coil Home

MTE Corporation - Driving Power Quality

We have never installed a VFD without a line reactor in any system we put into a plant. . . the cumulative effect of several VFDs on line at the same time is far greater than you might think and when we have upwards of 100 VFDs all running at the same time on a single power distribution system with no detectable problems . . . I'd say the line reactors are pretty effective.

Are the inverters 1- or 3-phase input? Even if they presents a perfectly resistive load, if they are 1-phase and two of them are on the same phase you can get a heavy imbalance.

Thank-you for your input, what you have all said sort of bears out my thinking about 'noise' in the supply although it seems I was off base regarding power factor.

Fulmen: The Air Con Compressors are 3 phase output rated at 15 HP each.

Dave: I regret that we have no shortwave radios to try what you suggest, we're at sea at the moment this could take a while to sort but at least I know what the cause of the syptoms is and have a line in on the cure.

Motion: Thanks for those links, I will look into the line reactors and how/where we'd fit them.

Using link reactor on the DC bus will help. A good inverter system will use a single input dual output incoming Ac xfmr with delta and wye outlets. A 12 pulse diode arrangement with dual inputs at 30 phase shift to remove harmonics.

Get a power quality meter and you can check the incoming AC for harmonics in the voltage and current waves. An bad diode on the the inverter can cause DC backfeed which will really cause some headaches.

My next step would be to put a scope on the mains feeding the phase loss relay. I'd want to know what disturbance is causing it to trip. It may be that it's a brief voltage dip, and the problem would be easily fixed by installing a delay of some sort in that relay.

http://stevenengineering.com/pdf/45RELAY_RM4U_RM4T.PDF

Certainly your inverters are putting harmonic (high frequency) noise back on the line to some degree. But fixing that sounds expensive, and I'd want to be sure that's the cause first...

Jim McIntyre said:

My next step would be to put a scope on the mains feeding the phase loss relay. ....

Certainly your inverters are putting harmonic (high frequency) noise back on the line...... But fixing that sounds expensive, and I'd want to be sure that's the cause first...

Click to expand...

if scope avail of course this would be great step to see the noise so u can see the reduction after a 'fix.'

no HP size given, but I will assume about 5hp for 3ph compressors.... so 5% line reactor like we sell prob is only $ 100-150 each - not expensive in my book to fix this...... I wud not do the more std 3% since u have an issue, i wud do 5% to get more bang for my buck.

Engine Room Artificer said:

Thank-you for your input, what you have all said sort of bears out my thinking about 'noise' in the supply although it seems I was off base regarding power factor.

Click to expand...

Harmonics mean lower power factor, in general, and any direct-connected rectified supply, 3 phase or not, will have a lower power factor. Reactors will lower the current harmonics, and also improve the power factor somewhat.

Reactors will also lower the voltage to the VFD a bit, so larger is not always better..... usually not a big issue, but if voltage is already low (as it sound that it may be) that can be the difference.

But then a direct rectifier lowers the voltage to all the loads when the diodes turn on, so it may be an overall improvement there too.

mike_kilroy said:

if scope avail of course this would be great step to see the noise so u can see the reduction after a 'fix.'

no HP size given, but I will assume about 5hp for 3ph compressors.... so 5% line reactor like we sell prob is only $ 100-150 each -

Click to expand...

all 4 compressors are 15HP each, we have no scope on board, now I know what the problem is I can contract a specialist to diagnose what is required to rectify when we finish this trip in a month or sos time, dunno if that sort of tech will be availabe in Borneo but I might be surprised.

The 'noise' happens with any permutation of 3 of the 4 compressors running which possibly negates the possibility of a bad diode.

Thanks for all assistance so far, I was wondering if this would fall to the bottom unanswered. I'll put some pics on later so ya'll can see the boat etc.

may I assume the compressors are fairly fully loaded when operating?

If so, there may be a pretty significant drop in the peak of the AC when the diodes in 3 VFDs all begin conducting more-or-less at once. It can "clip" the AC, especially with a fairly high impedance source like a generator, and that may be enough to disturb the sensor.

The bus filter capacitors in the VFDs must be charged with a pulse of current twice per cycle on each of the 3 lines. Rectified 3 phase has a characteristic "double pulse" of current, which can be a significant load for that instant.

Installing inductors in-line will lower and "spread out" the pulses, and may be enough to keep the sensor happy.

I don't know how the sensors work, but I assume they look for the timing of the phases. Too much 'clipping" may cause the sensor to falsely detect a missing or late phase voltage, and set off the alarm.

Would I be right in thinking that the alarm is set off when the compressors start up? or does it happen after they have been on a while?

Can you reset the alarm? If so, does it go off again without the compressors re-starting?

since it takes 3 of the 4 ON to make the issue, it is either a combination of that much switcher noise or as JST points out wave distortion.... so if it is noise, the amount of load will probably make little difference since the majority of the noise generated in a vfd is just due to the switching and so much from the amount of current flowing, so another test like I am about to suggest if possible won't make a hill of beans difference. But if it is waveform distortion then amount of load will effect it. So, can you turn the vfd speed down to say 1/2 speed? that would cut the input power pulled in 1/2 also, so now instead of 15*3 load you have 7.5*3 load - equiv to just 1.5 vfd running, which according to your test should not cause the sensor false trips.... If this works, can you simply run the compressors at slower speeds to reduce input power and let them run a tad longer to get thru this trip with all 4 turned on (I assume that is better than not all turned on?)?

JST said:

may I assume the compressors are fairly fully loaded when operating?

Would I be right in thinking that the alarm is set off when the compressors start up? or does it happen after they have been on a while?

Can you reset the alarm? If so, does it go off again without the compressors re-starting?

Click to expand...

Honestly Jerry, I think they run on full load when on, and the inverters just soft start them, they only ever show 50Hz when running. Difficult to say if the alarm sounds as they start or when they have been running a while as the bridge is some distance away. The panels with the watertight door electrickery in are somewhee else again. If I'm sat in the control room and see 3 are on load I can ring the bridge and guarantee the alarms are lit; (not sounding as I've disconnected it)!

The alarm self resets and we haven't been able to replicate it with anything else that is driven from inverters, even the bow thruster which has a 85kW inverter.

was gonna put some pics on so you can see, but can't seem to link facebook pics easily

Re the power factor issue, since the DC power supply capacitors are charging and discharging at a cyclic rate, reading as a sawtooth on an oscilloscope, the start and end of diode conduction will not be symmetrical on the AC peak. That will translate as a capacitive shift in the power factor. It would be interesting to add more capacity, which should reduce the shift by making the reducing the ripple.

Just conjecture.

Bill

reducing the ripple VOLTAGE will occur with more capacity (although it's hard to get into the case to add it, and using the DC terminals needs boxes, etc).

However, the CURRENT pulse may be still high..... less voltage change, more capacity, same net charge transfer. it "might" help, since the distortion should theoretically occur a bit higher on the waveform.

The current seems to be the issue, that or harmonics (the 85 kW thruster almost surely has inductors or other stuff to fix the power factor/current draw issue).

Three 15 HP inverters are getting up there near 40 kW, probably significantly more in kVA, so they are a comparable order of load, and with no correction, might be a serious wave distortion creator.

The right way is to look at the current draw and voltage..... you can't do that, so we have to guess. But going from what I know to happen in similar situations, I suspect waveform distortion. Could be harmonics, depending on how the sensors work.

Either way, inductors are 95% + likely to fix the issue.

Here's a couple of papers on VFD line harmonics:

http://www.mirusinternational.com/d...Meeting Harmonic Limits on Marine Vessels.pdf

http://www.eagle.org/eagleExternalP...armonicsinElecPowerSystems/Pub150_ElHarmonics

I recall about 20 years ago there was a big push in Europe to deal with what was called the poor Power Factor of Off the Line power supplies. It was actually the line current harmonics that was the problem. I remember a slew of PWM converter controller chips that sensed the line voltage & current and generated a bus voltage without causing line harmonics. I don't know of any VFDs that do this but I suppose some of them might and if the line reactor doesn't solve the problem then a VFD incorporating this type of converter might. Come to think of it, I looked in a newer Mazak turning center and the Mitsubishi drive have a DC bus converter with the spindle and axis drives powered from it rather than the AC bus directly. I know it regenerates energy back to the line during deceleration. I wouldn't be surprised if it also minimized line harmonics as that function would be relatively easy to implement.

Cheers,

Duane

I found another paper on Yaskawa's site that might be more helpful:

http://www.yaskawa.com/site/dmdrive.nsf/link2/CAMK-822TJA/$file/PP.HarmonicMitigation.01.pdf

Cheers,

Hmmm...

Engine Room Artificer said:

Dave: I regret that we have no shortwave radios to try what you suggest, we're at sea at the moment this could take a while to sort but at least I know what the cause of the syptoms is and have a line in on the cure.

Click to expand...

I would think that you'd have a radioman with a maritime HF SSB set, and most are general-coverage recievers. If that's the case, have your sparky tune down to about 1000khz and see what kind of racket he hears when the VFDs kick on.

DK :-)

I would like to thank all the contributors to this thread; with some of the included data, I went to the OEM of the doors that have the alarm sensor and laid out the case.

They agreed and have come up with a different sensor which will be a voltage sensitive relay rather phase sensitive, which we have ordered.

It is true we are dealing with the symptom rather than the cause but this is the only equipment that exhibits any sort of problem so hopefully we will all be pleased.